PIPING ENGINEERING CELL

Design of Tall Pressure Vessels

PROF. A. S. MOHARIR
Indian Institute of Technology Bombay

Introduction
Several chemical engineering processes require tall pressure vessels with cylindrical body and suitable
closure shapes. Common examples include distillation columns, absorption towers etc. The calculation of
pressure thickness for these towers are governed by the same formulae discussed earlier. However, by
virtue of their height, these vessels are also subject to several other non-process loads. For example, they
face wind load. Wind velocity increases as we go above the ground level. It is believed that wind velocity
is zero at ground level and is proportional to 1/7 power of height from ground level. Wind blowing at a
velocity is obstructed by these vessels. The kinetic energy of the wind is converted to pressure acting on
the vessel. The force created due to this can cause the tower to overturn. It has to be strong enough to
withstand this wind load. It may require increasing the wall thickness over and above the pressure
thickness. Similarly, tall structures are more prone to failure due to seismic vibrations of the Earth. If the
location of the process plant falls in a seismic zone which has had, or is likely to witness, earthquake, the
tower design must be checked for this aspect. A tall tower itself may support several other structures. For
example, a reboiler of a distillation column may be supported from the column itself. There could be a
platform or a ladder or a piping system whose weight is supported by the column etc. These loads also
could cause eccentric load on the column and add to its tendency to bend. This needs to be considered at
design stage. This chapter deals with such non-process loads on tall columns and presents design
considerations in brief.

Wind Load Considerations

When the wind is obstructed by a structure which comes in its way, its kinetic energy gets converted to
pressure energy. This pressure acts normal to the surface of the structure which obstructs the wind.
Product of surface area obstructing the wind and the pressure created on this surface because the wind
was obstructed is the force acting on the structure. This force will overturn the structure if the structure is
not strong enough. Even if the structure does not collapse because of wind, it will have a tendency to bend

1
PIPING ENGINEERING CELL

in the direction of the wind. This bending moment (also called overturning moment) will create stresses in
structure. If these stresses cross a certain threshold, the structure will be deemed to have failed.

Higher the wind velocity and more the area of the structure obstructing the wind, more will be these
stresses. These stresses can be reduced only by increasing the thickness of the structure. For the design to
be safe, we may have to increase the thickness of the tall column. The thickness should not only be
adequate to withstand the internal pressure, but also the wind load in addition to that.

As said earlier, design is always for worst conditions. The tall column design, therefore, should be
adequate to withstand the highest wind speed expected at the location where the plant is located. The
worst wind velocity is normally taken as the highest wind speed recorded at the location in the past. If the
geographical location is prone to cyclones, such as coastal belts etc., one would have to find out the
highest wind speeds recorded in the past. This data is normally available with meteorological
departments. The wind monitors are generally at 10 m or more height from ground level. The recorded
velocities are at this height. However, the wind speed will be more as we go higher and higher. As
mentioned earlier, wind velocity increases with height from ground level as per a one-seventh power law.
It is proportional to the ‘height to the power 1/7’.

What is used as wind velocity is what is reported for a location, without explicitly correcting for the
height. The velocity can be called as a design velocity. It is normally the worst wind speed at that
geographical location in the past 100 years or so. Let the design wind velocity be V km/hr (KMPH) or
miles/hr (MPH) etc. If wind blowing at a known velocity is obstructed by a flat surface perpendicular to
wind direction, the pressure created by the wind will be the maximum. If the object obstructing the wind
has a curved surface (just as our cylinder is), the wind at same velocity will create a somewhat lower
pressure on the shape because it can flow around the surface with relatively more ease. Therefore, shape
of the obstruction is also important. If the obstruction is offered by a flat plate, a commonly used relation
between wind velocity and wind pressure is as follows.

PW = 0.0025 VW2,

where, VW is the wind velocity in Miles per hour and PW is the wind pressure in pounds per square foot,
i.e. psf. The wind pressure is reduced by a factor, called shape factor, to account for shape of the
obstructing object. For cylinder, the shape factor is 0.8. The wind pressure can thus be calculated as:

PW = 0.8 x 0.0025 VW2 = 0.002 VW2

2
PIPING ENGINEERING CELL

This pressure is expected to be created on the projected surface of the cylinder as seen by the wind. If the
column is insulated, the wind will see the cylinder as of diameter which includes the insulation. It is
called ‘insulated’ diameter of the cylinder. If the cylinder height is H, The projected area of the insulated
cylinder will be

A = HDO

The force exerted by the wind on the cylinder will be then:

F = PW A = PW H DO

in appropriate units. If wind pressure is in psf (pounds per square foot), the height and diameter should
also be in feet. The projected area will then be in ft2, and force F will be in pound force.

If the cylinder has some attachments to it such as platforms, ladders, some auxiliary equipment supported
by the cylinder itself (such as say a reboiler supported by the tall vessel itself), or piping, their projected
area will also obstruct the wind and cause a force on the vessel which supports them.

Therefore, the area calculated above should include such areas also. But such appended areas could be
distributed along the vessel perimeter. For example, the ladder and platform could be on the eastern side
while the equipment could be on the western side and another ladder and platform on the northern side of
the vessel. If wind is blowing from east towards west, both the ladders and platforms are covered by the
projected area of the column. The equipment area is separate and seen as obstructing the wind. The
projected area should thus be the projected area of the insulated cylinder plus the projected area of the
attached equipment. If the wind blows from north to south, the equipment contribution will be ignored,
while both the ladders/platforms will be separately seen by the wind and their projected areas should be
counted wile calculating the force on the column. We have no control on wind direction. Therefore, the
projected area should be the maximum area we see as we take a 360 degrees tour around the column.
Attachments to the vessel are actually distributed along the perimeter for some other reason as we will see
later. The above point should therefore be kept in mind.

There is another important consideration that needs to be taken into account. Just as we took into account
the shape of the obstructing object, we also account for the features of the neighborhood of the tall vessel
we are designing. The wind velocity is generally measured by placing the wind monitor at not less than
10 m from ground level and possibly atop a reasonably tall building. Measurement is generally in urban
areas. For example, it would be atop a building where meteorological laboratory is situated. However, if it
is situated in an area with scattered trees, but not too many tall structures, the readings could be different.

3
PIPING ENGINEERING CELL

If it is situated at a location where the wind is not obstructed at all, such as say on a nearby sea shore, it
would measure the velocity of the wind as it is. If the location of the plant is in an area which has a lot of
buildings around it on all sides, the wind will lose some of its energy while blowing through the
surrounding structures. If, however, the location of the equipment is such that there is not much
population of tall structures or tall trees around it, the tall vessel will see the full fury of the wind. Even in
areas where there are tall structures or trees, if the vessel height is going to be more than such surrounding
things, it will still face unobstructed wind, at least on its part which rises above the average height of the
other structures in the vicinity. It is not possible to actually quantify effect of actual surroundings on wind
velocity or wind pressure. The effect is captured by multiplying the wind pressure by what is called as a
‘Gust Response Factor’ which captures the effect of locational aspects and vessel height broadly. The
locational features are captured by categorizing the location in terms of its Exposure Coefficient as
follows.

Exposure B: Urban/suburban areas

Exposure C: Open Terrain with scattered obstructions

Exposure D: Flat, unobstructed areas

For each exposure class, gust response factors for various heights of tall columns are suggested. These are
tabulated in the following table.

Coefficient G (Gust response factor combined with Exposure Coefficient)

Height
Above ground, ft (m) Exposure B Exposure C Exposure D

0-15 (0 – 4.6) 0.6 1.1 1.4

20 (6.1) 0.7 1.2 1.5
40 (12.2) 0.8 1.3 1.6
60 (18.3) 0.9 1.4 1.7
80 (24.4) 1.0 1.5 1.8
100 (30.5) 1.1 1.6 1.9
140 (42.7) 1.2 1.7 2.0
200 (60.9) 1.4 1.9 2.1

The gust response factor (G) and use it in the calculation of the wind force F as follows.

4
PIPING ENGINEERING CELL

F = G PW A

This lateral force acting on the column will tend to bend the column in the direction of the wind. The
bending moment will be maximum at the bottom of the column. This bending will create tensile axial
stresses on the side from where wind is reaching the column and compressive axial stresses on the other
side. For example, if the wind is blowing from East towards West, the East side of the column will suffer
tensile stresses and the West side compressive stresses. The East side is called the ‘Windward’ side and
the West side as ‘Leeward’ side, in this case as shown in the figure below:

The important point to note is that the wind forces cause axial stresses only and do not contribute to the
Hoop’s stresses in a cylinder. Also, these stresses are not of same nature. They are tensile on the
windward side and compressive on the leeward side. These are of same magnitude.

For our case, projected area A of the object above the skirt bottom is the entire projected area of the
column and skirt. If total height of the column including the skirt is H, The force acting on the column
will be

5
PIPING ENGINEERING CELL

F = PW Dins H

F is uniformly distributed across the height H. The column is firmly affixed to the skirt and the skirt to the
foundation. The bending moment M at the foundation level (skirt bottom) is given as follows.

M = F (H/2)

It is not difficult to appreciate this as the uniformly distributed force F can be seen as acting at the centre
of mass (at H/2).

The induced stress at the skirt bottom is given as follows.

S = M / [(π R2) tS]

where ts is the skirt thickness, and R is the radius of the column. The skirt has no net process fluid
pressure acting on its wall. Hence, above is the only stress acting in the axial direction on the skirt bottom.
As per our design criteria, as long as this does not cross allowable stress of the skirt material at design
temperature, the skirt can be said to be safe. The minimum thickness of the skirt that is required which
can withstand wind load is thus given as follows.

tS ≥ M / [(π R2) Sa]

Skirt does not suffer process fluid corrosion, and hence no corrosion allowance needs to be provided. We
can apply mill tolerance (if applicable) and then recommend the next higher nominal plate thickness
available. The formula for the skirt’s regulation thickness is adequate provided the fabrication of the skirt
does not involve welding. Otherwise, the allowable stress should be reduced by multiplying it by weld
joint efficiency, E. It should be appreciated that E should be the weld joint efficiency of the girth welding.
The applicable formula should be:

tS ≥ M / [(π R2) E Sa]

The pressure vessel is subjected to internal pressure and its regulation thickness is required to withstand
the Hoop’s stress due to pressure. The vessel is itself subjected to wind load and one should find out the
stress (due to the force exerted by the wind) at the bottom of the vessel (at the joint of vessel to the skirt).
What is important to note is that the force on the structure above the point where one wants to calculate
the bending moment or stress is only important. The force due to wind pressure acting above lower
tangent of the pressure vessel, FT, relevant in this case is:

FT = PW DIns (H-hT)

6
PIPING ENGINEERING CELL

This will be uniformly distributed along the pressure vessel length (H-hT). It can also be visualized to act
at the centre of this length. The bending moment MT at the bottom tangent of the vessel (where skirt is
jointed to the vessel) is thus given as follows.

MT = FT (H – hT) / 2

The bending moment can be used to calculate axial stress. The stress will depend on the vessel thickness.
The vessel thickness that is required to handle the wind load can then be calculated. The formulae are as
used earlier except that the bending moment value is different.

S = MT / [(π R2) t]

If wind load was the only load which induces stresses in the pressure vessel, then the thickness of the
vessel to withstand the wind load (tW) should be:

tW ≥ MT / [(π R2) E Sa]

This thickness is required by the wind load alone. If wind load is the only load, then the vessel thickness
should be this plus corrosion allowance. The corrosion provided thickness will then be further corrected
for mill tolerance and the next available commercial plate thickness will be recommended. However, if
the vessel also has internal pressure, the pressure itself will induce stresses, both circumferential (Hoop’s)
and axial. The axial stress will then be a sum of stress developed due to internal pressure and stress
developed due to wind load. Looked otherwise, pressure will demand some thickness and the wind load
will demand some thickness. The total thickness required to simultaneously withstand the axial stress due
to internal pressure and wind can be calculated. If this is more than the pressure thickness required to
withstand Hoop’s stress due to pressure, then the axial stress will govern the thickness choice.

Hoop’s stress due to internal pressure is double the axial stress due to pressure. The thickness calculation
of a cylinder was based on the Hoop’s stress which was limiting. If it is set equal to allowable stress, the
axial stress due to pressure will be only half of the allowable. The cylinder designed for internal pressure
can thus handle additional axial stress generated by other loads (such as wind load) as long as this
additional stress is not more than half the allowable. In that case, the cylinder thickness is as
recommended earlier for pressure design. Design must take into account all loads and the stresses
developed by them in the vessel skin. A thickness which does not allow either the Hoop’s stress or the
axial stress due to all these loads to cross the allowable stress in dirty corroded condition is the safe
thickness.

7
PIPING ENGINEERING CELL

Codes give a formula to calculate thickness required to keep axial stress at allowable stress for a cylinder
subjected to internal pressure. It is as follows. It should be noted that E should be the girth welding
efficiency.

tP = P DO /[4 Sa E + 0.8P]

The vessel wall thickness necessary to keep axial stresses due to internal pressure and wind load is thus
(tP + tW). At the same time, vessel wall thickness required to keep Hoop’s stress due to internal pressure is
as per the regulation thickness formula as given below.

t = P DO /[2 Sa ES + P Y]

Note that seam weld joint efficiency is involved in this formula. The thickness that is the maximum of the
two [(tP + tW) and t] will be what is required to keep both the Hoop’s and axial stresses within allowable
stress. To this, one would then add corrosion allowance, apply mill tolerance and recommend the next
available plate nominal thickness for vessel fabrication.

Seismic Load Considerations

Seismic activity manifests itself as a lateral force on the object resting on the Earth’s surface. This could
be a building or our tall pressure vessel. This lateral force is also vibratory in nature. It is believed that an
Earthquake is a mixture of several vibratory forces with varying frequencies of vibration. One therefore
talks about seismic spectrum for a location on Earth. Geophysicists have studied seismicity of Earth in
great detail. The Earth has been divided into several seismic zones depending on its seismic activity. The
zones are called Zone 1 to Zone 4. Zone 2 is further divided into Zone 2A and 2B. The zones are ordered
as per increasing seismic activity. The order is Zone 1, Zone 2A, Zone 2B, Zone 3 and Zone 4. Thus a
structure in Zone 4 is likely to experience more severe seismic activity than an identical structure in Zone
3 and so on. One must, therefore, know the seismic zone for a location where the process plant is located.

To design a tall column so that it can withstand seismic activity expected in the applicable seismic zone,
the first step is thus to know the seismic zone the location belongs to. With that basic information
available for any plant location, one can start using a simplified procedure to do seismic design of a tall
tower on a skirt.

The approach is similar to what we followed for calculating bending moment for wind load, bending
stress due to wind to get the skirt/vessel thickness one needs to withstand this load. We get the force that

8
PIPING ENGINEERING CELL

will act on the column due to seismic activity, decide its distribution along the height of the column
(Force due to wind was assumed to be uniformly distributed along the height. Force due to seismicity is
non-uniformly distributed as we will see later.), calculate the bending moment of this force at base of the
skirt or lower tangent of the vessel or at any height along the vessel, relate the moment to axial stress it
will develop at any location, calculate the wall thickness that will be required to withstand this seismic
load and to keep the seismic stress within allowable stress of the MoC.

The shear force V due to seismicity acting on the skirt-column structure is calculated as follows.

V=ZICW/4

The calculation involves seismic zone factor (Z), seismic coefficient (C) and the weight of the structure
(W). Weight of the Vessel (W) has to be estimated carefully. It is the maximum weight of the vessel
during its life time. A good way is to think of the weight at three stages: weight of a freshly erected (but
not yet commissioned) vessel, weight of the erected vessel during hydrotest, and weight of the vessel
during operation. The maximum of these three weight estimates is taken as W in the formula to calculate
seismic shear force, V.

Seismic Zone Factor

The Seismic Zone Factor Z depends on the seismic zone. The Seismic Zone Factor is a numerical
multiplier to be used in the formula for seismic shear force calculation. The values are as given below:

. Zone 1: 0.075

. Zone 2A: 0.15

. Zone 2B: 0.2

. Zone 3: 0.3

. Zone 4: 0.4

Seismic Coefficient

9
PIPING ENGINEERING CELL

The seismic coefficient C is calculated as per the following formula.

C = 1.25 S/T2/3

S is the ‘Soil Characteristic Coefficient’. It depends on the type of soil that is found at the site of plant
location. Some call S as Site Coefficient as well. Just as we need to know the seismic zone our site falls
into to get seismic zone coefficient, we need to know the soil characteristic to get the value of soil
characteristic coefficient.

The value of the Soil Characteristic Coefficient S is suggested as follows.

. For stiff or dense soil condition prevailing up to a depth of 200 ft from surface: S= 1

. For stiff or dense soil condition prevailing beyond a depth of 200 ft from surface: S = 1.2

. For soil profile ≥ 40 ft but with ≤ 40 ft of soft (or medium) clay : S = 1.5

. For soil profile with more than 40 ft of soft clay: S = 2.0

In short, as higher and higher depth of soil is comprised of soft clay, the more and more is the soil
characteristic coefficient and hence higher the seismic shear force acting on the structure.

T is the natural or fundamental period of vibration of the skirt-vessel structure. Period of vibration is of
course the reciprocal of frequency of vibration. It is calculated as per the following formula:

T = 0.0000265 (H/D)2 [WD/(H tskirt)]0.5

D is the diameter of the vessel, H is the height of the tower from skirt base as earlier, W is the weight of
the vessel-skirt assembly (maximum of erection, hydrotest and operating weight as elaborated earlier) and
tskirt is the thickness of the skirt. The formula is useful only with specific units for various quantities. This
is true for most formulae given so far.

To avoid resonance, the period of vibration of our tall tower, T, calculated as above should be restricted
as given below.

T ≤ Ta

where, Ta = 0.8 [W H /(V g)]0.5. Ta is the maximum allowable period of vibration for our structure of
given weight (W), given height (H), and given seismic shear force (V). g is the acceleration due to
gravity. All quantities must be in appropriate units.

10
PIPING ENGINEERING CELL

The only way to bring T below Ta is to increase the skirt thickness. Having got T, one can calculate V, the
seismic shear force. Having got this force acting on the structure laterally, one can proceed to calculate
the bending moment at the bottom of the skirt, bottom of the vessel, at any other location on the vessel
etc. and get contributions to axial stress in skirt and vessel wall and therefore the skirt thickness and
vessel thickness required specifically to keep axial stress within allowable stress.

The procedure here after is best understood with reference to a column schematic as below. H is the total
height of the skirt and vessel from ground level. hT is the skirt height as earlier (normally 4 ft.). X is the
distance of any level on the structure as measured from the top tangent of the vessel.

Unlike wind force which was distributed uniformly across the column, the seismic force V, calculated as
discussed earlier, is visualized to be distributed as follows.

1. A part of the force V is visualized as acting at the top tangent of the vessel, i.e. at height H from
the skirt bottom. Let us call this part of the force as Ft. Ft is calculated as follows.

Ft = 0 (if T ≤ 0.7 second)

Ft = 0.07 T V (if T > 0.7 second)

11
PIPING ENGINEERING CELL

V is the seismic shear force calculated as discussed earlier and T is the natural period of vibration
of the skirt-vessel structure in seconds calculated as per formula given earlier.
2. The balance seismic force (V – Ft) is deemed to act in a distributed fashion across the entire
height H. The distribution is non-uniform, but linear. The force is zero at the skirt bottom and
increases linearly as we go up to the top tangent. This distribution is also equivalent to a point
force (V – Ft) acting at a height 2H/3 from the skirt bottom or H/3 from the top tangent.

If we consider the equivalent picture of the force distribution as Ft acting at height H (or X = 0) and (V -
Ft) acting at height 2H/3 (or X = H/3), the bending moment MX at various distances X from the top
tangent can be written as follows.

MX = Ft X for X ≤ H/3

MX = Ft X + (V- Ft)(X – H/3) for X > H/3

We can get bending moment M at base of skirt by substituting X = H in the above formula.

M = Ft H + (V- Ft)(2 H/3)

Similarly, bending moment at the lower tangent of the vessel can be calculated by substituting X = (H-hT)
in the formula.

The bending moment at the skirt bottom can be calculated. Bending stress is related to bending moment
as per formula given earlier while discussing wind load. Similarly, one can calculate bending moment at
lower tangent of the vessel and bending stress corresponding to it in the vessel at that location. One can
similarly calculate the wall thickness demanded by the seismic load at the bottom tangent of the vessel so
that it does not cross allowable stress of the MoC. Let that be tS.

For the vessel, we have maximum thickness demand at the lower tangent and it is by each one of the
loads. The pressure demands tP, the wind demands tW and the seismic load demands tS. The guideline is to
consider only the maximum thickness of the two required, one for wind load and another for seismic load.

Eccentric Load

12
PIPING ENGINEERING CELL

Tall columns need to consider one more load. It is the load acting on the column because of various
attachments such as auxiliary equipment, platforms, ladders etc. If the column supports the weight of
these attachments and the attachments are not distributed uniformly along the column cross-section, the
column will face some net weight on one side with its centre of gravity away from the column axis. Such
a weight is called eccentric load and will cause a bending moment to act on the column. The bending
stresses due to such an eccentric load will also be axial in nature and will demand some wall thickness for
themselves just as pressure, wind load and seismic load did. Once we get the bending moments at the base
of the skirt and the lower tangent of the vessel, we know how to convert it to axial stresses and the
thickness requirement.

Consider a weight W with its centre of gravity separated by a distance of ‘e’ from the axis of skirt/vessel.
‘e’ is termed as the eccentricity of the load. It creates a bending moment M in the vessel and skirt wall. It
is given as:

M=We

This may be considered as same at the skirt base as well as any location on the vessel. The bending stress
for both the skirt and the vessel will then be M/[π R2t] if t is the wall thickness of skirt or vessel. The
thickness required to withstand this load without the stress crossing the allowable stress of the MoC will
then be M/[π R2EG Sa]. The weld joint efficiency should be of the girth welding as discussed earlier. A
subscript G is used for the weld joint efficiency as a reminder of this. This thickness should be considered
while deciding the skirt thickness as well as vessel thickness in addition to the demand made by other
loads.

Column Deflection

Another small consideration which has a bearing on the skirt thickness is the deflection of the column at
the top tangent from the original tower axis due to wind load. Generally, a deflection of about 6 inches
per 100 feet height of column is considered as acceptable. The deflection can be adjusted by adjusting the
skirt thickness. Thicker the skirt wall, lesser will be the deflection.

The wind force F at the skirt bottom can be used to calculate the deflection ∆L at the top as follows.

∆L = F (12 H)3/(8 E I)

13
PIPING ENGINEERING CELL

E is the Young’s modulus of elasticity and I is the moment of inertia of the skirt shape. Skirt is a cylinder
with known diameter and thickness and its moment of inertia can be calculated easily. Moment of inertia
of the skirt can be calculated as follows:

I = R3 π tskirt

The maximum allowed deflection in inches can be calculated as

∆Lmax = 6 H /100

If ∆L is more than ∆Lmax, then the skirt thickness can be increased. This will reduce ∆L as per the
formula given and by suitably increasing the skirt thickness, it can be brought below ∆Lmax.

Conclusion

Most pressure vessel design courses avoid detailed discussion of tall pressure vessel (also called tower)
design covered in this chapter. The reason is that the design is considered as too laborious and is very
demanding on the inputs that are required. At the same time, it is an important design task as many
vessels in a process industry can be tall. Even if the vessels are not tall, a stack (or chimney) in a process
plant is a tall structure. Metallic stacks are often used. It also should be designed using above principles
because stack faces all the loads except that it does not need to be designed for internal pressure because
effluent gases are just slightly above the atmospheric pressure (their pressure could be just a few mmWC
more than the atmospheric). For example, its pressure could be 5 mmWC above atmospheric. mmWC is
mm of Water Column. Remember 1 atmosphere pressure is about 10000 mmWC.

14