.

.: ,

I
bi
,'L
{I,
$,I I
design &application of
Paraflow plate heat exchangers
kj I
While the original idea for the plate heat
exchanger was patented in the latter half of
the past century, the first commercially CONTENTS
successful design was introduced in 1923 INTRODUCTION 3
by Dr. Richard Seligman, the founder of T H E PARAFLOW AND I T S
APV. Initially, the plate type exchanger PRINCIPLE 4
consisted of a number of cast gunmetal 2 THERMAL PERFORMANCE 10
plates enclosed within a framein a manner
3 T H E PROBLEM O F FOIJLING 16
quite similar to a filter press. Although this
basic principle remains unchanged, the 4 COMPARING PLATE AND
TUBULAR EXCHANGERS 27
early 1930s saw the development of plates
5 RECOVERY O F PROCESS HEAT "'
pressed in thin gauge stainless steel and
subsequently, in various other ductile 6 PARAFLOWS IN SEA WATER
metals. Other refinements in Paraflow en- COOLING SYSTEMS
gineering boosted operating pressures from IRROSION AND HEAT
rRANSFFR 45
about 15 psig to 350 psig in some current
machines, increased liquid flow capacities
to as much as 11,000 GPM, and provided a
wide selection of gasketing materials to
meet varying temperature and p r o d r t
characteristic parameters.
Considered for many years to be an ap-
proach to heat transfer that was limited to
hygienic heating and cooling duties within
the dairy and brewery industries, the plate
heat exchanger has proved to be an ex-
tremely flexible and economic unit for
virtually any process. Not only is it widely
used in the dairy and food processingfields
but with the development of high volume
flow units and both plates and sealing
materials that resist corrosion and elevated
temperatures, the Paraflow has gainec
acceptance for chemical, Detrochemical.
phaimaceutical, industrial and marine
applications.
This handbook descri t'araflow
and its operating principle, compares it I
with conventional tubular exchangers, and The late Dr. Richard Selieman.
 the Paraflow
and its principle
The Paraflow is the original plate type heat exchanger designed by APV to provide
maximum efficiency and cost effectivenessin handling thermal duties while mini-
mizing maintenance downtime and floor space requirements.
the kame
The Paraflow plate heat exchanger as shown in Fig. 1consists of a stationary head
and end support connected by a top carrying bar and bottom guide rail. These form a
rigid frame which supports the plates and moveable follower. In most units, plates
are securely compressed between the head and follower by means of tie bars on
either side of the exchanger. In a few models, central tightening spindles working
against a reinforced end support are used for compression. When Paraflows are
opened, the follower moves easily along the top bar with the aid of a bearing sup-
ported roller to allow full access to each individual plate.
With the exception of some sanitary models which are clad with stainless steel,
Paraflow frames are fabricated of carbon steel and are finished in chemical resistant
epoxy paint. Frame ports accept bushings of stainless steel or alternative metals
which, with various types of flanged or sanitary connections,form the inlet and out-
let. nozzles.Byusingintermediateconnectorp1ates
.. - asshown inFigure2, unitscanbe

PLATE PACK

t
c,
 FIG. 2 Two section Paraflow
with connector plate

3 Cutaway of Paraflow
I FIG.
plate shows turbulence during
passage of product and
service liquids.

the plates
The closely spaced metal heat transfer plates
have troughs or corrugations which induce tur-
bulence to the liquids flowing as a thin stream
between the plates (Fig.3).
The plates have comer ports which in the
complete plate pack form a manifold for even
..

fluid distribution to the individual plate pas-

lSTLHXLlO

P
sages (Fig. 4). - *

the gaskets
OPEN
The seal between the plates is established by a
peripheral gasket which also separates the thru-
port and flow-areas with a double barrier. The
interspace is .vented to atmosphere to prevent
FIG. 5
cross contamination in the rare event of leak-
age (Fig.5).
5
 unique interlocking gaskets
As an exclusive feature, Paraflow heat
exchanger plates have interlocking
gaskets in which upstanding lugs and
scallops are sited intermittently around
the outside edges. These scallops ensure
that there are no unsupported portions
of the gaskets and, in combination with
the patented form of pressed groove,
provide mechanical plate-to-plate s u p
port for thesealing system. The upstand-
ing lugs (Fii. 6)maintain plate alignment
in the Paraflow during pack closure and
operation.Thegrooveform provides 10046
peripheral support of the gasket, leaving
none of the material exposed to the out-
side. In addition, the gasket/groove
design minimizesgasket exposure to the
process liquid.
plate arrangement
Comparison of Paraflow plate arrange-
ment to the tube and shell side arrange-
ment in a shell and tube exchanger is
charted in Fig. 7. Essentially,the number
of passes on the tube side of a tubular
unit can becompared with the number
of passes on a plate heat exchanger. The
number of tubes per pass also can be
equated with the number of passages
per pass for the Paraflow. However, the
comparisonwith the shell side usually is FIG.7 Pass arrangementcomparison:
more difficult since with a Paraflow, the plate vs. tubular
total number of passages available for the flow of one fluid must equal those avail-
able for the other fluid to within *l.The number of cross passes on a shell, however, .
can be related to the number of plate passes and since the number of passagedpass
for aglate is an indication of the flow area, this can be equated to the shell diameter.
This is not a perfect comparison but it does show the relative parameters for ,
each exchanger.
With regard to flow patterns, the Paraflow advantage over shell and tube designs
is the ability to have equal passes on each side in full counter-current flow, thus
obtaining maximum utilization of the temperature difference between the two
fluids. This feature is particularly important in heat recovery processes with close
temperature approaches and even in cases with temperature cross-overs.
Whenever the thermal duty permits, it is desirable to use single pass, counter-
current flow f& an extremely efficient performance. Since the flow is pure counter-
flow, correction factors required on the LMTD approach unity. Furthermore, with
all connections located at the head, the follower is easily moved and plates are more
readily accessible.
6
 FIG. 10
Troughs are formed
at opposite angles
FIG.9 to the center line
Corrugations pressed in adjacent plates.
into plates are
perpendicular to the
liquid flow.

plate construction
Dependingupontype,some~~esemptoYdisgrmal~wwhiceothersaredesignedfaa
vertical flow (Fii. 8). Plates are pressed in thicknesses between 0.020 and 0.036
inches (0.5 - 0.9) and the degree of xtcckmd * kading.isit. The mort
severecaseoccurswhenonepmxssliqrtidkogmttingatthehighmtwwking-
sure and the other at
the plate and res
typical0.1-0.2 inchgap.
be provided to maintain the gap and th@is done by twodiffemH plate fonns.
One method is to press pips ino a plate with deep WBShboIfd carrugah to
2 of heat transfer .edace @is.9).
a)rrugltianswithmain-
plates are arranged 80 that ax-
0 2 tot sqwreinchdarea.The
and the cross pattern fonns a
tortuous path that promotes substantial liquid turbulence and thus, a very high
nsfer coefficient.
mixing and variable length plates
To obtain optimum thermal and pressure drop performancewhile using a minimum
number of heat exchanger plates, mixing and variable length plates are availablefor
severalAPV Paraflow plate heat exchanger models. These plates are manufactured
to the standard widths specifiedfor the particular heat exchanger involved but are
offeredin different cormgation patterns and plate lengths.
Since each type of plate has its
own predictable performance
characteristics, it is possible to
calculate heat transfer surface
which more precisely matches
the required thermal duty with-
out oversizing the exchanger. X. 11 Low H'I'U passage
This results in the use of fewer
plates and a smaller, less expen-
sive exchanger frame.
To achieve mixing, plates
which have been pressed with
different corrugation angles are
combined within a single heat
exchanger frame. This results
in flow passages that differ sig-
nificantly in their flow character-
istics and thus heat transfer
capability from passages created
by using plates that have the
same corrugation pattern.
For example, a plate pack (Fig.
11)of standard plates which have
a typical 50" corrugation angle
(to horizontal) develops a fixed
level of thermal performance
(TU)per unit length. As plates
of 0" angle (Fig. 12) are substi-
tuted into the plate pack up to a maximum of 50% of the total number of plates, the
thermal performanceprogressively increases toa level that typically is twice that of
a pack containing only 50"angle plates. *-

Thus, it is possible for a given plate length to fine tune the Paraflow design in a
single or even multiple pass arrangement exactly to the thermal and pressure drop
requirements of the application.
Of more recent developmentareplates of fixed width with variable lengths which
extend the range of heat transfer peaormance in terms of HTU. This is proportional
to the effective length of the plate and typically, provides a range of 3 to 1from the
longest to the shortest plate in the series. As shown in Fig. 13,mixing also is avail-
able in plates of varied lengths and further increases the performance range of the
variable length-plate by a factor of approximatelytwo.
This extreme flexibility of combining mixing and variable length plates allows
more duties to be handled by a single pass design, maintaining all connectionson the
stationary head of the exchanger to simplify piping and unit maintenance.
8
-.

plate size and frame capacity

Paraflow plates are available with effective heat transfer area from 0.28 f t 2 to 50 ft2
and up to 600 of any one size can be contained in a single standard frame. The largest
Paraflow can provide in excess of 30,000 ftz of surface area. Flow ports are sized in
proportion to the platearea and contrd the maximum permissibleliquid throughpiit
Fig. 14. Flow capacity of the indi-
vidual Paraflow, based on a maxi- iw j
mum port velocity of 20 fps ranges
from 15 GPM in the 'Junior' to
. OO GPM in the Model SR235.
11O
-This velocity is at first sight some-
what high compared toconventional
pipework practice. However, the
high fluid velocity is very localized
in the exchanger and progressively
is reduced as distribution into the
flow passages occurs from the port
manifold. If pipe runs are long, it is
not uncommon to see reducers fitted
in the piping at the inlet and exit
connections of high throughput
machines.
q ,1 1 !
25

lo 0 ,o
PORT DIAMETER (INCHES)

FIG .14 Throughput vs. port dia. at 14 ft/se

plate materials
Paraflow plates may be pressed from 304 or 316 stainless steels, Avesta 2 L A l O or
254SLX, nickel 200, HastelloyB-2, C-276or G-3,Incoloy 825,Inconel 625, Monel400,
titanium or titanium-palladium as required to provide suitable corrosionresistance
to the streams being handled.
gasket materials
As detailed in Fig. 15,various gasket materials are availableas standard which have
chemical and temperature resistance coupled with excellent sealing properties.
These qualities are achieved by specifically compounding and molding the elasto-
mers for long term performance in the APV Paraflow.
Since the temperatures shown are not absolute, gasket material selection must
take into consideration the chemical composition of the streams involved as well as
the operating cycles.
._ APPROX. MAXIMUM
OPERATING TEMP. APPLICATION
GASKET MATERIAL.
. PARACRIL General aqueous service,
275F 135C aliphatic hydrocarbons
(Medium-Nitrile)
PARATEMP High temperature
300F 150C resistance for a wide range
(EPDM)
of chemicals and steam

PARADUR Mineral oils, fuels,

400F 2 : vegetable and animal oils
(F'luoroelastemer)
PARAFLQR 400F 2---; Steam, sulfuric acids
(Fluoroelastomer)
FIG. 15 Paraflow gasket guide
 2 thertnd performance
The Paraflow plate heat exchanger is used most extensively in liquid-liquid duties
under turbulent flow conditions. In addition, it is particularly effective for laminar
$ow heat transfer and is used in condensing,gas cooling an applications.
turbulent flow
For plate heat transfer in turbulent flow,thermal performance can best be exempli-
fied by a Dittus Boelter type equation:

Nu = (C) (Re)" (Pr)m(&) where

Nu Nusselt number hDe

Re VDe
Reynolds number 7

Pr CPP
Prandtl number 7
De Equivalent diameter,
(2xaverage plate gap) I OW RATPIPASSAGE LBSiHR

Sieder Tate correction factor -

F , Performancedetails:
Series HX Paraflow
and reported values of the constant and exponents are
C = .15 to .40 m = .30 to .45
n = .65 to .85 x = .05 to .20
Typical velocities in plate heat exchangers for waterlike fluids in turbulent flow are
1to 3 ft/s (0.3-0.9m/s) but true velocities in certain regions will be higher by a factor
of up to four due to the effect of the corrugations. All heat transfer and pressure drop
relationships are, hawever, based on either a-velocitycalculated from the average
plate gap or on the flow rate per passage.
Fg.1illustrates theeffect of velocity for waterat 60"Fon heat transfercoeffiuents.
This graph also plots pressure drop against velocity under the same conditions.The
film coefficientsare very high and can be obtained for a moderate pressure drop.
One particularly important featuie of the Paraflow is that the turbulence induced
by the troughs reduces the Reynolds number at which the flow becomes laminar. If
the characteristic length dimension in the Reynolds number is taken at twice the
average gap between plates, the Re number at which the flow becomes laminar
varies from about 100 to 400 according to the typeof plate.
To achieve these high coefficients, it is necessary toexpend energy. With the plate
unit, the friction factors normally encountered are in the range of 10 to 400 times
those inside a tube for the same Reynolds number. However, nominal velocities are
10
 only transfers heat from one side and therefore the heat load is reduced.
When the plate unit is arranged for multiple pass use, a further correction factor
must be applied. Even when two passes are countercurrent to two other passes, at
least one of them must experience co-current flow. This correction factor is shown
in Fig. 2 against a number of heat transfer units (HTU = temperature rise of the
process fluid divided by the mean temperature difference).As indicated, whenever
unequal passes are used, the correction factor calls for a considerable increase in
area. This is particularly important when unequal flow conditions are handled. If
high and low flow rates are to be handled, the necessary velocities must be main-
tainedwith the low fluid flow rate by using an increased number of passes. Although
the plate unit is most efficientwhen the flow ratio between two fluids is in the range
of 0.7-1.4, other ratios can be handled with unequal passes. This is done, however,at
the expense of the LMTD factor.
less plate fouling
The question of how to specify a fouling resistance for a plate heat exchanger is dif-
ficult to resolve. Manufacturers generally specify 5%excess HTU for low fouling
duties, 10%for moderate fouling, and 5 2 0 % excess for high fouling. The allowed
excess surface almost always is sufficient even though in many cases, it represents
-
a low absolute value of fouling in terms of (Btu/hr ft2 OF)-? If a high fouling resis-
tance is specified, extra plates have to be added, usually in parallel. This results in
lower velocities, more extreme temperatures during startup, and the probability of
higher fouling rates. Because of the narrow plate gaps and in particular, because of
the small entrance and exit flow areas, fouling in a plate usually causes more prob-
lems by the increase in pressure drop and/or the lowering of flow rates than by
causing large reductions in heat transfer performance. Increasing the number of
plates does not increase the gap or throat area, and a plate unit sometimes can foul
more quickly when oversurfad.
Unless the customer has considerable experience with both his process and the
plate heat exchanger, he should allow the equipment manufacturer to advise on
fouling and the minimum velocity at which the exchanger should operate.
lamina+flow
The other area suitable for the plate heat exchanger is that of laminar flow heat
transfer. It has been pointed out already that the Paraflow can save surface by han-
dling fairly viscous fluids in turbulent flow because the critical Reynolds number is
low. Once the viscosity exceeds 20-50CP,however, most plate heat exchanger de-
s&s fall into the viscous flow range. Consideringonly Newtonian fluids since most
chemical duties fall into this category,in laminar ducted flow the flow can be said to
be one of three types: 1)fully developed velocity and temperature profiles (i.e., the
limiting Nusselt case), 2) fully developed velocity profile with developing tempera-
ture profile (i.e., the thermal entrance region),or 3)the simultaneous development of
the velocity and temperature profiles.
The first typeis of interest only when considering fluids of low Prandtl number
and this does not usually exist with normal plate heat exchanger applications. The
third is relevant only for fluids such as gases which have a Prandtl number of about
one. Therefore, consider type two.
I2
I As a roughguide for plate heat exchangers, the rateof the hydrodynamicentrance
length to the correspondingthermal entrance length is given by

--
rLth -1.7Pr
Lhyd
Plate heat transfer for laminar flow follows the Dittus Boelter equation in this form:

where L = nominal plate length

y tne range 1.m to 4.50)
c = constant for each plate ( u s w ~ in
x = exponent varying from 0.1to 02 depending upon plate type
~

&pressure loss, the friction factor can be taken as f = a where a is a constant

charactexistic of the plate. Re
It can be seen that for heat transfer, the plate heat exchanger is ideal because the
1 dued d is small and the filmcoefficients are proportional to d -23. Unfortunately,
hm,thepressure loss is proportionalto (d)-4and the pressure drop is sacrificed
the heat transfer.
1 thesecorrelatbs, it is possible tocalculate the film heat transfer coefficient
re loss for laminar tlow. This coefficient combined with the metal
t for the service fluid together with the foul-
the overall coefficient. As with turbulent
flew, an a f i m n c e has the LMTD to allow for either end effect
!
correction for small plate packs and/or concurrency caused by having concurrent
flow in some passages. This is particularly important for laminar flow since these
exchangers usually have more than one pass.
1
lnd liquid/liquid,
Over many years, AFVhas built up considerableexperienain the designand use of
paraflow plate heat exchangersfar process a p p l i c a b s that fall outsidethe "rial
turbulent liquid flow that is common in chemical operations. The hafhw, for
example, can be used in laminar flow duties, far the evaporation OC ftuidswith rela-
tively high viscosities, for cooling various gaaies, and for condensingappliicatiOns
where pressure drop parametem are nut @vt?r&restrictive.
7.
cQnd;nsing
of the most important heat trnafer
tgnbf vapors - a duty which &en is Gamiad out
, exchanger but is entirely feasible in the plate type
! determining factor is pressure drop-
For those condensingduties where pqrmissible
there is no doubt but that the tubular unit is most efficieJ3t, Under such pr;sgaarpe
, drop conditions,only a portion of the kn@h of a Paraflow plate w ~ l be d rteed anB
substantial surface area would be wasted. However, when less restxidive premrre
dropg are available the plate heat e x c h a q p becomes an excellent condenser since
~

very high heat transfe$--ts am obtainedand the cusatioa am becarried

out in a single pass across the plate.
 pressuredropof condensingva
The Dressure drop of condensing ,
steam in the passages of plate heaj
exchangers has been investigated
experimentallyfor a series of different
Paraflow plates. As indicated in Fig. 3
which provides data for a typical unit,
the drop obtained is plotted against
steam flow rate per passage for a
number of inlet steam pressures.
It is interesting to note that for a FIG.3 Ste
set steam flow rate and a given duty, I
the steam pressure drop is higher COUNTERCURRENT -
when the liquid and steam are in
countercurrent rather than cocurrent --. I
----
CO-CURRENT ----

L-
flow. This is due to differences in
temperature profile. - I

Fig. 4 shows that for equal duties L

and flows, the temperature difference Ec
for countercurrent flow is lower at
the steam inlet than at the outlet with
most of the steam condensationtaking
place in the lower half of the plate. PLATE LENGTH
The reverse holds true for cocurrent
flow. In this case, most of the steam FIG.4 Temperatureprofile during
condenses in the top half of the plate, condensation of steam
the mean vapor velocity is lower,and a reduction in pressure drop of between 10-4096
occurs. This difference in pressure drop becomes lower for duties where the final
approach temperature between the steam and process fluid becomes larger.
The pressure drop of condensingsteam therefore is a function of steam flow rate,
pressure and temperature difference. Since the steam pressure drop affects the
saturation temperature of the steam, the mean temperature difference in turn
becomes a function of steam pressure drop. This is particularly important when
vacuum steam is being used since small changes in steam pressure can give signifi-
cant changes in the temperature at which the steam condenses.
By wing an APV computer program and a Martinelli Lockhart type approach to
the problem, it has been possible to correlate the pressure loss to a high degree of
aqxracy. Fig. 5 cites a typical performance of a steam heated Series R4 Paraflow.
From this experimental run during which the exchanger was equipped with only a
small number of plates, it can be seen that for a 4-5 psi pressure drop and above, the
plate is completely used. Below thaffigure, however, there is insufficient pressure
drop available to fully use the entire plate and part of the surface therefore is flooded
to reduce the pressure loss. At only one psi allowable pressure drop, only 60%of the
plate is used for heat transfer which is not particularly economic.
This exampleFhowever,well illustrates the application of a plate heat exchanger
to condensingduties. If sufficient pressure loss is available,then the plate type unit
is agood condenser. The overall coefficientof 770 Btu/hr * ft2 OF for 4-5 psi pressure
- -
loss is much higher than a coefficient of 450-500 Btu/hr ft2 OF which could be ex-
14
 5,000 lbsl
et water temperature 216'1
et steam temperature 250"
tal number of F'-'--

Pressure loss psi

pected in a tubular exchanger for this type of duty. However, the tubular design
would, for shell side condensation, be less dependent on available pressure loss and
for a one psi drop, a 450-500Btu overall coefficient still could be obtained. With the
plate, the calculated coefficient at this pressure is 746 Btu but the effective coeffi-
-
cient based on total area is only 605l of that figure or 445 Btu/hr ft2 OF.
0

gas cooling
Plate heat exchangers also are used for gas cooling with units in service for cooling
moist air, hydrogen, and chlorine. The problems are similar to those of steam heat-
ing since the gas velocitychanges along the length of the plate due either to condensa-
tion or pressure fluctuations. Designs usually are restricted by pressure drop so
machines with low pressure drop plates are recommended. A typical allowable
pressure loss would be 0.5 psi with rather low gas velocities giving overall heat
transfer coefficients in the region of 50 Btu/hr. ft2 OF.
evaporating
The plate heat exchanger also can be used for evaporation of highly viscous fluids
when as a Paravap the evaporation occurs in the plate or as a Paraflash the liquid
flashes after leaving the plate. Applications generally have been restricted to the
soapand food industries. The advantageof these units is their ability toconcentrate
viscous fluids of UD to 5ooo centiwise.

It hAs been shown; therefore, that the plate heat exchanger is a relatively simple
machine on which to carry out a thermal design. Unlike the shell side of a tubular I
exchanger where predicting performance depends on baffldshell leakage, baffle/
tube leakage and leakage around the bundle, it is not possible to have bypass
streams on a Paraflow. The only major problem is that the pressure loss through the
ports can cause unequal distributibn in the plate pack. This is overcome by limiting
the port velocity and by using a port pressure loss correlation in the design to allow
for the effect of unequal distribution.
The flow in-a plate also is far more uniform than on the shell side. Furthermore,
there are no problems over calculation of heat transfer in the window, across the
bundle or of allowing for dead spots as is the case with tubular exchangers. As a
result, the prediction of performance is simple and very reliable once the initial
correlations have been established. I

Is/
- I
 the problemof fouling

ty, variability and the need tocarry out experimental work on

A Typical waterlwater tubular design B R405 Paraflow, waterlwater duty

- -
Clean overall coefficient 500 Btu/hr ft2 "F Overall coefficient lo00 B a r ft2 O F
Single pass - pressure loss 9 psig

te that for a typical waterlwater duty in a plate heat exchanger, it would be

necessary to double the size of the unit if a fouling factor of .0005was used on each
side of the plate (Le. a total fouling of .001).
Although fouling is of great importance, there is relatively little accurate data
availableancttherather conservative fwresquoted in Kern (process Heat Transfer)
are used all too frequently. It also may be said that many of the high fouling resis-
tances quoted have been obtained from poorly operated plants. If a clean exchanger,
for example, is started and run at the designed inlet water temperature, it will
e x d its duty. To overcome this, plant personnel tends to turn down the cooling
water flow rate and thereby reduce turbulence in the exchanger. This encourages
fouling and even though the water flow rate eventually is turned up to design, the
damage will have been done. It is probable that if the design flow rate had been
maintained from the onset, the ultimate fouling resistance would have been lower.
A similar effect can happen if the cooling water inlet temperature falls below the
des& f m and the'flow rate is again turnddown.
six types of fouling
& d l y speaking, the typesof foulingexperienced in most CPI operations can be
divided into six faiI'ly distinct categories. First is crystallization- the most common
type of fouling which occurs in many process streams, particularly cooling tower
water. Frequently superimposed with crystallizationissedimentation which usually
is caused by deposits of particulate matter such as clay, sand or rust. From chemical
reactionond~lynrerization oftencomes a build up of organic products and polymers.
The surface teniperature and presence of reactants, particularly oxygen,can have a
very significant effect. cokingoccurs on high temperature surfaces and is the result
of hydrocarbon deposits. Orgtznic material growth usually is superimposed with
crystallization and sedimentation and is common to sea water systems. And corm
 used in these units are noncorrosive.
The other four types of fouling, however,
are most important. With certain fluids
such as F l i n g tower water, foulingcan

From Fa.1,it is apparent that the fouling process is time dependant with zero foul-
ing initially. The foulingthen builds up quite rapidly and in most cases, levels off at a
certain ti otic value as represented by curve A. At this point, the rate
of deposi that of removal. Not all fouling levels off, however, and
curve B shows that at a certain time the exchanger would have to be taken off line
for cleaning. It should be noted that a Paraflow is a particularly useful exchanger for
this type of duty because of the ease of access to the plates and the simplicity
of cleaning.
In the case of crystallizationand suspended solid fouling, the process usually is of
the type A. However, when the fouling is of the crystallization type with a pure
compound crystallizingout, the fouling approaches type B and the equipment must
. be cleaned at frequent intervals. In one particularly severe foulingapplication, three
Series HMB Paraflows are on a 4% hour cycle and the units are cleaned in place for
1%hours in each cycle.
Biologicalgrowth can present a potentially hazardous foulingsince it can provide
a more sticky type surface with which to bond other foulants. In many cases,
however, treatment of the fluid can reduce the amount of biologicalgrowth. The use
._
of germicidesor poisons to kill bacteria can help.
-.
lower resistance
&generally isconsideredthat resistanaduetofouHng~lolrarnith~~wplate
heat exchangers than with tub& units. This is the result of f k Paraaew
advantages:
1. There is a high degree of turbulence which increases the rate of foulant
removal and results in a lower asymptoticvalue of fouling resistance.
2. The velocity profileacmes a plateisgcood. There are noaonesof low velocity
comparedwith certain areas on the shell side of tubular exchangers.
3. Corrosion is maintainedat an absolute minimum.
4. A smooth heat transfer surface can be obtained.
5.In certain m l i n g duties using water to cool organics,the very water

r:
 I - TUBUL

Effect of velocit
and turbulence

film coefficientmaintains a moderately low metal surface temperature which

helps prevent crystallization growth of the inversely soluble compoundsin the
cooling water.
The, most important of these is turbulence. HTRI (Heat Transfer Research In-
corporated)has shown that for tubular heat exchangers,foulingis a function of
velocity and friction factor. Although flow velocities are low with the plate
exchanger, friction factors are very high and this results in lower foulingresist
The effect of velocity and turbulence is plotted in Fig. 2.
Marriot of Alfa Lava1has produced a table showing values of foulingfor a numbe
of plate heat exchanger duties.
'
Ruid Fouling resistance Fluid Fouling resistance '
(hr .ft2 * "F/Btu) (hr ft2 * OF/Btu)
Water Oils, lubricating .00010-.00025
Demineralized or Oils, vegetable .Ooo1o-.m
distilled . m 5 Solvents, organic .oooO5-.00015
soft .OOolO Steam . m 5
Hard .00025 Process fluid, general .oooO5-.00030
Cooling tower (treated) .00020
Sea (coastal)or estuary .00025
Sea(ocean) .OOol5
River, canal .00025
Engine jacket .m
These probably represent about one half to one fifth of the figures used for tubu-
lars as quoted in Kern but it must be noted that the Kern figures probably are con-
servative, even for tubular exchangers.
APV, meanwhile, has carried out test work which tends to confirm that fouling
v&ies for different plates with the more turbulent type of plate providing the lower
foulingresistances. In testingan R405 heatinga multi-component aqueous solution
containing inverse solubility salts -(i.e., salts whose solubility in water decreases
with increasing temperature), it was learned that the rateof foulingin the Paraflow
was substantially less than that inside the tubes of a tubular exchanger.The tubular
Unit had to be cleaned every 3 or 4 days while the Paraflow required cleaning about
once a month. +-

Additional tests on coolingwater foulingwere sponsored by APVat Heat Transfer

/ Research,Inc. (HTRI) with the test fluid beinga typical treatedcooling tower water.
Results of these experiments are as follows:
18
objectives
In many streams, fouling is an unavoidableby-productof the heat transfer proc
Fouling depositscan assume numerous types such as crystallization, sediFentat
corrosion, and polymerization. Systematic research on fouling is relatively recen
and extremely limited (Refs. 1,2,3)and almost exclusively concentrated on water
the most common fluid with fouling tendencies. The few research data which exis
usually are proprietary or obtained from qualitative observations in plants. How-
ever, it generally is recognized that the prime variables affecting foulingbuildup are
flow velocity, surface temperature and surface material. In the case of water, the
water quality and treatment must also be considered.
The importance of fouling on the design of heat exchangers can be seen from the
rate equation

where U -
= overall heat transfer coefficient (Btu/hr - ft2 OF)
-
h,, h, =film coefficientsof the two heat transferring fluids (Etu/hr ft2 * -r 1
= fouling resistance (hr * ft2* OF/Btu)
Rf
It is obviousfrom Equation 1that the higher the film coefficients,thegreater effect
the foulingresistance will have on the overallcoefficient and therefore on the size of
the exchanger.
In tubular heat exchangers, water-side heat transfer coefficients in the order of
magnitudeof 1000Btuhr sft2-'Farequitecommon. In plateexchangers,thecoeffi-
-
cients are substantially higher, typically around 2000 Btu/hr ft2 O F . Assuming
that both types of equipment operate with water-water systems,overall clean coeffi-
-
cients of 500 and 1000Btu/hr f t 2 O F respectivelyare obtained. Using a typical foul-
-
ing resistance of 0.001 hr * ftz 'F/Btu (equal to a coefficient of 1000B t u h r - ft2* OF),
the inclusion of the fouling will cause the tubular exchanger size to increase by a
factor of 4 while for the plate exchanger,the correspondingfactor is 7.
This example clearly demonstrates the crucial importance of fouling,especially in
plate exchangers. Yet, fouling resistances which are unrealistically high often are
specified and invariably have been based on experiences derived from tubular
equipment. The common source of water fouling resistances is TEMA(4' which
recommends values of Rf spanning a tremendous range between 0.0015 and 0.005
-
hr * ft2 OF/Btu.
Flow velocity, as.mentioned earlier, is a yucial operating parameter which in-
flueices the foulingbehavior.For flow insidebf tubes, thedefinition of flow velocity
and the velocity profile is straightforward. But in plate exchangers, flow velocity is
maracterized by constant fluctuationsas the fluid passesover the corrugations. It is
postulated that this induces turbulence which is superimposed on the flow velocity
as a factor which diminishes fouling tendencies. This has been observed qualita-
tively in practical applicationsand confirmed by unpublished APV research.
test apparatus and conditions
The plate heattxchanger (PHE) tested was an APV Model 405 using APV Type R40
stainless s t k l plates 45 inches high and 18inches wide. The plate heat transfer area
was 4 sq ft with a nominal gap between plates of 0.12 inch. Due to the plate corruga-
tions, the maximum gap was 0.24 inch. Seven plates were used creating three
19
countercurrent passages each of the cooling water and the heating medhm. A
schematic diagram of the installation is shown in Fig. 3.
The PHE was mounted on the HTRI Shellside Fouling Research Unit f-SFRU)
together with two small stainless steel shell and tube exchangers. The PHE was
heated using hot steam condensate and the fouling was determined from the
degradationof the overall heat transfer coefficient. Simultaneously,tests also were
run on an HTRI Portable Fouling Research Unit (PFRU) which uses electrically
heated rodswith the cooling water flowing in an annulus. The SFRU and PFRU are
described in a paper by Fischer et a1.n
The fouling tests were conducted at a major petrochemical plant in the Houston,
Texas, area. The test units were installed near the cooIing tower basin of the plant.
The cooling tower operatingcharacteristicsare summarized in Table 1.The 140,OOO
GPM system had two 1600 GPM sidestream filters. Filter backwash accounted for
most of the blowdown. The makeup water for the cooling water system came from
several sourcesand was clarified with alum outside the plant. The water treatment
used is summarized below:
Chromate-zincbased inhibitor for corrosion control (20-25 ppm chromate)
Organic phosphonate and polymer combination as dispersant (2 ppm organic
phosphonate)
*Polyphosphateas anodic passivator (5-6ppm total inorganic phosphate)
Chlorine for biological control (continuousfeeding of 386 lbdday)
*Biocidefor biological control (15 ppm biocide ondweek)
*Sulfuricacid for pH control (pH of 6 to 6.5)
Le& procedure
The velocity for the PHE is defined by V=
PAC
whereV =velocity
w =massflowrate
p =fluiddensity
4 =cross-sectionalflow area

FIG.3 APV plate heat exchanger with 3 parallel munterarrent parallel passages
for each of the cooling water and the hot water.

20
Tower and Circulation System

birculation Rate
Temperature Difference Across
Number of Sidestream Filters

Corrosion Rate from Carbon Steel

Coupon Tests

Water Description

. _,_._ __..__.._._:ion UnclarifiedSan Jaci

Blowdown from coolers Clarified San Jacin

and sidestream filter=

ypical Composition
otal hardness as CaC03, ppm
:alcium as CaCo3,ppm
lagnesium as CaCO,, ppm
Methyl Orange Alkalinity as CaC03, ppm
Sulfate as SO4,ppm
Chloride as CI, ppm
Silica as SO2,ppm
Total inorganic phosphate as
Orthophosphate as PO4,ppm
PH
Specific conductance, microm

I *..
Cooling water system characteristics

The cross-sectionalflow area is based on the compressed gasket spacing. In other

words. the flow area is computed as if there were no corrugations but smooth plates
instlad. The surface temperature is defindas the fluid-deposit interface tempera-
ture. Since the PHE was heated by constant temperature steam condensate, the
metal temperature remained constant and the surface temperature decreased as
foulingbuilt up. In addition, due to the counterflow nature of the PHE, the surface
temperature was not constant fmm inlet to outlet. The surface temperature re-
ported here is the initial surface temperature at the midpoint of the plate.
Conditions of operation were selected so that all the SFRU test exchangers
operated roughly at the same nominal flow velocity and surface temperature. Dur-
ing the operatbn, weekly water samples were taken for chemical analysis. At the
end of each test, deposits were photographed and sampled for chemical analysis.
The exchanger plates then were cleaned and the unit reassembled and new test
conditionsestablished. *.
 test and result desc
Five test series were performed testing several velocities and surfacetemperatures.
Typicalfouling time histories are shown in Fg.5dongwith theoperatingconditions.
Notice that the PHE fouling resistances establish a stable asymptotic value after
about 600 to 900 hrs. of operation. The results of the other tests are shown in Fg.6
only as the values of the asymptotic fouling resistances, as these are the data re-
quired for design.
The chemical analysis of the fouling deposit was made after each test and the re-
sults are summarizedin Table 2. The primary elementsfound were phosphorus,zinc,
and chromefrom the water treatment and silicon from suspended solids in the water.
-The photograph of the plates of the PHE shown in Fg. 4 at the termination of Test
2 indicates that fouling only occurred in the upper third of the plates. This is the
region near the hot water inlet and cold water outlet, a region of high surface tem-
perature. Figure 5 shows a surface temperature profile for Test 2 conditions of 2.8
Wsecand an inlet bulk temperatureof W F .From the water chemistry parameters,
the saturation temperature for calcium phosphate above which precipitation is
expected was calculated to be 163'F6).The shaded region on Fg.7 indicates the ex-
pected precipitation region and correspondsclosely to the fouled region seen in Fg.4.
The surfacetemperatures during someof the tests were higher than thoseexperi-
e n d in normal PHE operation. This is the result of the setup condition criteria that
the three exchangers on the SFRU operate at the same midpoint surface tempera-
ture. Since the ratioof flow rateto-surfacearea for the shell side of the shell and tube
exchangem is twice that for the PHE, the mling water had a longer residence time
in the PHE than in the tububr exchanger.As a result, the lower flow rateto-surface

FIG. 4 Fouled plates, Test 2

-
f'
22
 c>
.i
a

5.
hSl
&.

FIG. 6
Asymptotic fwtingreaistance
versus velocity with surface
tempemturema$taeter.

I Series 2 1 Series 3 I Series 4 I Series 5 I Series 6

Loss on Ignition
Phosphorus
Aluminum
Silicon 18 16 13
1 Calcium 4 15
Chrome 5 I 13 I 15 15 I 9
Iron 2 7 ?. I 4 5 13
1 Zinc 2 1 2 0 1 1 2 1 1 3 1 7
Sodium
Maonnsium

1.The loss on ignition (LOI) is performed by drying the sample to constant weight at
220F (105OC) and then combusting at 1470F (800OC). LO1 represents bound water
and organics present in the deposit. An organic test on samples of Test Series 5
showed that about k of the LO1 was organic.
2. Aluminum, silicon, and iron come from the suspended solids. Aluminum comes from
operaron of the clarifier. The silicon is not from magnesium silicate since little
magnesium is in the analyses.
3.Chrome, zinc, and phosphorus are derived from the water treatment.
4. Low vatues of calcium indicate little or no tricalcium phosphate. Most of the calcium
comes from association with zinc and polyphosphates in the treatment.
 W'

9
W

2
5
9

:.
5
v)
130

120

110

40

OUTLET

FIG.7 Surface temperatureprofile with indicated

calcium phosphate precipitation
s a larger temperature rise of the cooling water. The result is the
temperature profile s .7. Although cocumnt operation
ted asa correctionfor this d higher hot water flow
adable. Cmsquentty, some of these tests was
slightly penalized due to part of the surface being above the critical temperature for
tricalcium phosphate precipitation, a condition which normally would not be en-
countered in industrial operations.
It is difficult to make meaningful comparisons between the fouling tendencies of
different types of heat exchangers. Unless carefullydone, the results are misleading.
This is particularly true when attempting to compare the foulingexperience in the
PHE to that of tubeside operation as in the PFWU. Thegeometries, surface area, and
operational characteristics are very different even though the coolingwater maybe
the same. Such comparisons often are attempted using TEMA(4)recommended
fouling resistances.
TEMA recommends a foulinn resistance of 0.002 hr * ft2 * "F/Btu for the water
system used. However, the maximum fouling resistance measured in the PHE was
-
less than 0.0005 hr * ft2 "F/Btu - only 25 percent of the TEMA recommendation.
This confirms the earlier assumption that applying TEMA-recommended fouling
resistances for a shell and tube exchanger to the PHE seriously handicaps the perfor-
mance ratings. Becauseof the inherently high heat transfer coefficients, the effects
of fouling resistances are more pronounced. Consider a PHE operating at 1.5 ft/sec.
From Table 3, the fouled overall coefficient using a TEMA fouling resistance of
-
0.002 hr ft2- "F/Btu is about one half of the overall coefficient using the measured
24
 I

and tube exchangers.

Figure 8 compares the performance of the PHE with typical tubeside data and
with the TEMA-recommendedfouling resistance.
. A direct comparison of the PHE and the shellsideexchangers can be made through
the deposit analyses. A sample analysis is given in Table 4. The amounts of
phosphates, calcium, chrome, and zinc are somewhat higher for the PHE. The
operation of the PHE at a higher surface temperature than the shellside units re-
sulted in more crystallization fouling. However, the amount of silicon for the shell-
side unit is much higher than for the PHE. The high silicon concentration is due to
sedimentation in the baffle-shellcomers of the bundle, while the high turbulence
promoted by the corrugated plates in the PHE minimizes sedimentation fouling. In
-.
heavily sediment-ladenwaters, the PHE would be especially superior.
conclusions
Although the tests were performed on one water system only, experience with
tubular data indicates that the overall trends generally are valid, i.e., plate ex-
changers should be designed to substantially lower values of fouling than would be
used on tubular equipment for the same stream. However, the comparison is not
always that straightforward,as flow velocity itself is not a valid criterion. The wall
shear stress in a PHE operating at 2.8 ft/sec. is equivalent to that in a tubeside
exchanger operatingat 8.2 ft/sec. On the basis of typical industrial velocities, it was
found that the PHE at 1.5ft/sec. fouls about one half as much as a tubesideexchanger
operating at 5.9 ft/sec. Furthermore, because of the high heat transfer coefficients
 TABLE 4 Comparisonof deposit
analyses from plateexchanger
and shellsidetest exchanger
for Series 5

typical to the PHE, surface temperature may differ from those in tubular equipment
and should be carefully watched.
The turbulence inducing corrugation pattern prevalent in heat exchanger plates
produces very high local velocities. This results in high friction factors and there-
fore, high shearing. This high shear, in turn, results in less fouling in plate ex-
changers than in tubular units. For cooling water duties, Heat Transfer Research
Incorporated(HTRI)has shown that the plate exchanger fouls at a much lower rate
than either the tubeside or shell side of tubulars. TEMA recommendations for foul-
ing therefore should never be used for plate units since they probably are five times
the value found in practice. In particular, the plate exchanger is far less susceptible
to foulingwith heavy sedimented waters. This is contrary to popular belief. It must
be noted, however, that all particulate matter must be significantlysmaller than the
plate gap and generally particles over 01 inch in diameter cannot be handled in any
plate heat exchanger.
It has been shown that quoting a high fouling resistance can negate a plate heat
exchanger design by adding large amounts of surface and thereby overriding the
W i t s of the high coefficients. Fouling design resistance, therefore, should be
chosen with care, keeping in mind th with a Paraflow unit, it always is possible to
3
add or subtract surface to meet exac ouling conditions.
Disclaimer: As a general policy, HTRI as the project operator disclaims responsi-
bility for any calculations or designs
R&"
1. Tabodc, J., Aoki, T.;Ritter, R.B., Palen, J.W., ahd Knudm, J.G., "Fouling - The Major
UnresolvedProblem in Heat Transfer, Parts Iand 1I," CEP, 68, Nor. 2 817 (1972)
2 Wtor, J.W., Marner, W.J., and Ritter, R.B., 'The History and Status of Research in Fouling of
-
H a t Exchangers in Cooling Water Servi&," Paper No. 7SCSME/CSChE 19 presented a t the
18thNational Heat Transfer Conference, St. Louis, Missouri, August 8-11,1976
3. Morse, Robert W. and Knudsen, JameiG., "Effect of Alkalinity on the Scaling of Simulated
Cooling Tower Water," Paper No. 76SME/CSChE - 24 presented at the 16th National Heat
Transfer Conference, St. Louis, Missouri, August 8-11,1976
4. Sandbrdsof Tubular Exchanger ManufacturersAssociation.5th Ed., New York, 1968
5, F h r , P., Suitor,fW., and Ritter, R.B., "Fouling Measurement TechnquesandApparatus,"CEP,
71, No. 7,66 (1975)
8. Grwn, J. and Holmes, J.A., "Calculation of the pH of Saturation of Tricalcium Phosphate," J.
A-r Water Works Assn., 39,No. 11 (1947)
 comparing plate
and tubular exchangers
In forming a comparison between plate and tubular heat exchangers, there are a
number of guidelines which will generally assist in the selection of the optimum
exchanger for any application.In summary, these are:
1. For liquidlliquid duties, the plate heat exchanger usually has a higher overall
heat transfer coefficient and often the required pressure loss will be no higher.
2. The effective mean temperature difference will usually be higher with the
plate heat exchanger.
3. Although the tubeis the best shapeof flow conduit for withstanding pressure,
it is entirely the wrong shape for optimum heat transfer performance since it has
the smallest surface area per unit of cross sectional flow area.
4. Because of the restrictions in the flow area of the ports on plate units, it is
usually difficult,unless a moderate pressure loss is available,to produce economic
designs when it is necessary to handle large quantities of low density fluids such
as vapors and gases.
5. A plate heat exchanger will usually occupy considerablyless floor space than a
tubular for the same duty.
6. From a mechanical viewpoint, the plate passage is not the optimum and gas-
keted plate units are not made for operating pressures much in excess of 300psig.
7. For most materials of construction, sheet metal for plates is less expensiveper
unit area than tube of the same thickness.
8. When materials other than mild steel are required, the plate will usually be
more economical than the tube for the application.
9. When mild steel construction is acceptable and when a close temperature ap-
proach is not required, the tubular heat exchanger will often be the most economic
solution since the plate heat exchanger is rarely made in mild steel.
10. Plate heat exchangers are limited by the necessity that the gasket be elasto-
meric. Even compressed asbestos fiber gaskets contain about 6% rubber. The
maximum operating temperature therefore is usually limited to 500F.
heit transfer coefficients
Higher overall heat transfer coefficientsare obtained with the plate heat exchanger
Tompared with a tubular for a similar loss of pressure because the shell side of a
tubular exchanger is basically a poor design from a thermal point of view. Consider-
able pressure drop is used without much benefit in heat transfer due to the turbu-
lence in the separated region at the rear of the tube. Additionally, large areas of tubes
even in a well designed tubular unit are partially bypassed by liquid and low heat
transfer areas thus are created.
Bypassingin a plate type exchanger is less of a problem and more use is made of
the flow separation which occurs over the plate troughs since the reattachment
point on the plate gives rise to an area of very high heat transfer.
For most duties, the fluids have to make fewer passes across the plates than would
be required through tubes or in passes across the shell. Since a plate unit can carry
27
 FIG.1 Duty: Demineralized waterlsea water. Tocool 864,000lb/hr of water from 107F
to 82.4"F using 773,000lbs/hr of 72OF sea water

I
TUBULAR DESIGN PLATE DESIGN
rneaaunE LVSS I rneaaune LUSS

13.380 53.382 320A120 42 305 301 .3 2.2 92 HX

hydrocarbon water 120892
B 884,ooO 77,300 107 A82.4 19 19 12,100 10.8 8.8 3,630 R10
water sea water 99.1 B 71.6
c 17,500 194.ooO 140AlW 2.9 4.5 1.830 3.0 3.7 445 R5
solvent water 95879
D 148,650 472,500 !222A100 10.8 8.0 1,500 4.5 10 380 R4
desalter saltwater WEK
effluent
FIG.2 CASE STUDIES OF PLATE HEAT EXCHANGERS AND TUBULAR DESIGNS
All tubular designs were carriedout with the aid of the HTRI program ST3

out thedutywithonepass for both fluidsin manycases,thereductionin thenumber

of required passes means less pressure lost due to entrance and exit losses and con-
sequently, more effective use of the pressure.
mean temperature difference
A further advantageof the plate heat exchanger is that the effective mean tempera-
ture difference is usually higher than with the tubular unit. Since the tubular is
always a mixture of cross and contra-flow in multi-pass arrangements, substantial
correction factors have to be applied to the logmean temperature difference. In the
plate heat exchanger where both fluids takethe same number of passes through the
ut& -$heLMTD correction factor is usually in excess of 0.95. As is illustrated in
Fie. 1,this factor is particularly important when a close or relatively close tempeta-
ture approach is required.
In practice, it is probable that the-& water flow rate would have been increased
to reduce the number of shells in series if a tubular had to be designed for this duty.
While this would reduce the cost of the tubular unit, it would result in increased
operating costs. 1

design case studies

Fig.2 covering a number of case studies on platevs. tubular design demonstratesthe
remarkable heat transfer performance that can be obtained from Paraflow units.
28
 I

If a close temperature approach is required, however, the plate unit always will cost
less and where stainless steel or more exotic materials are required for process
reasons, the plate unit usually will cost less.
physical size . 1
.

One important advantage of the plate over a tubular unit is that for a particular
duty, the plate heat exchanger will be physically smaller and require far less floor
space. This is shown most graphicallyin Fg.3where a SeriesR405 Paraflow is being
installed next to tubular units that need twice the amount of space for the identical
duty. It is further illustrated in thevolumetric comparisons of Fig. 4.
If the tightened plate pack of a Paraflow is regarded as a rectangular box, each
cubic foot contains from 50 to 100 ft2 of heat transfer area according to the type of
plate used. Allowing for metal thickness, the contained liquid is some 80%of this
volume. Thusifhe total of both heating and cooling media is about five gallons. EX
pressed in another way, the liquid hold up per square foot for each stream varies
according to plate type from about .06 gallons down to .03gallons.
By comparison,one cubicfoot of tubular exchanger of equilateral triangular pitch
29
 TUBE

Liquid containment
Heat transfer Heat transfer per sq ft of
area per cubic Liquid contained area per cubic heat transfer
ite footof per sq ft of heat Tube tube pitch foot of area (average of t
3h exchanger transfer area dia. tube dia. exchanger both sides)

r n .2
0.06 gal 2 1.5 10 ft2 0.32 gal

0.03 gal % 1.5 40 ft2 0.085 gal

FIG.4 Volumetric comparison,plate vs. tube

with a tube pitch/tube diameter ratio of 1.5has a surface area of 10ft2for 2 inch OD
tubes or 40ft2for 1/2 inch OD tubes. The average contained liquid is proportionately
0.27 gallons/ft2 down to 0.07 gallons/ft2 of heat exchanger area with no allowance
for the headers. If the heat transfer coefficient ratio between platehubular is con-
servatively taken as two, the plate exchanger volume to meet a given duty varies
from 1110th to 115th of that of the tubular. For a lower tube pitch/tube diameter
ratio of 125, the comparison becomes 117th to 114th. These facts demonstrate why
the Faraflow plate heat exchanger is referred to as compact:
thermal limitations
While it would appear offhand that the plate heat exchanger always provides a better
performance at usually a lower price than the tubular exchanger, consideration
must be given to the thermal as well as mechanical limitations of the plate type
machine. These usually are based on allowable pressure loss.
For single phase liquiuiquid duties, the plate heat exchanger can be designed for
moderately low pressure loss. However, if the pressure loss across any plate passage
which has liquid flowing downward is lower than the available liquid static head,
the plate will not run full and performance therefore will be reduced. This is termed
low plate rate. Use of a plate below the minimum plate rate is inadvisable since it
causesa wastage of surface area and results in unreliable operation. It is, however,
possible to function below the minimum plate fate in a single pass arrangementby
making sure that the low plate rate is operated with a climbing liquid flow.
these problems are not quite so severe with a tubular exchanger and therefore,
aperation at a moderately lower available pressure loss is possible.
-_
conclusion
To summarize, the gasketed plate heat exchanger generally will be the most em-
nomical heat exchanger for liquiuiquid duties providing the material of construc-
tion is not mild aeel and providing the operating temperatures and pressures are
below 500F and 300 psig respectively. For other types of duties such as gas cooling,
condensation or boiling, the plate heat exchanger can be a very economical type of
unit if the pressure loss allowed is sufficient to utilize the very high heat transfer
performance characteristics of the plate.
30
recovery of process heat L
Since it is quite clear that never again will energy be as inexpensiveas in the past, it
therefore is necessary to conserve this natural resource-to recover more of the
process heat that currently is dissipated to waterways and atmosphere. Someof this
heat can be recovered with the aid of high efficiency heat exchangers which can
operate economically with a close temperature approach at relatively low pumping
powers. One type of unit that is particularly suited for this duty is the APVParaflow
plate heat exchanger.For many applications, this equipment can transfer heat with
almost true counter-current flow coupled with high coefficients to provide efficient
and inexpensiveheat transfer.
Unfortunately, the plate heat exchanger has been considered by many chemical
engineers to be suitable only for hygienicheat transfer duties. This, of course, is not
so. Nowadays, many more plate type units are sold for chemical and industrial use
than are sold for hygienicapplications. A further mistake is the claim that the plate
heat exchanger can be used only for duties when the volumetric flows of the two
fluids are similar. Again, this is not true, although it must be stated that the plate
heat exchanger is at its most efficient when flows are similar.
plate or tubular?
Since plate and tubular heat exchangers are the most widely used types of heat
transfer equipment, it is well to draw a brief comparison of their respective heat
recovery capabilitiesfor the energy-consciousplant manager.
While the plate heat exchanger does have mechanical limitations with regards to
withstanding high operating pressures above 300 psig, it is more efficient thermally
than shell and tube units, especially for liquidlliquid duties. In many waste heat
recovery applications, however, both pressure and temperature generally are mod-
erate and the plate type unit is an excellent choice since its thermal performance
advantage becomes very significant for low temperature approach duties. Higher
overall heat transfer coefficients are obtained with the plate unit for a similar loss of
pressure because the shell side of a tubular basically is a poor design from a thermal
point of view. Certain pressure drop is used without much benefit in heat transfer
on the shell side due to the flow reversingdirection after each cross pass. In addition,
even'in a well designed tubular heat exchaitger, large areas of tubes are partially
bypassed by liquid and areas of low heat transfer thus are created. Conversely, by-
passing of the heat transfer area is far less of a problem in a plate unit. The pressure
loss is used more efficiently in producing heat transfer since the fluid flows at low
velocity but with high turbulence in thin streams between the plates.
For most duties, the fluids also have to make fewer passes across the plates than
would be required either through tubes or in passes across the shell. In many cases,
the plate heat exchanger can carry out the duty with one pass for both fluids. Since
there are fewer-passes, less pressure is lost due to entrance and exit losses and the
pressure is used more effectively.
A further advantage of the plate heat exchanger is that the effective mean tem-
perature difference usually is higher than with the tubular. Since in multi-pass
arrangements the tubular is always a mixture of cross- and contraflow, substantial
 FIG. 1

I
1 2 Ib/tn* pressure loss 25 Ib/in? pressure loss
Heat
recovered
(Btulh) (%) Price

3,600,000 60 5,500

4,500,000 75 6,450

5,100,000 85 8.400

5,400,000 90 11,000
~

5,700,000 95 \ 5 0.83 4.16 1580 18,100

hot liquid 60,000 lblh of water at 200F

cold liquid 60,000 lblh of water at 60F

correction factors have to be applied to the log mean temperature difference. In the
plate unit for applicationswhere both fluids take the same number of passes through
the exchanger, the LMTD correction factor approaches unity. This is particularly
important when a close or even relatively close temperature approach is required.
thermal perlor lllallce h a ~ d
Although the plate heat exchanger now is widely used throughout industry, precise
thermal performance characteristics are proprietary and thus unavailable.It is pos-
sible, however, to size a unit approximately for turbulent flow liquidlliquid duties
by use of generalizedcorrelations which apply to a typical plate heat exchanger.The
basis of this method is to calculate the heat exchanger area required for agiven duty
by assuming that all the availablepressure loss is consumed and that any size unit is
available to provide this surface area.
For a typical plate heat exchanger,the heat transfer can be predicted in turbulent
flow by the following equation (1):

The pressure loss can be predicted from 2fG2L

AP=-
equations (2) and (3). gPDe ' (3)

Obviously, equation (1) cannot represent accurately the performance of the many
differenttypes of plate heat exchangers that are manufactured. However,plates that
have higher or lower heat transfer performance than given in equation (1)usually
will give correspondingly.higheror lower friction factors in equation (2). Experience
indicates that the relationship for pressure loss and heat transfer is reasonably con-
sistent for well designed plates. In the Appendix, equations (Al)-(A3) are further
developed to show that it is possible for a given duty and allowablepressure loss to
32
c

FIG. 2 Two shrouded Paraflow units provide an optimum return o estment with energy and cost
savingregenerationof 88 and 81 percent.

predict the required surface area. This technique has been used for a number of
years to provide an approximate starting point for design purposes and has given
answers to *20%.For accurate designs, however, it is necessary to consult the
manufacturer.

111 ~ I I Y ICLUVCI y dpplication, it always is necessary to consider the savings in

IIML
the cost of heat against the cost of the heat exchanger and the pumping of fluids.
Each case must be treated individually since costs for heat, electricity,pumps, etc.
willvary from location to location.
One point is obvious. Any increase in heak recovery, and thus heat load, results in
a decrease in LMTD and considering a constant heat transfer coefficient, subse-
quently in the cost of the exchanger. This effect is tabulated in Fig. 1. Because the
cost of an exchanger increases considerably for relatively small gains in recovered
heat above the 90% level, such applications even with the plate heat exchanger
must be studied closely to verify economic gain. The economic break-even point is
far lower for a tubular exchanger. Situations where it is advantageous togo above
90%recovery usually involve duties where higher heat recovery reduces subsequent
heatingor cdingof the process stream. High steam or refrigeration costs therefore
justify these higher heat recoveries.
As shown in Fig. 1, the cost of increasing heat recovery from 85%to 90%at a
constant pressure loss of 12 lbs/in2is $2600. From a practical standpoint, going from
90% to 95%requires a significantly higher pressure loss and nearly doubles the
 .-"-
I FIG,3. Heat recovery -comparison between plate and tubular heat exchangers

Duty. To heat 1,300,000 Ib/h of water from 73F to 97OF using 1 300.000 Ib/h of water at 107OF.
Available pressure loss 15 Ib/inz for both streams.
Heat recovery = 70.5%
Number of heat transfer units (HTU) - A

Tubular
Heat transfer area 13,100 ft'
Heat transfer coeffi 386 -
Btu/hr * ft2 OF
Heat transfer coeffi 271 Btu/hr. ft2. O F
Effective mean tem 8.7 "F
Pressure drop. hot 15/13 Ib/in*
Fouling resistance: 0.0005/0.000 (Btu/hr * ft2 * O F)-1
Pass arrangement 4 tube side baffled
3 shells in series
Approximate price $135,000
Mild Steel
This example demonstrates that
PHE with only a moderate pressure loss.

exchanger cost. However, even with this 95%heat recovery and assuming steam
costs at $6.OO/lOOO#, payback on the plate heat exchanger would take 530 hours.
The plate heat exchanger thus provides a most economic solution for recovering
heat. This degree of heat recovery cannot be achieved economically in a tubular
exchanger since the presence of crossflowand multipass on the tube side causes the
LMTD correction factors to become very small or, alternately, requires more than
one shell in series. This is shown in the Figure 3 comparison.
As detailed, this example illustrates that the plate heat exchanger has consider-
able thermal and therefore price advantage over the tubular exchanger for a heat
recovery of 70%.Since the overall heat transfer coefficient and the effective mean
temperature differenceboth are much higher for the plate unit, reduced surface area
is needed. Furthermore, because of crossflow temperature difference problems in
the tubular, three shells in series were needed to handle the duty within the surface
area quoted. Using only two shells would have resulted in a further 40%increase in
surface area.
The small size of the plate heat exchanger also results in a saving of space and a
lower liquid hold up. For this type of heat recdvery duty, a stainless steel plate heat
exchanger almost always will be less expensive than a mild steel tubular unit.
Mhough the tubular exchanger physically will be capable of withstanding higher
temperatures and pressures, there is a considerableand for the most part unneces-
sary penalty to pay for these features both in price and size.
For heat recovery duties in excess of 7096,the plate heat exchanger will become
increasingly more economic than the tubular.
typical applicatJons
One of the more common uses of regeneration is found in many of the nation's
breweries where it is necessary to cool huge amounts of hot wort before it is dis-
charged to fermentation tanks. Typical in scope is an operation where 850 barrels
per hour of wort (220,000lbshr) are cooled from 200F to 50F by means of 242,000
34
 l b s h r (R = 1.l:l) of water entering at 35F and being heated to 165F. The result:
approximately 33,000,000Btu/hr are saved, there is an excellent water balance for
use throughout the brewery, and no water is discharged to the sewer.
For a dairy, regeneration usually involves the transfer of heat from pasteurized
milk tocooler raw milk entering the system. After initially heating 100,OOOIbshr of
milk from 40F to 170F by high temperature, short time pasteurization, 90% regen-
eration permits cooling of the milk back down to 40F with a savings of 11,700,000
Btu/hr. Only 10%of the total heat or coolingmust be supplied and no coolingmedium
s u c h as city water is used and discarded.
Chemically, there are many and varied regenerative applications. For desalina-
tion, APV has supplied a number of Paraflows which are virtually perpetual motion
machines. These units achieve 95%regeneration in heating 87,000lbshr (175 GPM)
from 70F to 197F while cooling a secondary stream flowing at 78,000 lbshr from
214F to 73F. Savings in the Btu load in this case are 11,050,OOOper hour.
While flow rates in hot oil applications are quite low in comparison,temperatures
are very high. It is possible to cool 400 GPH of vegetableoil from 446Fto 266Fwhile
heating 400 GPH of oil from 200F to 400F with 80% regeneration.
And in the production of caustic soda where very corrosive product streams are
encountered, Paraflows with nickel plates are being used tocool 10,000lbshr of 72%
NaOH from 292F to 210F while heating 14,000 lbs/hr of
to 169F.
anddollarsavings
To examine a hypothetical case of Paraflow regeneration from the viewpoint of effi-
ciency and dollar savings, consider the following process duty:
DUTY Heat 100 GPM of fluid #1 from 50F to 200F while cooling 100 GPM of
fluid #2 from 200F to 50F.
Under ordinary conditions, steam required to heat fluid #1 would be in the nature of
7500 lbs/hr with an equivalent coolingrequirement for the second stream of 625 tons
of refrigeration. Using 85%regeneration, however, the energy needs are drastically
reduced

FLUID #1

7.5 x $6.00 = $45.OWhr

W.00x 24 x 306 = $3W,ZiB
1.125~$6.00=$6.7!Vhr
#Mxzy mm I
 "I
I I I
cool 100 GPM cool 100 GPM
200OF -+50' F 200F- 725F
by heating fluid #1
from 50' - 177.5' F

equivalent -7500000
= 625tons .15 x 625 = 93.5 tons
ref rigeration 12,000 to cool to supplemental
required 50" F cooling to lower

using available 85'F

cooling tower water 200OF -
200 GPM to cool
90F
200 x 60 = 12,000 GPH
temp to 50' F
none

none
assuming average
water cost: S.05
per 1000 gal.
water cost/hr 12 x .05 = $.60/hr none
annual water cost $60~24 x 365 = $5250 none

At this point in the process, fluid #2 in a conventional system has been cooled to only
90F by means of 85F city water and will require supplemental refrigeration for
final cooling to 50F.At the same time, fluid #2 in an APV regenerative system has
been cooled to 72.5"F by means of 85%regeneration and must be cooled further
to 50F.
Dum COOLFROM COOLFROM
WF - 50F 72.5"F - 50F
suppkmental 50'000
= 166.5 tons
refrigeration E,o00
required
power required 166.5x 1.2x 0.746= 149 KWH x 1.2x 0.746= 83.7KWH
assumingaverage electrical cast: 64/KwH
annual - 6x 149x 24 x 365= $78,341 -
6x 83.7x 24 x 365=$43,993
supplemental 100 100
refrigerationcost

stssm $394,200 $ 59,130 $335,070

tower
cooling 5,250 none 5,250
water

refrigeration 78,341 43,993 34,348

$477,791 $103,123 $374,668 annually

Note: Capital cost of a Paraflow plate heat exchanger for the above duty would be approximately
$17,000.

36
 u& since As = L x wetted Derimete

by the mean tem-

A further term for

and it is Dossible to eliminate L from

equation (A5)in equations (B),

(A3) and

bulent flow it is

he equation reduces to:

That is, the film heat transfer

empirical factor 2 to account for port

pressure loss and other deviations from
equations (112) to (A4).
solved using the HTU a

liminated toproduceageneral equation. fluid one divide

For any plate:
\
U-A
vhere m is the total mass flowrate an
Lf the total flow area.
De is definedby APV as four times the empirical modification of the constants
flowarea in a plate divided by the wetted
erimeter. Since the plate gap is small
ompared with the width then
 Paraflows in sea
cooling systems
The use of sea water as a cooling medium is far from new. It has been common prac-
tice with powered ships for many decades and, indeed, it is a result of technical
experiences on board ship that many of the materials used in the handling of sea
water and the practices related to its use have been developed. On shore, the exten-
sive use of sea water is a somewhat later development although it now is common
practice at coastal sites and becoming even more so with the increasing shortage of
good quality industrial cooling water. The rapid development of the global offshore
industry and of amphibious chemi
more intensive use of this medium.
basic essentials of a sea water cooler
Sea water is a very complex liquid. It can vary in cleanliness and salinity and, in
addition, is an excellent nutrient medium supporting many forms of animal and
plant life. Regardless of variations, two common potential problems can arise in all
cooling operations. Corrosion and fouling, particularly biological fouling, affect both
downtime and maintenance costs. It is essential, therefore, that coolingequipment
which is intended to employ sea water should embody a number of basic features.
Coolers should be constructed from corrosion resistant materials. They should
minimize deadareas, i.e., regions of low velocity which could permit organisms to
settle and develop,or silt to deposit. They should be designed so far as is possible to
discourage surface fouling.
The traditional materials for handling sea water are copper alloys such as
aluminum brass and the cupronickels specifically developed for marine use. Even
with marine applications, however, there are disadvantages to these materials,
principally that of being erosion-prone. This limits the upper level of velocities that
can be used in the design of the cooling equipment. Shipboard applications of sea
water, however, are relatively simple in that the medium to be cooled usually is
non-corrosive, being either lubricating oil or clean water used in the jackets and
pistons of the ships prime mover.
For other industries, particularly chemical, the cooling problem is more complex.
It frequently is found that a more sophisticated material has to be employed, not
only to combat the corrosivity of the sea water but also that of the medium which is
being cooled. The latter may be any of the wide,range of process streams handled by
the modern chemical plant. Typical of these materials are titanium, titanium/
palladium alloy, Incoloy 825 and the various Hastelloys. Of these, the one which
appears to begaining the widest prominence is commercially pure titanium although
even this is by no means a panacea and other materials often are needed. Unfortu-
nately,metals or alloys which can cope successfully with this dual corrosion situation
generally tend to be expensive. The list of attributes required of cooling equipment
using sea water therefore should include a high rate of heat transfer in order to
minimize the extent of usage of these high cost alloys.
One class of heat exchanger,the plate and frame type, in large measure possesses
these requirements since it is designed to avoid pockets and crevices. This auto-
matically fulfills one of the criteria for its use with sea water and other biomedia.

I 38
 carbon
2 865.810 774.594 107 A 82 12,110 19 19 3,633 11 8.7
water sea 9 9 B 72

In operation, the plate heat exchanger pack is clamped tightly between the head and
follower but the unit easily is dismantled for cleaningor maintenance. It also has the
further advantage of flexibility in that heat transfer surface can either be added to or
deleted from the original configuration.
It will be recognized immediately that when the plate pack is assembled, the
geometry of the flow channel on both sides of the plates is identical. Thermal and
pressure drop predictions thus can be made much more accurately than, for instance,
in the case of shell and tube (S and T) heat exchangers. In tubular exchangers, the
shell side geometry in particular is so complex that large margins of safety have to
be allowed for the film coefficient on this side of the unit. Furthermore, geometrical
similarity of the plate heat exchanger passages and equal access to both sides of the
plates removes the tubular exchanger dilemma of which shall be the shell side fluid
and which the tube side.
Without doubt, the most important feature of plate units in the context of this
paper is that the overall heat transfer coefficient achieved on liquid to liquid heat
transfer duties is of the order of two to three times that which can be achieved even
in a well designed tubular exchanger. The criterion of economy in the use of costly
constructional materials is readily met. Typical overall design coefficients for a
-
sweet water/sea water PHE would lie in the range of 600-800Btu/hr ft2 O F . Table 1
compares areas and pressure drops for S aqd T and PHE designs for typical indus-
trial duties, two of which are for a sea water cooling function. The S and T designs
were prepared with the assistance of the HTRI (Heat Transfer Research Inc.) S.T.3
program.
problems with direct sea water _cooling
Plate heat exchangers are, of course, extensively used in many process industries
for direct sea water cooling, i.e., the employment of sea water on one side of thecool-
ing surface and product liquid on the other. The main advantage is that the full dif-
ference between the product temperatures and those of the sea water can be used,
resulting in a smaller heat transfer area for agiven thermal duty. It is rare, however,
that any technological system possesses only advantages. Direct cooling, while
widely practiced, is no exception to this rule. A typical chemical industry example,
thecooling of high strength sulphuric acid by sea water, illustrates this point. Over
39
certain temperature ranges, product corrosiveness is such that a relatively un-
sophisticated and, therefore, moderately priced material such as 316 stainless steel
could be used quite readily. At these temperatures,however, 316 stainless steel will
corrode in sea water, so a more sophisticatedmaterial is needed. High nickel alloys
like Incoloy 825 which probably represent the next step up in sophistication and
price usually are avoided as well. While they are well suited to contact with the
sulphuric acid, there still is some risk on the sea water side.
Titanium, next upward in price, would be ideal for sea water but does not suffi-
ciently resist acid corrosion. Therefore,the final choice (Fig. 1) often has to lie with
an alloy such as Hastelloy 'C'which not only is the most expensive of those men-
tioned but also has a comparativelylow thermal conductivity. In addition toa higher
cost per pound, it sometimesis necessary with Hastelloy 'C'to provide agreater area
to do the same work. Differences in metal thermal conductivity on high efficiency
heat exchangers can be quite significant. Stainlesssteel and titanium haveconduc-
-
tivities of the same order of magnitude-about 112 Btu/hr * ft OF but Hastelloy 'C'
-
is about half this at 62.4 Btu/hr ft OF. For normal plate thicknesses and typical
PHEfilmcoefficientsof about 2025Btu/hr-ft2."Ffor waterand670Btu/hr.ft2."F
for strong acid, the additionalresistanceof Hastelloy 'C'would result in an increase
in area of 740% for this acid cooler.
-
alternative solution indirect cooling
This corrosiondilemma can be solved by i
circuit carrying clean and noncorrosive water between the acid and the sea water
(Fii. 2). The closed loop water initially flows through the sulphuric acid coolers
which can be of any material that suits the corrosive conditions, then rejects the
heat gained to other coolers which generally are fabricated of titanium to resist sea
water. As a bonus, the acid coolers are not sthject to cooling water fouling and can
remain on stream longer without cleaning.
The disadvantagesare that this system requires a pump for the closed hop water
and that the temperature differences between acid and sea water have to be appor-
tioned between the process cooler and the clean water/sea water cooler. There is
some compensation for this loss of temperature difference since less fouling allow-
ance need be built into the acid cooler design, and in this particular example,by being
able to revert to a material with a higher thermal conductivity. For both acid and
clean watedsea water heat exchangers,it is essentialthat high efficiency liquid-to-
liquid units be chosen. This is where PHEs are finding widespread application.
the Hunterston 'B' power station
The acid cooling example illustrates the philosophy behind the increasingadoption
 hr krcal problem areas within many diffpent types of
uclear or fossil-fk.al etatiom, chemical or fertil-
&&u&s, offshoreoil platform, or pulpad pper plants that d r e substantial
quantitiesof coolingwater.
The choice of closed loop cooling for the reactor and auxiliary systems for
Hunterston 'B' nuclear power station of the South of Scotland Electricity Boardis a
-case. Located at aooBstal sitesthe natural choice of^ medium Lser
water. Corrosive~uids,however,anmtbedowed into^^^AU reactor
emling circuits must be of e x t d y high intejgrity and myst utifise water with
WanB m i o n and fouling characten' s t i c s . I t k - a * & m f o r

P
 treated water or other liquid

I oil productionplatford

-.
I marineinstailation

FIG. 3
Closed circuit
cooling system

effected by a chlorine generator to avoid the potential risks associated with storage
of substantial quantities of liquid chlorine.
fouling
Biological fouling is, of course, not the only type to be expected in heat exchangers.
Taborek et a1 (2) and Fischer et a1 (3) classify fouling into six groups:
Crystallization
Silting (particylate sedimentation)
Chemical reaction and polymerization
Coking
Organic material growth (i.e., bio-fouling)
Corrosion
Depending upon the type of duty, these can occur singly or in almost any combina-
tion. For sea water plate heat exchangers, however,it is possible toeliminate at least
three. The choice of titanium removes the problem of corrosion and, at the same
time, neither polymerization nor coking are features of this type of cooling duty.
Cooper and Suitor(,)observe that the plate heat exchanger fouls less than the shell
and tube sides of tubular exchangers under typical operating conditionsand also is
much less sensitive to silt fouling. Other than biological fouling which already has
been discussed, the remaining type is crystallization.
Liquid hold-up volumes of plate heat exchangers are very small. Even in the
largest units, the particle dwell time for this type of duty is only of the order of 3-5
seconds so that crystallization, which requires a substantial time to develop, is
muchreduced. In the general chemical induary, there are many examples of plate
heat exchangers being deliberately selected for conditions where crystallization
vpould be known to occur in any other class of equipment. Thus, although the plate
type unit is not totally immune to crystalline fouling, the rate of formation is much
slower. In most cases of mixed fouling, the plate unit reaches an asymptotic fouling
level after a relatively short operating time.
closed circuit cooling
Using closed circuit cooling systems (Fii. 3) to serve complete in
gained widesp&d acceptance and is of particular value at the plant design stage. It
givesgreater freedom of choiceof construction materials and permits moreeconom-
ical process equipment design due to the virtual elimination of coolingwater fouling
allowances. A further advantage of closed circuit cooling sometimes is overlooked
42
 FIG.4 Ammonidurea fertilizer plant under construction with 13 plate heat =&angers as central coolera.
but nevertheless is important. It is common practice to design the central cooler
station on worst case: Le., summer temperatures, so that cooler sizes are fixed by
this criterion. By simply recycling some of the heated sea water, it is possible to
maintain summerconditions throughout the year, thereby avoiding seasonal fluc-
tuations in the temperature of the closed circuit water. This is important in some
installationssuch as closed circuit water used for jacket water cooling of large diesel
power sets.
It cannot be denied that handling sea water is troublesomein general, not only in
process plant but also in the sea water pipelines. From the viewpoint of subsequent
maintenance, there is a distinct advantage in keeping these pipe runs as small in
diameter and as short as possible. The high overallheat transfer coefficientsof plate
heat exchangers coupled with the fact that most are designed for almost 10096 true
mnter-current flow results in less sea water being required in the central coolers
than if any other type of heat exchangers were used. piping and pumps
therefore may be smaller.
There are a number of plants currently under consid where the process
installationwill be about 2-3miles inland fnnnthecoast. If dircct sea water cooling
is used, the problems of 4-6 miles of flow and return sea water Cufvert are obvious.
Instead,theseplantsarebeingdes~~wiwiththecentralcodingstationat theshare.
Sea water pipelines are s h r t and of relatively small diametez. Sina the pipelines
betw&n shore and the plant will carry only&tn noncorrosive!water, a relatively
troublefree installatim will result.
+. I shows a plate heat exchanger installation in the Mddb b t during the
construction of a new ammonidurea fertilizer plant..he thirteen- act
as central coolers for the whole of theprocess plant so that sea water piping extends
onlyfrom theintakethroughthetreatmentplant, t h e n t o t h e m h a n d b a c k t o t h e
sea.The position of the underground manifoldsfor the sea water and closed ckuit
water flow and return can be located easily in the photograph, the stub pipes for
subsequent conpection to the heat exchaxgersbeingplainly visible.
The& coolers provide almost 40,OOO GPM of recirculating cooling water for the
main plant process items. As in the case of the sulphuric acid cooler, isolationof sea
water from the process plant enabled the designers to select materiaisto suit mn).
sion characteristics of process streams being h a n W at any point, and -1y
43
 ILLSTAINLESS (TUBESANDSHELL)
8T-STAINLESSTUBES. MILD STEELSHEL
8T-MILD STEELTUBESANDSHELL
YC
2LL STAINLESS CONTAC

500 1000
.
HEATING ARE

FIG. 5 Relativecosts: plate vs. tubular exchangers

%M!M
j7*-*

5 * ' ~ f ~ w'.3=m I 1 s
resulted in lower cost materials being used. With no corrosion problems at the
central coolers, any fouling which does occur can be due only to the fouling charac-
teristics, if any, of the process streams. w!"d

costs
Refere
expensive metals and alloys. Fig: 5 shows relative costs per unit area of stainless
steel PHEs against three different material combinations for tubular units. On this
basis, PHE surface always is less expensive than any tubular unit partially or
wholly constructedof stainless steel. Thegraph indicates that the unit area cost of a
mild steel shell and tube exchanger is lower than that of a plate heat exchanger with
stainless steel surfaces. However, since smaller areas are needed for plate type
exchangers, it can happen on a duty for duty basis that the plate unit sometimes is
less expensive than an all mild steel tubular.
Where more sophisticatedmaterials are mandatory for both classes of exchangers,
the price differential rapidly widens in favor of the PHE-by a factor that can at
times exceed 4:l as was the case for the ammonidurea fertilizer plant previously
mentioned. World Bank financing was involved and international bidding was by
two plate and five tubular exchanger manufacturers. Both plate heat exchanger
bids were for titanium construction. The lowest priced tubular offer which used
mild steel shell and aluminum brass tubes was 2.2 times the PHE prices while the
lowest priced titanium tubed units were 4.1 times.

-R
1 South of Scotland Electricity Board (Research Division) Internal Report No. RD 2/70 - 'SeaWater Testing
of Plate Heat Exchanger Materials.' Author, McKay, J.D
2 er. R B , Paten,J W and Knudsen,J G -'Fouling, The Major UnresolvedProblem in
hemical Engineering Progress,68, No 2, pages 59-67. February 1972
3 , Ritter, R B - 'Fouling Measurement Techniques,'Chemical Engineering Progress,
4

44
corrosion
and heat transfer
7
Insofar as the chemical industry and certain segments of food processing have to
deal with widely ranging corrosive processes, engineers must be particularly con-
scious of the need for detection and replacement of sections of corroded equipment.
The ease with which this is achieved with the plate heat exchanger must, in no
small part, account for the widespread acceptance of this method of heat
However, by careful attention to detail at the design stage and observa
implementation of the correct operating and maintenance schedules, any
plate replacement because of corrosion should be minimized.
As describedin the initial section of the APV Heat Transfer Handbook, the des
principles, methods of construction and operating characteristics of the plate h
exchanger are fundamentally differentfrom the more traditional forms of heat trans
fer equipment such as the shell and tube heat exchanger. Since the configuratio
andgeometry of the plate heat transfer surfacesare unique,a new materials specifica
tion philosophy frequently has to be adopted when consideringmaterials of construc
tion if a plate exchanger is to be operated on a corrosive process or service stream.
catering for general corrosion
Since the flow plates of a plate heat exchanger are quite thin (0.024 inl0.036 in.) in
contrast toother heat transfer surfaces (0.048 in. andgreater) corrosionallowances
normally quoted for chemical plants in standard reference books tend to be mean-
ingless. Such works will give a Grade A rank to a corrosion rate of 0.005 in./year or
less - a rating which is realistic when consideringmaterial 10times this thickness.
However, a corrosion of 0.005 in./year with metal of 0.024 in. nominal thickness
represents over 50% reduction in thickness in just three years of use. To most
chemical engineers, this would be unacceptable. As a general guideline, the maxi-
mum permissible corrosion rate for a plate heat exchanger is 0.002 in./year and at
this level, the concept of general corrosion in plate units is brought into line with
normally accepted levels for other pieces of process equipment.
The implications are that in a specific corrosive environment, a change from
tubular exchangers to plate units may necessitate an upgrading of the alloy used for
the K a t transfer surface. In spite of this decasional need for a more expensive
material of construction, however, the advantage of using thin gauge material and
theinherently high heat transfer coefficients of the plate heat exchangers frequently
means that the actual cost of the required heat transfer surface is less. The added
attraction of using a more highly alloyed material is that some degree of extra cor-
rosion protection is built into the equipment, thus permitting excursions from
normal design conditions into the more corrosive regimes which frequently are
encounteredin the startup and shutdown of a modern chemical plant.
materials for heat exchanger plates
Before describingthe various types of corrosionwhich may occur, a summary of the
composition and corrosion characteristics of the metals and alloys from which heat
exchanger plates are produced is appropriate.
 Bal.
IO IU 2% Bal.
20 25 4% 2 Bal.
20 18 6 % Bal.

99
<1 Bal. 28 2
15% Bal. 16 5
22% Bal. 6% 2 20
18% Bal. 9 3
21% 42 3 2 Bal.
Bal. 30

304:The most inexpensivegrade of the austenitic stainless steels. While offeringa

general corrosion resistance to a range of organicand inorganicproducts, it exhibits
a poor resistance to sulphuric and hydrochloricacids. It is particularly susceptibleto
crevice corrosion induced by cooling waters containing chloride. The marginal cost
savings which can be achieved by its use as an alternative to 316 stainless steel
rarely is justified.
316 A general purpose stainless steel suitable for use in a wide range of environ-
ments and with limited applicationsin sulphuric acid. The presence of molybdenum
in the alloy confers a level of resistance to chloride-inducedcrevice corrosion which
will permit its successful application in most coolingwater duties.
AVESTA 254s- This alloy is of particular value for duties involving sulphuric
acid, and especially phosphoric acid where there is contamination with chloridd
fluoride ions. The high nickel content imparts a g d degree of resistance to stress
corrosion cracking, a failure mode frequently associatedwith pittinglcreviceattack
in316 stainless steels. An alloy which offers an excellent compromisebetween price
and corrosion resistance for a wide range of acidic and chloride containing
environments.
AVESTA 254SMO An improved hade of 316 resulting from the increased level of
molybdenum which imparts better resistance to chloride-inducedpittingkrevice cor-
rosion. Of use in brackish waters where the corrosion resistance of 316 is inadequate.
The performance in mineral acids is significantly better than 316. In general terms,
the corrosion resistance improves with increasing molybdenum content.
*l*ys
NICKEL 2 0 0 U s e of this material tends to be confined to high strength 0 5 0
46
 percent) caustic soda at high temperatures (up to B.Pt) although it is susceptible to
chloride-induced crevice corrosion by brackish cooling waters.
HASTELIXIYB-2:An expensive alloy with a useful resistance to hydrochloric acid
and many sulphuric acid concentrations. Its disadvantage lies in its inability to
withstand oxidizing environments and contamination of acid streams with ions
such as ferric and cupric. If the acid is aerated, it can have a catastrophic effect on
corrosion resistance. It is particularly good for handling the products of Friedel
Craft reactions.
HAsTELIAlY C - 2 7 6 An expensive alloy which is almost immune to attack by
chloride ions at low pH values. It has an extremelygood resistance toa wide range of
sulphuric acid concentrations and can be used with certain strengths of hydrochloric

acid. It is particularly resistant to wet process phosphoric acid. One of the few alloys
suitable for use in hot, concentrated sulphuric acid.
HASTELIXIY 6-3 Possessing a high nickel and molybdenum content, the alloy is
suitable for dilute and middle ranges of sulphuricacid concentrations. It is very good
where contamination with chloride and fluoride ions occurs (Le., gas scrubbing
duties)and exhibits an improved resistance in such environments over Incoloy 825,
albeit at an increased price.
INCONEL625 Suitable for many environments, including chloride ions and acids.
It provides a compromise solution where conditions demand something better than
Incoloy 825but do not necessitate the extremely good corrosion resistance (and high
p r i ~of~Hastelloy
) C-276.
INCOLOY 825 A reasonably priced alloy with a good resistance to high strength
caustic soda and virtual immunity to stresscorrosion cracking. It is the workhorse
for duties involving numerous sulphuric acid concentrationsat lower temperatures
but it is susceptible to chloride-induced crevice corrosion.
MONEL400:Particularly good for acid chloride environments and, under certain
circumstances, brackish and saline waters. It has excellent resistance to high
strength caustic soda (30-50percent) at high temperatures, but it is susceptible to
attack by mercury which occasionally is present as a contaminant. In such cases,
Incoloy 825 is preferred.
other metals
TITANIUM The material par excellencefor chloride solutions. It is virtually
immune to attack by sea water upto 248F (120C) and to other chloride solutions
guch as calcium chloride. It is not immune to attack by some amcentrated chloride
solutions at temperatures over 212F (100C)and therefore in duties where this
condition prevails or is likely to arise through leakage, titanium-palladium is pre-
ferred. Titanium has excellent Mistance to concentrations of nitric acid up to 70
percent although its resistance to dilute sulphuric acid is similar to that of 316. A
useful performance in hydrochloric acid can be achieved,espeCially if the acid is con-
taminated with oxidizing ions such as ferric or cupric.
TITANIUM-PALLADIUM The addition of0.15percent palladium to titanium has
a pronounced effect in increasing corrosion resistance. In general terms, it produces
an all around improvement in resistance to mineral acids (sulphuridhydro-
~

chloric), especially in reducing condition, i.e., spin bath liquor.

I
47
 general or uniform corrosion
This type of corrosion commonly is encountered with carbon and low alloy steels
(rusting)and,as the name suggests,is a form of attack which results in a moreor less
uniform deteriorationand thinning of the metal. It is, however, rarely encountered
with stainless steels and more highly alloyed metals and when it does occur, it is
almost exdusivelythe result of attack by certain mineral acids or certain particularly
corrosive organic acids (e.g. hot concentrated oxalic acid solutions). Insofar as
general co&ion rates for most metals and
alloys are well documented, this type of at-
tackis relatively easy to predict and the selec-
tion of suitablematerials,bearingin mind the
criterion of tolerable metal loss previously
mentioned, presents few problems. Because
of its predictability, this could well account for
the fact that it is so rarely encountered.Metal
which has been the subject of general corro-
sion assumes an etched appearance and the
original metal surface morphology usually is
lost. F m r e 1shows a typical example of this
form of corrosion.
General corrosionalways is a temperature
related phenomenon. The higher the temper-
ature, thegreater is the corrosion. In any heat
transfer situation, it is not sufficient to surface. Note pristine condition of
assume that the maximum bulk liquid tem- area outside product contan
perature represents the most corrosive con-
dition. T h e temperature of the heatinglmling medium always will dktate the
actual skin metal temperature. In a cooling duty, it is true to say that the bulls liquid
represents the most corrosivecondition since metal around the inlet portswill be at
this temperature even though the heat transfer surface may be cooler. On the other
hand, the most corrosive conditions in a heating duty exist where the heating
medium first contacts the product via the heat transfer surface. The actual maxi-
mum skin metal temperature,therefore, must be determined before selection of the
correct material can be effectd
9

pitthgcorrosion .
One very common cause of equipment corrosion failure is
attrtck characteristically is caused by &bride or chlotine
quently, a largenumber of environmentsexist within the pmsesss irdarstrigs which
potentially are corrosive and for which correct material selectioa isc r k b l .
Pitting corrosion damage is highly localized and tends to be deep and undercutin
nature. Fwre 2 shows pitting upon a stainless steel pipe and, as can be seen, the
surface damage is little more than a series of randomly sited small hoses with the
adjacent metalbeing entirely unaffected and retaining its original polished appear-
ance. Viewed in cross-section, the insidious nature of pitting eorrodh pdiapsis
easier to appfeciate.Underneath the surface, the metal has been corrodectawayin
an undercut fashion and complete perforation of the metal is im
largely to the product from the corrosion reaction being more
4a
 FIG.2 Pittingcorrosion.Inset
shows a micrograph of a section
through a corrosion pit.

initiatingprocess stream. Additionally, these products also tend to be more dense and
it is not uncommon to see pitting grow under the influence of gravity.
From all the foregoing, it will be appreciated that the actual loss of metal is negli-
gible when compared with the bulk. This highlights the inadequacies of corrosion
tables which quote weight loss or mean loss of metal thickness.
The austenitic stainless steels are susceptible to this type of attack and in very
broad terms, their susceptibility is an inverse function of the molybdenum content
of the alloy. For example, 304stainless steel (no molybdenum) is particularly prone
to pitting corrosion by neutravacidic solutions containing as little as 50 ppm of
chloride. This, therefore, limits the use of this grade of stainless steel to all but the
pmbt of coolingwaters and seriously restricts the chemicalenvironments in which
it CBR be employed successfully. On the other hand, 316 stainless steel (246% molyb
denum) offers a reasonably good resistance to pitting attack and as a very rough
guide, 150ppm chlorideconcentration is taken as a safe maximum Ievel. Thus, this
grade of steel is suitable for use in a wide raw of industrial waters, even cooling
towerwaters,assumingthatthewatetqualityutcnlatinginsuchasystemLmain.
and the chemistry of it is coRtrdled.Jtmwt beemphashd that all the fare-
going is for typical waters. Other factors such as pH or scaling tendency can have a
influenceontheperformanceoBthevarioYsmaterials.Specificrecommenda-
therefore,can be made only in the light of all the iononthe
particular stream.
F b r m o r e ~ i ~ ~ a r ~ s t r e a l n s o o n t a i n i R
alloys such as Avesta 254SLX (434% molybdenum) or Av&b 254S#o (Ssa molyb-
b u m ) may be required. For sea water cooling applications (2l,OOO ppm chloride),
the 6%molybdenum alloys can be used for lower temperatures but at high tempera-
tctres, alloys such as Inconel 625 may be needed or even titanium which is immune
to this form qf damage at temperatures up to 2 4 8 O F (UOOC).
Chlorination of water supplies is a common practice and although the 300series
stainless steels are suitable in potable waters containing up to 2 ppm midud
chlorine,moreh~lyalloyedmateriaismaybenecessaryw~tkereere~~.
FIG.3 Crevice corrosion at the contact
points of a heat exchanger plate. Inset
shows a micrograph of a section through
a corrosion site.

FIG.4 Typical stress corrosion

cracking of a 304stainless steel
storage tank. Inset shows a
section through a crack.

As with chloride solutions, titanium offers immunity even to saturated chlorine

solutions such as may be encountered in chlorine production units.
Brines basedon sodium or calcium chloridecommonly are used in coolant circuits
and these too can initiate pitting corrosion of stainless steels. For their successful
use in equipment constructed of 300 series stainless, strict pH control and deoxy-
genation is essential. During shutdowns, draining and rinsing of brine residuals is
essential to avoid intense pitting attack. While the precautionary rquirements for
successful use of 316can be enforced in food processing plants, it is most uncommon
for them to be fulfilled in an industrial chemical plant. Under these circumstances,
the use of the 300 series steels is inadvisable. As a material able to withstand
chemical abuse, titanium ideally should be employed.
.. 1

crevice corrosion
Mlkmever pittingcomion is encountered, it is not uncosnmoat;forc-ion
to be found as well. Crevices inherently occur within modern proaess ghat at a
number of places, i.e., pipeline joints, valves, and within plate he& ex&.
F w3shows crevice corrosion on a heat exchanger plate at a pIabbpb&aWact
point. Crevice conditions also can be created on heat transfer s d m by the
presence of scale or product debris.
Crevicecorro$on is, in fact, essentially similar to pittingcomiouexcept that by
definition it can occur only under crevice conditions where several added features
can exist which are capable of adverselyinfluencingthe pittingcorrosionprocess. It
is caused by exactly the same range of chloride and chlorine bearingenviranments.
This similarity with pitting is clearly evident from the cross-sectional view.
 There is a divergenceof opinion as to the reaction mechanism which will initiate
crevice corrosion. It originally was considered that it resulted from an electro-
chemically induced breakdown of the protective oxide film on the metal surface
caused by a difference in concentration of an ionic species or oxygen within the
crevice and in the bulk liquid surrounding the crevice. Modern views tend to favor a
mechanism whereby, due to the restricted irrigation of the liquid trapped in the
~ crevice,restricted replacement of oxygen in the crevicefluid causes chloride or other
corrosive ions to migrate into it and create very high concentrations. This, in turn,
causes accelerated attack.
The preventive measures for crevice corrosionare, as one might expect, the same
as those for pitting corrosion. While certain steps such as keeping plant surfaces
clean will alleviate the risk, such measures rarely are practical on equipment located
in a continuouslyoperated industrial complex. For such equipment, correct material
choice is the all important factor. It is worth restating that 304 stainless steel is
totally unsuited for coping with crevice corrosion and although 316 offers some
degree of resistance, for many situations the highly alloyed stainless steels or
titanium may be necessary.
As with pitting corrosion,values quoted in corrosion tables
mean metal penetration rate are meaningless.
stress corrosioncracking (S.C.C.)
Yet another form of corrosion damage which frequently is encountered and which
can cause extremely rapid equipment failure is stress corrosion cracking.This form
of damageoccurs when a metal under the influence of stress comes into contact with
certain highly specificenvironmental conditions. While these vary depending upon
the metal, in the case of the chromiudnickel alloys it is most commonly chloride
ions. It also is a feature of stainless steels that S.C.C.only occurs in environmental
conditions when the temperature is in excess of 140F(60C).
Figure 4 shows the appearance of a typical stress corrosion failure, the cracking
being a network of highly branched cracks. It will be noted that the cracking is
transgranular, i.e., the cracks pass through the metal grains rather than following
the grain boundaries. From a point of view of diagnosing plant failures, this trans-
granular route of the crack path is of vital significance in helping to classify the
corrosive agency as chloride ions.
The stress component of the failure mechanism may be present either as an inbuilt
stress resulting from welding and fabricational processes or alternatively, may be
*due to process pressure conditions. In either case, it is clear that in modern process
plant the presence of stressed metal is unavoidable. It also must be pointed out that
other failure processes can play a role in stress corrosion by actingas stress raisers,
complementing the existing stress levels. For example, it is not uncommon to find
stress corrosion cracks initiating from within pits formed by crevice or pitting
corrosion, Figure 5.
The susceptibility of an austenitic stainless steel to S.C.C.is a function of the
nickel content of the alloy. It is an unfortunate fact that thegreatest susceptibility is
exhibited in stainless steels containing ca. 10%nickel which is the alloying level in
both 304 and 316. Where an environmental situation exists when there is a serious
risk of S.C.C.,the remedial action necessary is to select an alloy with a higher or
51
 FIG.5 Stress corrosion cracking associated
with crevice and pitting corrosion.

FIG.8
Fatigue crack with stress
mrrosion cracking emanating
f"the fractureface.

lower nickel content. The so-called duplex structurestainlesg steelswith a nominal

5%nickel and the ferritic stainless steels with zero nickel exbibit extremely j p d
resistance to S.C.C. Unfortunately, their mechanical propertKs and ductility are
such that it is not possible to form anything o*er than the (geumetricauy) simplest
heat exchange plate form although for the fabrication of other a0 process
equipment,they offer many advantagesover the 300series sta&su.stmaiS. l;lcreasr
ing the nickel content to 25%(Avesta 254SLX)imparts a very h&h degm of &-
tance to chloride induced S.C.C. while with an alloy containing4096 niclrel(Incdoy
825),it is claimed that the material exhibits immunity.
Before closing this section on S.C.C.,.brief mention must be made of a second
specific environmental condition which can cause this mode of failure-namely,
caustic alkalis. However, the conditionsrequired for such attack are quite severe. At
230F (llOC), for example, caustic concentrations of some 30.6096 are neceswy.
The unique feature of this type of S.C.C. which differentiates it from chioride in-
duced S.C.C. is that the cracking is intergranular (following the grain boumlanes
as shown in Figure 6.
52
 As previously mentioned, the remedial action is to employ a material of higher
nickel content. Although Monel400 can be used for high temperature duties involv-
ing caustic soda in the concentration range of 30-5096, for higher strengths and
temperatures nickel is the ultimate choice. However, both these materials have
disadvantageswhich have been covered in a previous section.
Mgue
Fatigue is theoretically an in-vacuophenomenon. More correctly,examplesof fatigue
-inplate heat exchangers, the incidence of which is very low, should be defined as
f
corrosion fatigue.
Fatigue is caused by the metal being allowed to flex under the influence of cyclicor
pulsating load conditions. The end result is a straight, unbranched crack as shown
4
in Figure 7.
It may be argued that insofar as fatigue is the result of a purely mechanical action,
it does not come into the field of corrosion. In many, if not most, instances however
there is a small promotional corrosion influence which has an adverse influence on
the time to failure. In some cases, the participation of the corrosion element in the
cracking process is evident. For example, stress corrosion cracks propagating from
the main fatigue crack (Figure 8)sometimes is observed. Although there is a corro-
sive element invariably associated with fatigue, it generally is considered that in the
absenceof a fatiguingconditionthe corrosive element is insufficientto initiate failure
from this source alone.
- - Numerous factors, both mechanical and environmental, can influencethe fatigue
hazard but in terms of avoiding failures in plate heat exchangers, there are only two
satisfactory preventive measures and these stem from the original definition of a
fatigue failure-the metal being flexed under the influence of a pulsating load. The
most common source of pulsating load originates from positive displacementpumps
and thus wherever possible,the use of centrifugal pumps is to be advised since these
produce a smooth pressure output. In correctly assembled and maintained process
plant, however, theequipment is less able to flex under any prevailing loads and thus
more severe conditions can be withstood before failure starts to occur. This high-
lights the need to correctly tighten plate packs within a plate heat exchanger frame.
For a plate heat exchanger, the necessary plate support comes from the plate-to-
, plate contact of support points which are pressed into the plates at calculated sites.
If these contact points become distorted or flattened either by adverse pressure con-
ditionsexistingwithin the plant or byoperatarsovertightening the plate packduring
assembly, this good support w illbe lost.
Figure 9 shows fatigue crackingon a heat exchanger plate near support pips that
have been flattened by overtighteningthe machine. This adversesituation occurred
in an effort to effect a seal with aggd and compressed gaskets that should have been
replaced. Although plates eventually were regasketed and the plate pack easily
compressed to the correct dimensions, fatigue cracking still resulted because of the
flattened contact points and loss of plate support.
erosion corrdon
Impingement attack or erosion corrosion is the increase in the rate of deterioration
on a metal attributable to the movement of a liquid over the surface. It usually is
characterized by the appearance of troughs or gullies on the surface. This type of
53
P

lY

magegenerally does not prove to be a problem with stainless steels but mo

is associated with the softer, copper based alloys. Figure 10, however, s
ion damage to stainless steel at an inlet orifice where the velocity of a
osive process stream was extremely high. In other areas where theveloci
greatly reduced, no attack was noted.
For agiven plant material of construction-product combination,flow velocity an
turbulence are critical factors in determining the erosion risk and for this reason, a
flow velocity maxima often is imposed where alloys sensitive to erosive conditions
are encountered.
When solutions contain abrasive particulate solids, even the stainless steels may
be subject to erosion although, because of their excellent resistance to this form of
deterioration, limiting the velocity usually is an adequate precaution.
conclusion
It has been possible in this paper to cover the performance of materials of construc-
tion for plate heat exchangers only in the broadest terms. There are many forms of
corrosion of which no mention has been made (hydrogen embrittlement, caustic
embrittlement, intergranular attack, galvanic corrosion) since these forms of
damage either are peculiar to materials not med in plate heat exchangersor are the
result of bad manufacturing techniques which also are not used in making plate
type units.
Corrosionengineering is a complex subject and a blind interpretation of corrosion
data presented in many reference works can lead to problems. Only by taking all the
factors into account at the specificdtion stage can materials offering optimum per-
formance be selected. For such a choice to be made, it is imperative that all the facts
are stated and the full and exact composition and maximum temperature of the
process streams given. For example, as little as 50 ppm of fluoride ions as an im-
purity in an acidic process stream can cause the catastrophic failureof titanium. On
the other hand, for a duty involving sulphuric acid which normally would demand
an expensiveplate material, the presence of oxidizing species such as ferric or cupric
ions will permit the use of a much less expensivealloy.

I 54
A APV
APV Crepaco Inc.
Executiveand Intmatknal Headquarten
9525 West Bryn Mawr Ave.; Rosemont, IL 60018
(708)678-4300 Fax:(708)678-4407
Enginawingand Manuhcturing
395 FillmoreAm., Tonawanda, NY 14150/(716) 692-3000
100 South CP Am., Lake Mills, WI 53551/(414) 648-8311
1303 Samuelson Rd., Rockford. IL 61109/(815) 397-5000
i65 John L. Dietsch Sq.,AWeboro Falls, MA 02783/(508) 695-7014

R.gknrlco8:
Los Angeles (Cerritos),CA/(2l3) 926-9700 Chicago (Rosemont),iL/(708)678-4300
Metropolin New York (Haworth, NJ)/(2Ol)381-9500 Nashville, TN/(615) 255-0342
-
Dallas (Irving),TX/(2l4) 257-3455 -

sales ottlces:
Minneapolis, MN Charlotte, NC Mt.Laurel, NJ Columbus OH
-- UT Seattle, WA
Salt Lake City,

In Canads
-
APV Canada, Inc. Toronto (416)850-1852 Montreal (514)737-0006
Vancouver (604)420-4344

lntemationdotfices:
Tokyo, Japan Mexico City, Mexico Taipei, Taiwan