You are on page 1of 5

RS1084

DESIGNING SHELL-AND-TUBE HEAT EXCHANGERS WITH


VELOCITY DEPENDENT FOULING

D Butterworth

HTFS Senior Consultant


Abingdon, Oxfordshire

ABSTRACT

This paper uses the "design-envelope" concept to explore the effect of velocity dependent fouling on the design of a
shell-and-tube exchanger. The case of tube-side fouling is considered in which the fouling is assumed to be zero
above a critical velocity and to have a high but constant value below this critical velocity. It is shown that the
inclusion of the velocity dependence has a profound effect on the final design result and therefore on the design
strategy. The effect of fouling both on heat transfer and pressure drop is considered. Cases are illustrated where
there are two separate design envelopes, one for the fouled exchanger and one for the clean exchanger. Conclusions
are drawn about the implications which this has on the specification, design and operation of exchangers.

NOTATION 1. INTRODUCTION
Roman Shell-and-tube exchangers are almost always
A Heat-transfer area [m2 ] designed using a constant, specified fouling
D Tube internal diameter [m] resistance. There are however many cases where the
f Friction factor fouling is known to depend on the local conditions of
k Thermal conductivity of tube-side stream temperature and velocity, even though these
[W/m K] dependencies are only known approximately. This
K Head-loss coefficient per pass for inlet, paper explores whether we should take account of
outlet and turnaround on the tube side such dependencies during the design process, just as
L Tube length [m] we take full account of the effect of local conditions
np Number of tube passes on heat transfer coefficients and friction factors. The
Pr Prandtl number of tube-side stream argument for assuming a constant fouling resistance
Q Heat duty [W] is, of course that we often only have a vague idea
r Tube-side fouling resistance [m2 K/W] about the level of the fouling and even less about any
R Thermal resistance due to tube wall and variations with process conditions. It is shown here,
shell-side stream [m2 K/W] however, that it can be worthwhile attempting to
Re Tube-side Reynolds number include the variations in fouling, even if only known
u Tube-side velocity m/s approximately, because they can profoundly
U Overall heat transfer coefficient [W/m2 K] influence the final design. Indeed, a case is shown in
which there are two quite separate design envelopes
Greek
for clean and dirty exchangers thus illustrating the
α Heat-transfer coefficient on tube side adage that you "get the fouling which you design for".
[W/m2 K]
∆p Tube-side ressure drop [Pa] This paper deliberately uses very simple calculation
∆Tm Mean temperature difference [K] procedures in order to illustrate the phenomena
ρ Tube-side stream density [kg/m3] clearly without getting distracted with details.
However, the main points which are made will apply
also if more rigorous calculations are used.

231
The case considered here is of a hydrocarbon which influence the pressure drop. This is done simply by
is being heated inside tubes using condensing vapor doubling the Blasius friction factor.
on the shell side. For simplicity, the shell side is
assumed not to be controlling (i.e. to have a very high No account is taken here of nozzle losses but it is a
coefficient and to have no important pressure drop simple matter to do this in a real design.
limit) and to be isothermal. The hydrocarbon is
2.2 Heat Transfer
assumed to have velocity dependent fouling. Having
this fluid on the tube side is in line with normal The heat load is calculated from
practice to put the fouling stream in the tubes to
facilitate cleaning. Q = UA∆Tm (3)

The design process is explored using the "design- As this paper concentrates on the tube side, the heat-
envelope concept" discussed elsewhere by transfer area, A, and the overall coefficient, U, are
(Butterworth, 1996) and which has been incorporated based on the tube internal diameter.
into a piece of educational software known as
DEVIZE. Both the design-envelope method and the The overall coefficient, is calculated in the normal
fouling effects discussed here will apply to other heat way by
exchanger types. Indeed, HTFS uses the design-
envelope method in its air-cooled exchanger program, 1 1
= R+r+ (4)
ACOL. This program also allows the user to input U α
the fouling resistance as a function of velocity and
local process conditions. Where the tube-side coefficient, α, is given by the
simple Dittus-Boelter relationship.
2. BASIS OF CALCULATION
k
As explained in the introduction the calculations are α = 0.023 Re 0.8 Pr 0.4 (5)
D
confined to the tube side stream and very simple,
though representative, methods are used.
The tube side fouling resistance, r, is taken as zero at
2.1 Pressure Drop high velocity and a constant, but high value below a
The tube-side pressure drop is calculated from certain critical velocity.

 R, is the combined resistance of the shell-side stream


L ρu 2 ρu 2 
∆p = n p  4 f +K (1)
 D 2 2  and the tube wall (referred to the tube side). An
assumed, low value is used in the calculations.

Where the friction factor is calculated from the 3. ENVELOPE METHOD


Blasius relationship
The objective of this method is to design an
f exchanger that achieves the heat load within the
= 0.0396 Re −0.25 (2)
2 constraints of the stream pressure drops (and any
other constraints which one may wish to include). In
The coefficient, K, is the number of velocity heads this case, the shell side pressure drop is ignored.
lost due to entrance exit and turnaround in each tside
side pass. A representative value of 1.8 is used in the For a given number of tubes and number of passes,
calculations. the tube-side velocity is first calculated. From this,
The above method is strictly for a clean tube. In using the above equations, the tube length is
some of the calculations, fouling is also assumed to calculated which will just use up the specified
pressure drop. This may be plotted as a point on a

232 RS1084
graph of number of tubes verses tube length. Table 1: Data used in calculation
Repeating this for different numbers of tubes gives a
locus of exchanger configurations which will just Item Value
satisfy the tube-side pressure-drop limit. This is Tube internal diameter 21 mm
illustrated in Figure 1. All exchangers above the line Number of passes (in base case) 2
will satisfy the pressure-drop criterion and are Required heat duty 10 MW
therefore valid designs, whereas those below the line Mean temperature difference 40 K
will not. (which in this case corresponds
to the LMTD)
The process may be repeated for heat transfer, giving Tube-side mass flow rate 100 kg/s
the line shown in the figure. Again, configurations Maximum allowable tube-side 0.3 bar
above this line will satisfy the heat transfer while pressure drop
those below will not. Hence, one may construct an Combined thermal resistance of 0.0001 m2 K/W
envelope, shown shaded of those exchangers which shell side and tube wall
will satisfy both the heat-transfer and pressure-drop (referred to the tube internal
criteria. Normally, the cheapest design is that with diameter)
the least number of tubes, which is easily determined Properties of tube-side fluid
from where the lines cross. Density 610 kg/m2
Specific heat 2850 J/kg K
Viscosity 0.18 mN s/m2
Thermal conductivity 0.088 W/m K
Prandtl number 5.83

5. RESULTS

5.1 Clean Case

The example shown in Figure 2 is the base case of a


clean exchanger. The exchanger with least tubes is
4.65 m long and has 303 tubes. It is clearly possible
to choose an exchanger with a particular tube length
if working to fixed tube lengths. However, this
Figure 1: Illustration of envelope concept
would be at the expense of more tubes. For example,
4. EXAMPLE a 5 m length requires 311 tubes and a 6 m length
requires 332 tubes.
Table 1 gives the details of the example which is
considered in the following calculations. The fluid
tube-side properties are those n-decane at 190oC.

Figure 2: Envelope for clean exchanger

233 RS1084
5.2 Fouling Affects Heat Transfer Only working together on avoiding fouling, the best
solution here may be to allow for more pressure drop
The previous case is calculated but now with fouling.
in the first place.
The exchanger is assumed to remain clean if the
velocity is kept very high. Hence zero resistance is
assumed if the tube-side velocity is greater than 2.5
m/s and 0.0005 m2 K/W is assumed if below that
velocity. In this case, the fouling does not affect the
pressure drop.

Figure 3 shows the resultant envelope. At around


380 tubes, the tube side velocity is at the critical
value of 2.5 m/s. For greater numbers of tubes, the
exchanger is fouled and so the heat transfer curve is
displaced sharply to the right. This gives the two Figure 4: Envelope for a 4 pass exchanger (fouling
operating regions shown. Focusing on the lower, only affects the heat transfer)
"clean" envelope gives the same designs as before. A
designer may however be concerned that the clean As a digression, this example illustrates the power of
operating envelope is small, making the installed the design-envelope method. If the results had not
exchanger sensitive to small changes in operating been plotted in this form, it is doubtful whether the
conditions. The designer may therefore play safe by designer, unless very experienced, would have
choosing the higher, "fouled" envelope. The design spotted the problem. Also, many of the current
with least tubes is then 9.00 m long and has 393 thermal design programs would have had difficulty in
tubes. finding both the clean and dirty designs.
5.3 Fouling Affects Both Heat Transfer and
Pressure Drop

Figure 5 shows the envelope for this case (with a 2-


pass exchanger). As already stated, the increased
pressure drop due to fouling is calculated simply by
doubling the clean friction factor. In practice, higher
pressure drops would be due to both an increased
friction factor and to flow area blockage.

The clean envelope is as before but now the fouled


Figure 3: Envelope when the fouling only affects the envelope is displaced higher because the pressure-
heat transfer drop locus is displaced to the left. The design for a
fouled exchanger is now 6.4 m long with 461 tubes.
Let us suppose that 9 m is too long an exchanger in Comparing with the clean case, the penalty for
this case. The designer might then consider going to allowing fouling is 52% more tubes and a 38%
4 passes to increase the tube-side velocity and hence increase in tube length.
increase the heat transfer coefficient. The result is
shown in Figure 4. This certainly shortens the
exchanger but at the expense of having many more
tubes. Despite forcing higher velocities, the clean
envelope is even smaller than before. The problem is
clear. The tube side pressure drop is limiting the
design. Hence, if both the designer and the user are

234 RS1084
ACKNOWLEDGEMENT
This paper was previously presented at the US
National Heat Transfer Conference, Pittsburgh, 20-22
August, 2000.

REFERENCE
BUTTERWORTH, D, (1996) "Visualize Your
Design of Shell-and-Tube Heat Exchangers",
Figure 5: Envelope when fouling affects both the heat Chemical Technology Europe, Vol. 3, No. 4,
transfer and the pressure drop July/August, pp20 -24.

6. CONCLUSIONS LIST OF TABLES


Using what is admittedly a simplified case, the paper 1. Data used in calculation.
illustrates clearly that including the effect of velocity
on fouling resistance has a profound influence on the LIST OF FIGURES
design process. It is shown that exchangers may be
1. Illustration of envalope concept
designed to be clean or to be dirty. For clean designs,
2. Envelope for a clean exchanger
it becomes evident that the way the exchanger is
3. Envelope when the fouling only affects the heat
operated is also important since the clean operating
transfer
envelope may be small. The use of the design-
4. Envelope for a 4 pass exchanger (fouling only
envelope concept is shown to be a good way of
affects heat transfer
highlighting problems and revealing possible
5. Envelope when the fouling affects both the heat
solutions.
transfer and the pressure drop

In the light of the points made here, the following


good-practice recommendations are made.
• More design software should be developed
which allows for the variation of the fouling
resistance with local conditions.
• Exchanger users should try to give more
information on fouling than just a single-value
fouling resistance. They should, at the very least
state the range of velocity (or wall/bulk
temperature) for which the specified resistance is
valid.
• Designers should then be expected to design
within the above limits and to set out the flow
and temperature operating limits beyond which
the exchanger will foul badly.
• Operators should be made aware of the above
operating limits.
• In difficult cases, dialogue should be retained
between the user and the designer in order to
modify the initial specification should this give
greater scope to minimize fouling

235 RS1084

You might also like