CoverInQuark:Layout 1

12/19/11

10:40 AM

Page 1

Steel Plate Engineering Data-Volume 1

Steel Tanks for Liquid Storage

Revised Edition - 2011

The material presented in this publication is for general information

only and should not be used without first securing competent advice
with respect to its suitability for any given application. The publication
of the material contained herein is not intended as a representation or
warranty on the part of the Steel Market Development Instituteor of
any other person named hereinthat this information is suitable for
any general or particular use or of freedom from infringement of any
patents. Anyone making use of this information assumes all liability
arising from such use.

Published by
STEEL MARKET DEVELOPMENT INSTITUTE,
A business unit of the American Iron and Steel Institute

In cooperation with and editorial collaboration by

STEEL PLATE FABRICATORS ASSOCIATION, Div. of STI/SPFA

Acknowledgements

cknowledgement is given to the important and valuable contribution made by members of

the Steel Plate Fabricators Association in reviewing and updating the material for
publication in this current edition.
A special note of appreciation is given to Stephen W. Meier, P.E., S.E. of Tank Industry Consultants
for his effort in updating this publication.
The Steel Market Development Institute gratefully acknowledges the continued investment of its
investor steel-producing companies in the steel pipe and tank markets.

Copyright Steel Market Development Institute 2011

STEEL MARKET DEVELOPMENT INSTITUTE

25 Massachusetts Avenue NW, Suite 800
Washington D.C. 20001
ii

Introduction

he purpose of this publication is to provide a design reference for the usual design of tanks
for liquid storage. Volume 1, "Steel Tanks for Liquid Storage, deals with the design of flatbottom, cylindrical tanks for storage of liquids at essentially atmospheric pressure. Volume 2,
"Useful Information on the Design of Plate Structures, provides information to aid in design of such
structures.
For unusual applications, involving materials or liquids not covered within these pages, nor
referenced herein, designers should consult more complete treatments of the subject material.
Part I contains general information pertaining to carbon plate steels. This section is most helpful
to readers who are not intimately familiar with steel industry terminology, practice and classification.
Part II deals with the particular carbon steels applicable to tanks for liquid storage.
Part III covers the design of carbon steel tanks for liquid storage.
Part IV covers materials, design, and fabrication of stainless steel tanks for liquid storage.

Inquiries for further information on the design of steel tanks should be directed to:
Steel Plate Fabricators Association
Division of STI/SPFA
944 Donata Court
Lake Zurich, IL 60047
www.steeltank.com
iii

Contents

Part I
Part II
Part III
Part IV

MaterialsGeneral .............................................................................. 1
MaterialsCarbon Steel Tanks for Liquid Storage......................... 7
Carbon Steel Tank Design................................................................. 11
Stainless Steel Tanks for Liquid Storage ......................................... 33

iv

Part I
MaterialsGeneral
Designation
ost of the steel specifications
referred to in this manual can be
obtained from the American Society
for Testing and Materials (ASTM).
Each ASTM specification has a number such
as A283, and within each specification there
may be one or more grades or qualities. Thus
an example of a proper reference would be
ASTM designation A283 grade C." In the
interest of simplicity, such a reference will be
abbreviated to "A283C."
ASTM standards are issued periodically to
report new specifications and changes to
existing ones having a suffix indicating the
year of issue such as "A283-C-03." Thus a
summary such as is provided here may
gradually become incomplete, and it is
important that the designer of steel plate
structures have the latest edition of ASTM
standards available for reference.

certain elements increases in the liquid portion

of the ingot. The resulting product, known as
RIMMED STEEL, has marked differences in
characteristics across the section and from top
to bottom of the ingot.
Control of the amount of gas evolved during
solidification is accomplished by the addition
of a deoxidizing agent, silicon being the most
commonly used. If practically no gas evolved,
the result is KILLED STEEL, so called because
it lies quietly in the ingot. Killed steel is
characterized by more uniform chemical
composition and properties than other types.
Although killed steel is a quality item, the end
result is often not so specified by name, but
rather by chemical analysis. Other deoxidizing
elements are used, but in general, a specified
minimum silicon content of 0.10% on heat
analysis indicates that a steel is "fully killed."
The term SEMIKILLED designates an
intermediate type of steel in which a smaller
amount of deoxidizer is added. Gas evolution
is sufficiently reduced to prevent rimming
action, but not sufficiently reduced to obtain
the same degree of uniformity as attained in
fully killed steels. This controlled evolution of
gas during solidification tends to offset
shrinkage, resulting in a higher yield of usable
material from the ingot. As a practical matter,
therefore, plates originating from ingots are
usually furnished as semikilled steel unless a
minimum silicon content of 0.10% on heat
analysis is specified.
The steels with which we are concerned are
either continuous cast or cast into ingots. The
ingots may be hot rolled to a convenient size
for further processing, or they may be rolled
directly into plates. The current practice is
mostly to use continuous casting of the steel.
The steel used for continuous casting is fully
deoxidized.

Definitions
At least a nodding acquaintance with the
terminology of the steel industry is essential to
an understanding of steel specifications. This
is especially true because, in common with
many other industries, a number of shop and
trade terms have become so thoroughly
implanted in the language that they are used
instead of more precise and descriptive
technical terms. The following discussions
may be of assistance.
Steelmaking Processes
Practically all steel is made by the electric
furnace process or the basic oxygen process.
ASTM specifications for the different steels
specify which processes are permissible in
each case.
Types of Steel
In most steelmaking processes, the principal
chemical reaction is the combination of carbon
and oxygen to form a gas. If the oxygen
available for this reaction is not removed, the
gaseous products continue to evolve during
solidification in the ingot. Cooling and
solidification progress from the outer rim of
the ingot to the center, and during the
solidification of the rim, the concentration of

Chemical Requirements
A discussion of the effects of the many
elements added to steels would involve a
metallurgical treatise far beyond the scope of
this work. However, certain elements are
common to all steels, and it may be of help to
briefly outline the effects of carbon,
manganese, phosphorus, and sulfur on the
properties of steel.
1

There are some exceptions to these rules in

High Strength Low Alloy (HSLA) steels.

CARBON is the principal hardening element

in steel, and as carbon increases, hardness
increases. Tensile strength increases, and
ductility, notch toughness and weldability
generally decrease with increasing carbon
content.
MANGANESE contributes to strength and
hardness, but to a lesser degree than carbon.
Increasing the manganese content generally
decreases ductility and weldability, but to a
lesser degree than carbon. Because of the more
moderate effects of manganese, carbon steels,
which attain part of their strength through the
addition of manganese, exhibit greater
ductility and improved toughness than steels
of similar strength achieved through the use of
carbon alone.
PHOSPHORUS. Phosphorus can result in
noticeably higher yield strength and decreases
in ductility, toughness, and weldability. In the
steels under discussion here, it is generally
kept below a limit of 0.04% on heat analysis.
SULFUR decreases ductility, toughness, and
weldability, and is generally kept below a limit
of 0.05% on heat analysis.
HEAT ANALYSIS is the term applied to the
chemical analysis representative of a heat of
steel and is the analysis reported to the
purchaser. It is usually determined by
analyzing, for such elements as have been
specified, a test ingot sample obtained from
the front or middle part of the heat during the
pouring of the steel from the ladle.
PRODUCT ANALYSIS is a supplementary
chemical analysis of the steel in the
semifinished or finished product form. It is
not, as the term might imply, a duplicate
determination to confirm a previous result.

Alloy Steel
Steel is usually considered to be alloy when
either:
1. A definite range or definite minimum
quantity is required for any of the
elements listed above in (1) under
carbon steels, or
2. The maximum of the range for alloying
elements exceeds one or more of the
limits listed in (2) under carbon steels.
Again, the HSLA steels demonstrate
some exceptions to these general rules.
High Strength Low Alloy Steels
These steels, generally with specified yield
point of 50 ksi or higher and containing small
amounts of alloying elements, are often
employed where high strength or light weight
is desired.
Mechanical Requirements
Mechanical testing of steel plates includes
tension, hardness, and toughness tests. The
test specimens and the tests are described in
ASTM specifications A6, A20, A370, and A673.
From the tension tests are determined the
TENSILE STRENGTH and YIELD POINT or
YIELD STRENGTH, both of which are factors
in selecting an allowable design stress, and the
elongation over either a 2" or 8" gauge length.
Elongation is a measure of ductility and
workability.
Toughness is a measure of ability to resist
brittle fracture. Toughness tests are generally
not required unless specified, and then usually
because of a low service temperature and/or a
relatively high design stress. Conditions under
which impact tests are required or suggested
will be discussed in connection with specific
structures.
A number of tests have been developed to
demonstrate toughness, and each has its
ardent proponents. The test most generally
accepted currently, however, is the test using
the Charpy V Notch specimen. Details of this
specimen and method of testing can be found
in ASTM-A370, "Mechanical Testing of Steel
Products," and in A20 and A673. Briefly
described, an impact test is a dynamic test in
which a machined, notched specimen is struck
and broken by a single blow in a specially
designed testing machine. The energy
expressed in foot-pounds required to break the

Carbon Steel
Steel is usually considered to be carbon steel
when:
1. No minimum content is specified or
required
for
chromium,
cobalt,
columbium,
molybdenum,
nickel,
titanium,
tungsten,
vanadium,
zirconium, or any other element added
to obtain desired alloying effect;
2. When the maximum content specified
for any of the following elements does
not exceed the percentages noted:
manganese 1.65, copper 0.60, silicon 0.60;
3. When the specified minimum for copper
does not exceed 0.40%.
2

specimen is a measure of toughness.

Toughness decreases at lower temperatures.
Hence, when impact tests are required, they
are usually performed near temperatures
anticipated in service.

Referring to Figure 1-1, if the designer has

selected a Charpy V Notch value of "x ft.-lbs,
as desirable under special service conditions, it
will be noted that the steel illustrated would
not be acceptable at temperatures lower than
about +35F in the as-rolled condition. In the
normalized condition, the same steel would be
acceptable down to about -55F, and if
quenched and tempered, to about -80F
together with an increase in carbon,
manganese, or other hardening elements.
Note, however, that heat treatment adds to the
cost and is indicated only when service
conditions indicate the necessity for increased
toughness and/or increased strength.

Grain Size
Grain size is affected by both rolling practice
and deoxidizing practice. For example, the use
of aluminum as a deoxidizer tends to produce
finer grains. Unless included in the ASTM
specification, or unless otherwise specified,
steels may be furnished to either coarse grain
or fine grain practice at the producer's option.
Fine grain steel is considered to have greater
toughness than coarse grain steels. Heattreated fine grain steels will have greater
toughness than as-rolled fine grain steels. The
designer is concerned only with the question
of under what conditions it is justifiable to pay
the extra cost of specifying fine grain practice
with or without heat treatment in order to
obtain improved toughness. Guidelines will be
discussed in later sections.

Classification of Steel Plates

Plate steels are generally defined or
classified in two ways. The first classification,
which has already been discussed, is based on
differences in chemical composition between
CARBON STEELS, ALLOY STEELS and HIGH
STRENGTH LOW ALLOY STEELS. The
second classification is based primarily on the
differences in extent of testing between
STRUCTURAL QUALITY STEELS and
PRESSURE VESSEL QUALITY STEELS1. It
should not be construed that these terms limit
the use of a particular steel. Pressure vessel
steels are often used in structures other than
pressure vessels. The distinction between
structural and pressure vessel qualities is best
understood by a comparison of the governing
ASTM specifications.
ASTM
designation
A6,
General
Requirements for Rolled Steel Plates for
Structural Use, covers a group of common
requirements and tolerances for the steels
listed therein, the chemical composition and
special requirements for which are outlined
under separate specification numbers such as
A36, A283, A514, etc.
Similarly, ASTM designation A20, General
Requirements for Steel Plates for Pressure
Vessels, covers a group of common
requirements and tolerances which apply to a
list of about 35 steels, the chemical
composition and special requirements for
which are outlined under separate ASTM
specification numbers.

Heat Treatment
POST-WELD HEAT TREATMENT consists
of heating the steel to a temperature between
1100F and 1250F, furnace cooling until the
temperature has reduced to about 600F and
then cooling in air. Residual stresses will be
reduced by this procedure.
NORMALIZING consists of heating the steel
to between 1600F and 1700F, holding for a
sufficient time to allow transformation, and
cooling in air, primarily to affect grain
refinement.
QUENCHING consists of rapid cooling in a
suitable medium from the normalizing
temperature. This treatment hardens and
strengthens the steel and is normally followed
by tempering.
TEMPERING consists of reheating the steel
to a relatively low temperature (which varies
with the particular steel and the properties
desired). This temperature normally lies
between 1000F and 1250F. Through the
quenching and tempering treatment, many
steels can attain excellent toughness, and at the
same time high strength and good ductility.
To illustrate the effect of heat treatment on
toughness and strength, refer to Figure 1-1.
The numerical values shown apply only to the
specific steel described. For other steels, other
values would apply, but the trends would be
similar.

Pressure vessel quality steels were previously known as

FLANGE and FIRE-BOX qualities, historically inherited terms
used to define differences in the extent of testing, but which
have no present-day significance insofar as the end use of the
steel is concerned.
1

The arc is formed between the work to be

welded and a metal wire which is called the
electrode. The electrode may be consumable
and add metal to the molten pool, or it may be
nonconsumable and of a relatively inert metal,
in which case no metal is added to the
workpiece.
In the welding of steel plate structures, we
are concerned principally with five variations
of arc welding:
1. Shielded metal arc process (SMAW)
2. Gas metal arc process (GMAW)
3. Flux-cored arc process (FCAW)
4. Electrogas or Electroslag welding
5. Submerged arc process (SAW)

Both A6 and A20 define tolerances for

thickness, width, length, and flatness, but for
the designer the important difference is in the
quality of the finished product as influenced
by the difference in the extent of testing. A
general comparison of the two qualities
follows:
1. Chemical AnalysisThe requirements
for phosphorus and sulfur are more
stringent for pressure vessel quality than
for structural quality. Both A6 and A20
require one analysis per heat plus the
option of product analysis. Product
analysis tolerances for structural steels
are given in A6.
2. Testing for mechanical properties
a. In general, all specifications for
structural quality require two
tension tests per heat, size bracket
and strength gradation. A6 specifies
the general location of the
specimens.
b. In general, all specifications for
pressure vessel quality require either
one or two transverse tension tests,
depending on heat treatment, from
each plate as rolled2 (and as heattreated, if any). This affords a check
on uniformity within a heat.
Specification A20 also specifies the
location from which the specimens
are to be taken.
3. Repair of surface imperfections and the
limitations on repair of surface
imperfections are more restrictive in A20
than A6.

Shielded Metal Arc Welding

In the early days of arc welding, the
consumable electrode consisted of a bare wire.
The pool of molten metal was exposed to and
adversely affected by the gases in the
atmosphere. It became obvious that to produce
welds with adequate ductility, the molten
metal must be protected or shielded from the
atmosphere.
This led to the development of the shielded
metal arc process, in which the electrode is
coated with materials that produce a gas as the
electrode is consumed which shields the arc
from the atmosphere. The coating also
performs other functions, including the
possible adding of alloying elements as well as
slag-forming materials which float to the top
and protect the metal during solidification and
cooling.
In practice, the process is limited primarily
to manual manipulation of the electrode. Not
too many years ago, this process was almost
universally used for practically all welding. It
is still widely used for position welding, i.e.,
welding other than in the down flat position.
For the down flat position some of the later
processes described below are much faster and
hence less costly.

Welding
Many plate structures are fabricated by
welding. A brief discussion of welding
processes follows.
Welding consists of joining two pieces of
metal by establishing a metallurgical bond
between them. There are many different types
of welding, but we are concerned only with
arc welding. Arc welding is a fusion process in
which the bond between the metals is
produced by reducing the surfaces to be joined
to a liquid state and then allowing the liquid to
solidify. The heat required to reduce the metal
to liquid state is produced by an electric arc.

Gas Metal Arc Welding

In the gas-shielded arc welding process, the
molten pool of metal is protected by an
externally supplied gas, or gas mixture, fed
through the electrode holder rather than by
decomposition of the electrode coating. The
electrode is a continuous filler-metal
(consumable) bare wire and the gases used
include helium, argon, and carbon dioxide. In
some cases, a tubular electrode is used to

2 The term Plate as rolled refers to the unit plate rolled

from a slab or directly from an ingot in relation to the number
and location of specimens, not to its condition.

Weldability
It will be observed from the above that all
arc welding processes result in rapid heating
of the parent metal near the joint to a very high
temperature followed by chilling as the
relatively large mass of parent plate conducts
heat away from the heat-affected zone. This
rapid cooling of the weld metal and heataffected zone causes local shrinkage relative to
the parent plate and resultant residual stresses.
Depending on the chemical composition of
the steel, plate thickness and external
conditions, special welding precautions may
be indicated. In very cold weather, or in the
case of a highly hardenable material, preheating a band on either side of the joint will
slow down the cooling rate. In some cases
post-heat or stress relief as described earlier in
this section is employed to reduce residual
stresses to a level approaching the yield
strength of the material at the post heat
temperature.
With respect to chemical composition,
carbon is the single most important element
because of its contribution to hardness, with
other elements contributing to hardness but to
lesser degrees.
It is beyond our scope to provide a definitive
discussion on when special welding
precautions are indicated. In general, the
necessity is dictated on the basis of practical
experience or test programs.

facilitate the addition of fluxes or addition of

alloys and slag-forming materials. Some
methods of this process are called MIG and
CO2 welding.
The gas-shielded process lends itself to high
rates of deposition and high welding speeds. It
can be used manually, semi-automatically, or
automatically.
Flux-Cored Arc Welding
This is an arc welding process wherein
coalescence is produced by heating with an arc
between
a
continuous
filler-material
(consumable) electrode and the work.
Shielding is obtained from a flux contained
within the electrode. Additional shielding may
or may not be obtained from an externally
supplied gas or gas mixture.
Electrogas or Electroslag Welding
This process is a method of gas metal arc
welding or flux-cored arc welding wherein
molding shoes confine the molten weld metal
for vertical position welding.
Submerged Arc Welding
Submerged arc welding is essentially an
automatic process, although semi-automatic
applications have been used.
The arc between a bare electrode and the
work is covered and shielded by a blanket of
granular, fusible material deposited on the
work ahead of the electrode as it moves
relative to the work. Filler metal is obtained
either from the electrode or a supplementary
welding rod. The fusible shielding material is
known as melt or flux.
In submerged arc welding, there is no visible
evidence of the arc. The tip of the electrode
and the molten weld pool are completely
covered by the flux throughout the actual
welding operation. High welding speeds are
achieved.
It will be obvious that the necessity of
depositing a granular flux ahead of the
electrode lends itself best to welding on work
in the down flat position. Nevertheless,
ingenious devices have been developed for
keeping flux in place, so that the process has
been applied to almost all positions except
overhead welding.

Figure 1-1
Typical Effect of Heat Treatment on Notch Toughness
Of a Fine-Grained C-Mn-Si Steel (1 Inch Thickness)

Part II
MaterialsCarbon Steel Tanks For Liquid Storage
Introduction
he intent of this publication is to
provide information that may be
useful in the design of flat-bottom,
vertical cylindrical tanks for the storage of
liquids at essentially atmospheric pressure.
Considerable attention has been directed to
tanks storing petroleum-based liquids or
water, which constitute most of the tanks built.
However, suggestions have been included for
storage of liquids meriting special attention,
such as acid storage tanks.
There are two principal standards in general
use in the U.S.: the American Petroleum
Institute (API) covering welded steel tanks for
petroleum storage, and the American Water
Works Association (AWWA) covering tanks
for water storage. The abbreviations API and
AWWA will be used for the sake of
convenience.
While API has developed and maintains
numerous
standards
related
to
the
construction, operation and inspection of
tanks, the API 650 Standard, Welded Tanks for
Oil Storage and the API 620 Standard, Design
and Construction of Large, Welded, LowPressure Storage Tanks are the commonly
used tank design basis standards.
AWWA has also developed and maintains
numerous tank-related standards for concrete
and steel tanks, included bolted and welded.
The most commonly applied and used in this
publication is AWWA D100- Welded Carbon
Steel Tanks for Water Storage.
Both API and AWWA permit the use of a
relatively large number of different steel plate
materials. In addition, the basic API Standard
650 and AWWA Standard D100 Section 14
provide refined design, construction and
inspection rules for tanks designed at higher
stresses in which the selection of steel is
intimately related to stress level, thickness and
service temperature, as well as the type and
degree of inspection. As a result, knowledge of
available materials and their limitations is
equally as important as familiarity with design
principles.
Useful information concerning plate steel in
general has been covered in Part I. It is the
purpose of this section to assist in the selection
of the proper steel or steels in the construction
of tanks for liquid storage.

Factors Affecting Selection of Steel Plate

As you will learn in more detail in Part III of
the publication, both the AWWA and the API
offer optional methods of shell design. The
AWWA basic procedures apply simplified
rules which use conservative allowable stress
levels. The optional design methods are based
on refined procedures that take into account
plate grade, service temperature, thickness and
higher standards of inspection.
It will be obvious that inasmuch as the
simplified design provisions of both standards
allow identical design stresses for any of the
permissible steels, economic considerations
will lead to the selection of the least expensive
steel that will be satisfactory for the intended
service.
Steel selection is not so simple and
straightforward in the case of tanks built in
accordance with either the API or the AWWA
Section 14 design provisions. Unstressed
portions of such tanks, including bottoms and
roofs, will probably be furnished as A36 unless
the purchaser specifies otherwise. The
selection of material for the shell demands
further attention.
The design provisions AWWA Section 14
resulted from a desire to utilize newer and
improved steels and modern welding and
inspection techniques to build tanks of higher
quality. The use of higher stresses demanded
attention to other properties of steel, primarily
toughness. An exhaustive discussion of
toughness is beyond the scope of this work,
but it can be pointed out that as the stress level
increases
and
temperature
decreases,
toughness becomes more important.
At the stress level existing in API and
AWWA basic design criteria tanks, experience
has demonstrated that the steels used in
combination with the specific welding and
inspection rules have proven adequate for the
service temperatures involved. Operating at
the field of higher stress levels of the optional
design methods requires steels having greater
toughness. Thanks to research in metals, such
steels are readily available. A number of
factors enter into making a proper selection.
For example, for any given steel, toughness
generally decreases as thickness increases. The
7

Although both the API and AWWA

standards permit the ordering of plates for
certain parts of the tank on a weight rather
than thickness basis, there is no longer any
economic advantage in doing so.

toughness of carbon steels is improved if part

of the hardness and strength is obtained by a
higher manganese content and lower carbon at
the same strength level. Fine-grained steels
exhibit greater toughness than coarse-grained
steels; this can be accomplished in the
deoxidizing process, and in heat treatment.
Thus as thickness increases and service
temperature
decreases,
more
stringent
attention must be paid to toughness from the
standpoint of materials selection and
fabrication.
The steels permitted by API and AWWA
Section 14 for use at these higher stress levels
have statistically demonstrated adequate
toughness for the thickness and temperature
ranges shown. The API standard includes an
Impact Exemption chart which establishes
requirements for impact testing, based on
thickness, temperature and type of material.
In the final analysis, the goal is to design the
most cost-effective but acceptable tank for a
given set of conditions. API and AWWA rules
permitting higher design stresses afford a
fairly wide selection of steels and stress levels
from which to choose.
A definitive treatment of economics is
beyond the scope of this work. Basically, the
factors involved are:
1. Cost of material
2. Weight of material as it affects freight
and handling
3. Fabrication, erection and welding costs
4. Inspection and QA/QC costs
None of these factors is necessarily
conclusive in itself. In any given case, the
lightest weight or lowest material cost may or
may not be the least expensive overall
depending on the relative importance of the
factors listed above. The tank fabricator is
usually in the best position to judge which
steel or combination of steels will permit
construction of the most economical, safe tank
based on current market conditions.
It is generally unwise to specify a more
expensive steel than can be justified by the
application.
There are material costs not associated with
quality. The cost of plates will vary according
to both width and thickness, and from this
consideration tank shell plate approximately 8'
wide will commonly be used. Particular
situations may dictate the use of wider or
narrower plates for all or part of a tank shell.
Plate widths of 10 ft to 12 ft are not
uncommon.

The Future
To this point, only those steels specifically
permitted by API or AWWA have been
discussed. Other steels have been used to a
minor extent by those thoroughly familiar
with the problems involved. Among these are
the materials referred to in Part I as high
strength low alloy steels, manufactured either
as proprietary, trade-named steels, or to ASTM
specifications. Some of these steels offer the
additional attraction of improved atmospheric
corrosion resistance, thus eliminating the
necessity for painting outside surfaces.
As is the case with all high-strength
materials, the designer and user must assure
themselves that factors other than strength
(toughness for example) are properly allowed
for in design and construction.
For obvious reasons, construction codes
often lag behind technical progress. The
extensive research facilities of individual steel
producers are constantly searching for ways to
better serve the needs of our modern economy.
But before any construction standard such as
those of API and AWWA can accept and
permit a new material, it must have been
established that it is suitable for the structure
in which it will be used.
Usually, but not always, acceptance by API
and AWWA implies prior acceptance by
ASTM. Primarily this is because ASTM
specifications clearly delineate the materials to
be furnished, whereas any departure from
ASTM requires that the standards involved
spell out the requirements in corresponding
detail. New ASTM steels may or may not
eventually find their way into the construction
standards, depending on economics and the
proven properties of the materials.
It should be left to those who have acquired
the necessary experience in tank design and
construction to pioneer in the use of materials
not approved by API or AWWA. The designer,
the user, and the fabricator assume added
responsibilities in working outside of
recognized industry standards. On the other
hand,
such
pioneering
by
qualified
organizations in the past led to the progress
represented by the refined procedures of
Section 14 of AWWA D100 and API-650.
8

As in the case of steels already approved by

API and AWWA, time and experience will
eventually lead to recognition of the steel or
combination of steels that will yield the
highest quality tank at least cost.

This page has been left blank intentionally.

Part III
Carbon Steel Tank Design
excellent. Very few tank failures have been
recorded under even abnormal conditions and
properly maintained steel tanks have endured
long past their original design lives. Before
applying them to tanks storing liquids other
than water or oil, the designer should consider
which philosophy best fits his circumstances. In
either case the design standards provide
minimum requirements for safe construction
and should not be construed as a design
manual covering all possible service conditions.

Introduction
art III will consider the design of flat
bottom, vertical, cylindrical, carbon steel
tanks for the storage of liquids at
essentially atmospheric pressure and near
ambient temperatures. Practically all tanks in
the United States within the scope of this part
are constructed in accordance with API 650
covering welded steel tanks for oil storage or
AWWA D100 covering welded steel tanks for
water storage.
Tanks of other shapes and subject to gas
pressure in addition to liquid head, and tanks
subject to extreme low or high temperatures
present radically different problems. Consult
ASME Section VIII, API 650 Appendices F and
M, and API 620 for further information.
API 650 and AWWA D100 contain detailed
minimum requirements covering inspection.
Any attempt to summarize the inspection
requirements of either standard would be
voluminous and dangerously misleading. It
will be the purpose of Part III to discuss only
those portions necessary to understand the
various design bases. Anyone concerned with
fabrication, erection, or inspection must obtain
copies of the complete standards.
There are basic differences between the
standards of API and AWWA. API 650 is an
industry standard especially designed to fit the
needs of the petroleum industry. The petroleum
tank is usually located in isolated areas, or in
areas zoned for industry where the probable
consequences of mishap are limited to the
owners property. The owner is conscious of
safety, environmental concerns and potential
losses in his operations, and will adjust the
minimum requirements to suit more severe
service conditions.
AWWA D100 is a public standard to be used
for the storage of water. The water storage tank
is often located in the midst of a heavily
populated area, often on the highest elevation
available. The consequence of catastrophic
mishap could not be tolerated in the public
interest.
The API 650 and AWWA D100 standards
have been in existence for many decades (since
the 1930s).
The performance of the tank
population throughout the U.S. has been

General Design Formula for Tank Shells

Membrane shell theory, as it applies to
cylindrical tanks of large diameter, is
elementary and needs no explanation here.
Starting with the basic premise that
circumferential load in a cylinder equals the
pressure times the radius, then expressing H
and D in feet for convenience, the
circumferential load at any level in a vertical
cylinder containing water weighing 62.4 #/ft3,
can be expressed as:
T = 2.6 HD
where T

(3-1)

the circumferential load per

inch of shell height
depth in feet below
maximum liquid level
tank diameter in feet

Then the minimum design thickness can be

expressed as:

t (inches) =
where

S
E
C

=
=
=

2.6 HDG
+C
SE

(3-2)

contained liquid specific

gravity
allowable design stress in psi
joint factor
corrosion allowance in inches

Obviously the ideal situation would be to

vary the thickness uniformly from bottom to
top, but since steel plates are rolled to a uniform
thickness, any given course of plates is uniform
throughout its width. Thus a course designed
for the stress at its lower edge will have excess
thickness at the top, which will help carry part
11

asymmetric loading criteria depending on the

design standard applied.

of the load in the lower portion of the course

above. API takes advantage of this and designs
each course of plates for the stress existing one
foot above the bottom of the course in question.
AWWA designs on the basis of stress existing at
the lower edge of each course.
Application of other methods of shell design,
such as the variable point method, is permitted
and explained in API 650.

LiveThe minimum roof live load shall be 15

psf to account for future maintenance loads and
possible accidental vacuum.
SeismicBecause of their flexibility and
ductility, flat-bottomed cylindrical steel tanks
have had an excellent safety record in
earthquakes. Steel has the ability to absorb large
amounts of energy without fracture.
Prior to the Alaskan earthquake of 1964, flat
bottom ground storage tanks had an excellent
record of surviving western hemisphere
earthquakes with essentially no effects other
than broken pipe connections or minor buckles.
In the Alaskan quake, the horizontal oscillations
of the tank contents caused vertical shell
stresses of sufficient magnitude to permanently
deform the shell in a peripheral accordion-like
buckle near the bottom (exaggerated elephant
foot buckling). But again, the properties of steel
were
sufficient
to
accommodate
this
deformation without fracture of the shell
plates.4
AWWA D100 and API 650 contain
recommendations for the seismic design of
tanks. The seismic design requirements in both
API and AWWA were recently updated to
follow the ASCE 7-05 and IBC 2006
requirements.

Loads To Be Considered
As outlined in the preceding section, the
thickness of the shell is determined by the
weight of the product stored. However, there
are other loads or forces which a tank may have
to resist and which are common to both oil and
water tanks.
WindHistorically for tank design, wind
pressure has been assumed to be 30 psf on
vertical plane surfaces which, when applying
shape factors of 0.6 and 0.5 respectively,
becomes 18 psf on the projected area of a
cylindrical surface, and 15 psf on the projected
area of a cone or surface of double curvature as
in the case of tank roofs. These loads are
considered to be the pressure caused by a wind
velocity of 100 MPH. For higher or lower wind
velocity, these loads are increased or decreased
in proportion to the square of the velocity ratio,
(V / 100) 2 , where V is expected wind velocity
expressed in miles per hour.
In recent years, the ASCE 7 has been the basis
of loads for the U.S. buildings codes. This
document is more advanced and includes
effects of escalation of wind speed with height,
increased wind speed along coastal regions and
other factors not considered in the original
simplified approach of the tank standards. This
newer method was adopted; but, for AWWA
the historical wind pressures were retained as
minimum design pressures.

Negative Pressure(such as partial vacuum)

Most tanks of this nature at some time will be
subject to a negative pressure (partial vacuum)
by design or otherwise. Approximately one-half
oz. per square inch negative pressure is built
into the shell stability formulae in AWWA D100
and API 650. AWWA D100 tanks are not
usually designed for negative pressure but
negative pressure due to the evacuation of
water
is
considered
in
the
venting
requirements. Occasionally API 650 tanks are
specified to resist a certain negative pressure,
usually expressed in inches of water column. To
meet these requirements the shell and roof must
be designed to resist the specified negative
pressure. See API 650 Appendix V for the
current design methods applicable to flat
bottom tanks for external pressure.

SnowSnow load is assumed to be 25 psf on

the horizontal projected area of the roof that has
a slope of 30 degrees or less with the horizontal
plane. If the roof slope is greater than 30
degrees, then the snow load may be zero. Snow
loads reduction may also be made in regions
where the lowest one day mean temperature is
5F or warmer. Fixed steel roofs on tanks are not
usually designed for nonsymmetrical loads but
if such load conditions are anticipated, these
should be considered by the designer.
Aluminum geodesic dome roofs may have an
12

should remain flexible to facilitate plate seams,

nozzles and other interferences. For example,
for a shell plate that is 10 feet long, it would be
advantageous to use three anchors per plate
and space the anchors at approximately 10.5
feet.
Obviously the anchor bolt circle must be
larger than the tank diameter, but care should
be taken so interference will not occur between
the anchor bolts and foundation reinforcing.
Volume 2 Part V provides design rules for
anchor bolt chairs.

Top and Intermediate Wind Girders

Open top tanks require stiffening rings at or
near the top of the shell to resist distortion or
buckling due to wind. These stiffening rings are
referred to as wind girders. In addition, some
tank shells of open top and fixed roof tanks
require intermediate wind girders to prevent
buckling due to wind. API 650 and AWWA
D100 provide differing design requirements for
intermediate wind girders and are explained in
the examples of Appendix A. The formula for
maximum height of unstiffened shell is based
on the MODIFIED MODEL BASIN FORMULA
for the critical uniform external pressure on
thin-wall tubes free from end loadings.

Corrosion Allowance
As a minimum for all tanks, bottom plates
should be l/4" in thickness and lap welded top
side only. If corrosion allowance is required for
bottom plates, the as-furnished thickness
(including corrosion allowance) should be
specified. The thickness of annular ring or
sketch plates beneath the tank shell may be
required to be thicker than the remainder of the
bottom plates and any corrosion allowance
should be specified as applicable to the
calculated thickness or the minimum thickness.
API 650 and AWWA D100 specify minimum
shell plate thicknesses based on tank diameter
for construction purposes. If corrosion
allowance is necessary, it should be added in
accordance with the respective standard. A
required minimum above those stated in the
standards may also be specified, but it should
be made clear if this minimum includes the
necessary corrosion allowance.
As a minimum for all tanks, roof plates
should be 3/16" in thickness and lap welded
top side only. If corrosion allowance is
necessary, it should be added in accordance
with the respective standard. A required
minimum greater than 3/16" in thickness may
be specified, but it should be made clear if this
minimum includes the necessary corrosion
allowance.
If corrosion allowance is necessary for roofsupporting structural members, it should be
added in accordance with the respective
standard. If a corrosion allowance requirement
different from the standards is necessary, it
should be made clear what parts of the
structure require the additional thickness
(flange or web, one side or both sides) and/or
the minimum thickness necessary.

Anchor Bolts
The normal proportions of petroleum tanks
are such (diameter greater than height) that
anchor bolts are rarely needed. It is quite
common, however, for the height of water tanks
to be greater than the diameter. There is a limit
beyond which there is danger that any empty
tank will overturn when subjected to the
maximum wind velocity. As a good rule of
thumb, if C in the following formula exceeds
0.66, anchor bolts are required:

2M
where
(3-3)
dw
overturning moment due to wind, ft. lb.
diameter of shell in feet
weight of shell and portion of roof
supported by shell, lb.
C=

w =

4M W
Design tension load per bolt = ND N

(3-4)

Where M and w are as above and

N = number of anchor bolts
D
=
diameter of anchor bolt circle, feet
The diameter of the anchor bolts shall be
determined by an allowable stress of 15000 psi
on the net section at the root of the thread with
appropriate stress increase for wind or
earthquake loading.
Because of proportionately large loss of
section by corrosion on small areas, it is
recommended that no anchor bolt be less than
1.25" in diameter.
Maximum desirable spacing of anchors as
suggested by API 650 and AWWA D100 is 10'0.This spacing is a matter of judgment and
13

API Standard 650

construction difficulties in order to perform the

work in a safe manner.

General
The following information is based on API
650. Anyone dealing with tanks should obtain a
copy of the complete standard.

Top Angle
Except for open-top tanks and the special
requirements applying to self-supporting roofs,
tank shells shall be provided with top angles of
not less than the following sizes:

Shell Design
API requires that all joints between shell
plates shall be butt-welded. Lap joints are
permitted only in the roof and bottom and in
attaching the top angle to the shell.
API 650 offers optional shell design
procedures. The refined design procedures
permit higher design stresses in return for a
more refined engineering design, more rigorous
inspection, and the use of shell plate steels
which demonstrate improved toughness.
The probability of detrimental notches is
higher at discontinuities such as shell
penetrations. The basic requirements pertaining
to welding, stress relief, and inspection relative
to the design procedures are important. Tank
shells designed in accordance with refined
procedures will be thinner than the simplified
procedure, and thus will have reduced
resistance to buckling under wind load when
empty. The shell may or may not need to be
stiffened, but must be checked. This is
discussed in the section on wind girders.

Tank Diameter
35 feet and less
over 35 to 60 ft. incl.
over 60 feet

Roofs
The selection of roof type depends on many
factors. In the oil industry, many roofs are
selected to minimize evaporation losses.
Inasmuch as the ordinary oil tank is designed to
withstand pressures only slightly above
atmospheric, it must be vented against pressure
and vacuum. The space above the liquid is
filled with an air-vapor mixture. When a nearly
empty tank is filled with liquid, this air-vapor
mixture expands in the heat of the day and the
resulting increase in pressure causes venting.
During the cool of the night, the remaining airvapor mixture contracts, more fresh air is
drawn in, more vapor evaporates to saturate the
air-vapor mixture, and the next day the cycle is
repeated. Either the loss of valuable "light ends"
to the atmosphere from filling, or the breathing
loss due to the expansion contraction cycle, is a
very substantial loss and has led to the
development of many roof types designed to
minimize such losses.
The floating roof is probably the most
popular of all conservation devices and is
included as Appendices to API Standard 650.
The principle of the floating roof is simple. It
floats on the liquid surface; therefore, there is
no vapor either to be expelled on filling or to
expand or contract from day to night.
Inasmuch as all such conservation devices are
represented by proprietary and often patented
designs, they are beyond the scope of this
discussion, which will be limited to the fixed
roofs covered by API Standards.
API 650 provides rules for the design of
several types of fixed roofs.
The most common fixed roof is the
column-supported cone roof, except for
relatively small diameters where the added cost
of a self-supporting roof is more than offset by
saving the cost of structural framing. The
dividing line cannot be accurately defined

Bottoms
Tank bottoms are usually lap welded plates
having a minimum nominal thickness of 1/4".
After trimming, bottom plates shall extend a
minimum of 1 inch beyond the outside edge of
the weld attaching the bottom to the shell
plates. The attachment weld shall be a
continuous fillet inside and out as shown in the
following table of sizes:
Maximum t of
Shell Plate
Inches
3/16
over 3/16 to 3/4
over 3/4 to 1-1/4
over 1-1/4 to 1-3/4

Minimum Size of Top Angle

2 x 2 x 3/16
2 x 2 x 1/4
3 x 3 x 3/8

Minimum Size of Fillet

Weld*
Inches
3/16
1/4
5/16
3/8

* Maximum size Fillet 1/2"

Butt-welded bottoms are permissible, but

because of cost, are seldom used except in
special services. Butt-welded bottoms are
usually welded from the top side only using
backing strips attached to the underside.
Welding from both sides presents significant
14

because different practices and available

equipment may affect the decision in any given
case. If economy is the only consideration, the
purchaser would be well advised to specify the
size of tank and let the manufacturer decide
whether or not to use a self-supporting roof.
A self-supporting roof is sometimes desirable
for special service conditions such as an internal
floating roof, or where cleanliness and ease of
cleaning are especially important.
All roofs and supporting structures shall be
designed to support dead load plus a live load
of not less than 15 psf.
Roof plates shall have a minimum nominal
thickness of 3/16 inch. Structural members shall
have a minimum thickness of 0.17 inch.
Roof plates shall be attached to the top angle
with a continuous fillet weld on the top side
only:
1. If the continuous fillet weld between the
roof plates and the top angle does not
exceed 3/16 inch and the slope of the roof
at the top-angle attachment does not
exceed 2 inches in 12 inches, and when the
cross-sectional area of the roof-to-shell
junction does not exceed
A=

W
201,000 tan

connection for the device and the

drawings should reflect the need for such
a device to be supplied by the customer.
The top angle may be smaller than
previously noted when a frangible joint is
specified.
3. Tanks less than 50 ft. diameter may not be
considered to have frangible roof joints
even when the provisions of item 1 are
satisfied.
Supported Cone Roofs Supported cone
roofs are usually lap welded from the top side
only with continuous full fillet welds. Plates
shall not be attached to supporting members,
and shall be attached to the top angle by a
continuous 3/16" fillet weld or smaller on the
top side if specified by purchaser.
The usual slope of supported cone roofs is
3/4" in 12". Increased slopes should be used
with caution. The columns transmit their loads
directly to the supporting soil through bases
resting on but not attached to the bottom plates.
Some differential settlement can be expected. A
relatively flat roof will follow such variations
without difficulty. As pitch increases, a cone
acquires stiffness, and instead of smoothly
following a revised contour, unsightly local
buckles may develop. In general, slopes
exceeding 1-1/2" in 12" may be undesirable.
Rafters in direct contact with the roof plates
may be considered to receive adequate lateral
support from friction, but this does not apply to
truss chord members, rafters deeper than 15", or
roof slopes greater than 2" in 12".
Rafters are spaced so that, in the outer ring,
their centers are not more than 6.28 feet apart at
the shell. Spacing on inner rings does not
exceed 5.5 feet. All parts of the supporting
structure shall be so proportioned that the sum
of the maximum calculated stresses shall not
exceed the allowable stresses as stated in the
appropriate section of API 650.

(3-5)

where W =

total weight of the shell and

roof framing supported by
the shell in pounds
= angle between the roof and a
horizontal plane at the roofto-shell juncture in degrees,
the joint may be considered to be frangible
and, in case of excessive internal pressure,
will fail before failure occurs in the tank
shell joints or the shell-to-bottom joint.
Failure of the roof-to-shell joint is usually
initiated by buckling of the top angle and
followed by tearing of the 3/16 inch
continuous weld at the periphery of the
roof plates.
2. Where the weld size exceeds 3/16 inch, or
where the slope of the roof at the topangle attachment is greater than 2 inches
in 12 inches, or when the cross-sectional
area of the roof-to-shell junction exceeds
the value calculated per Equation 3-5, or
where fillet welding from both sides is
specified, emergency venting devices in
accordance with API Standard 2000 shall
be provided by the purchaser. The
manufacturer shall provide a suitable tank

Self-Supporting Roofs Self-supporting cone,

dome or umbrella roofs shall conform to the
appropriate requirements of API 650 unless
otherwise specified by the purchaser.

15

such storage are referred

Appendices R and Q.

Accessories
API 650 contains specific designs for
approved accessories which include all
dimensions, thicknesses, and welding details.
For all cases, OSHA requirements must be
satisfied.
No details are shown, but specifications are
included for stairways, walkways and
platforms. All such structures are designed to
support a moving concentrated load of 1000 Ibs.
and the handrail shall be capable of
withstanding a load of 200 lbs. applied in any
direction at any point on the top rail.
Normally all pipe connections enter the tank
through the lower part of the shell. Historically
tank diameters and design stress levels have
been such that the elastic movement of the tank
shell under load has not been difficult to
accommodate.
With the trend to larger tanks and higher
stresses, the elastic movement of the shell can
become an important factor.
Steel being an elastic material, the tank shell
increases in diameter when subjected to internal
pressure. The flat bottom acts as a diaphragm
and restrains outward movement of the shell.
As a result, the shell is greater in diameter
several feet above the bottom than at the
bottom.
Openings near the bottom of the tank shell
will tend to rotate with vertical bending of the
shell under hydrostatic loading. Shell openings
in this area, having attached piping or other
external loads, should be reinforced not only for
the static conditions but also for any loads
imposed on the shell connections by the
restraint of the attached piping to the shell
rotations. Preferably the external loads should
be minimized or the shell connections relocated
outside the rotation area.

to

API

620

Molasses Tanks Molasses presents no

unusual problems other than the fact that its
specific gravity is about 1.48, and the shell
design must, of course, take this into account. It
is quite common to require such tanks to be
built in accordance with API 650.
It must be remembered that the API
Appendix A design stress of 21,000 psi at 85%
joint factor is predicated on the tank being full
of water during test, and that the actual stress in
petroleum service is usually considerably less.
Because molasses is heavier than water, the full
design stress is present in service. Thus if the
designer is depending on the long and
successful record of tanks designed in
accordance with API 650 Appendix A design, it
would be more consistent with the true
situation to use a somewhat lower design stress.
On the other hand, on tanks built to the basic
design of API 650 this difference between usual
petroleum service stress and design stress does
not exist. However, the addition of a corrosion
allowance is required when warranted by
service conditions.
Acid and Caustic Tanks To attempt a
comprehensive discussion of the subject of
storing acids and caustic solutions is beyond the
scope of this work. While stainless steel or other
high alloy materials are often required, some
acids and caustic solutions can be stored
successfully in carbon steel tanks, and the
following discussion will be limited to such
application.
In the absence of personal experience,
information concerning the corrosive properties
of many common solutions can be found in
chemistry and chemical engineers' handbooks
or in the publications of the National
Association of Corrosion Engineers. However,
it should be noted that very small differences in
content (such as slight impurities) or conditions
can influence the corrosive effect of many
chemicals.
As an example, concentrated sulfuric acid
does not attack carbon steel, whereas dilute
sulfuric acid is extremely corrosive. Thus
concentrated sulfuric acid can often be safely
stored in carbon steel tanks provided proper
precautions are taken to cope with dilute acid
that may form in the upper portions of the tank

Tanks Other Than for Oil or Water

There are many applications for steel tanks
other than the storage of oil or water. Since
most such applications are industrial in nature
for which no industry standard has been
developed, it is quite common to use API
Standard 650 as a basis for design and
construction. This is a logical approach
provided that problems peculiar to the contents
stored are taken into account.
Tanks designed to store liquefied gases at or
near atmospheric pressure are beyond the scope
of this document. However, those interested in
16

when acid fumes and water condensation meet

in the vapor space.
Thus one fundamental requirement for an
acid tank is that the interior of the tank be
smooth without crevices or pockets where
dilute acid condensation can collect. Selfsupporting roofs are good practice. If the design
of the roof or size of tank requires structural
stiffeners, it is desirable that they be placed on
the outside. If the roof is lap welded, it should
be welded underneath as well as the top. The
connection of the roof to the shell should
eliminate any pocket which might exist at the
top of a standard API tank.
When using Appendix A design basis of API
650, a lower design stress should be considered
for the same reasons as given under "Molasses
Tanks." The tank user should specify the
amount of corrosion allowance, if any is
required, for his particular purpose.
In the case of carbon steel tanks storing
caustic solutions, both the concentration and
temperature are important. Carbon steel tanks
should not be used if the combination of
concentration and temperature exceeds the
following values and may in some cases be
unsatisfactory below these limits:
50% and 120F
25% and 150F
5% and 200F
It is most important to make sure that the
specified design conditions are not exceeded in
service. Automatic temperature controls are
recommended.
In addition to ordinary corrosion, the
principal problem in caustic tanks is one
referred to as "caustic embrittlement" or "stress
corrosion cracking." In the presence of high
local stresses this type of corrosion can rapidly
result in cracks and leaks. Local stress
concentrations approaching the yield point can
exist at shell penetrations, in the vicinity of
welds and at other details. In caustic service
these are the points where stress corrosion
cracking can occur.
Thus, in the case of caustic storage tanks, all
fittings penetrating the shell or bottom, or any
permanent attachments welded to the interior
surface thereof, should be installed in a plate in
the shop and the entire assembly thermally
stress relieved. Essentially, this leaves only
main seam welding to be performed in the field.
Self-supporting roofs without structural
members immersed in the tank contents are
advisable. It is not necessary, however, to

eliminate crevices and pockets as is

recommended for acid tanks. For caustic tanks,
a standard API roof is acceptable.
Certain additional precautions in welding
should be taken for both acid and caustic tanks.
Lap welds in the bottom and the inside bottomto-shell fillet should be made in at least two
passes. Since the bottom-to-shell weld usually
consists of a fillet inside and out, it is advisable
to provide a water stop (complete penetration)
at each vertical shell joint so that if a leak does
occur in the inside fillet, channeling will be
limited to one plate length.
All other shell joints should be designed for
complete penetration and fusion. The inside
passes should be made first. The later welding
of outside passes will partially heat treat and
reduce residual stresses in the inside weld. If
anticipated corrosion indicates a bottom plate
thickness greater than 3/8", the bottom should
be butt welded and the same sequence
followed; i.e. weld the inside passes first.
Inasmuch as all welds create locally high
residual stresses, all brackets, welding lugs, etc.
should be kept to a minimum, be located on the
outside, and attached with small-diameter
electrodes to limit the heat input and
consequently the effect on the inside surface.
When the corrosive attack is considered
sufficiently severe to admit the possibility of
local penetration, but not severe enough to
warrant the expense of high alloy or clad steel
plates, the tank is sometimes supported on a
structural grillage to permit inspection from the
underside.

AWWA Standard D100

General
The following information is based on the
AWWA Standard D100. Anyone dealing with
tanks should obtain a copy of the complete
standard.
With the exception of shells, roofs and
accessories, many of the comments made in
connection with API tanks also apply to
AWWA tanks and will not be repeated here in
detail.
Bottoms may be either lap or butt-welded
with a minimum thickness of 1/4 inch.
AWWA does not specify top angle sizes, but
the rules of API represent good practice.

17

Shell Design
AWWA D100 offers two different design
bases the standard or basic design and the
alternate design basis as outlined in Section 14.
The alternate design basis permits higher
design stresses in return for a more refined
engineering design, more rigorous inspection,
and the use of shell plate steels with improved
toughness.
AWWA D100 Section 14 includes steels of
significantly higher strength levels and
correspondingly higher design stress levels.
This introduces new design problems. For
example, for A517 steels, the permissible design
stress of 38,333 psi will result in reaching the
minimum required nominal thickness several
courses below the tank top. It would be
uneconomical to continue the relatively
expensive steel into courses of plates not
determined by stress. The obvious answer is to
use less expensive steels in the upper rings. To
govern this transition, Section 14 adds the
following requirements:

the high water level will extend up into the roof

itself. The resultant upward pressure on the
roof is resisted by the combination of the roof
dead load and the weld joint between the roof
and shell.
AWWA requires that for all roof plate
surfaces in contact with water, the minimum
metal thickness shall be 1/4". Roof plate
surfaces not in contact with water may be
3/16".
As applied to rolled shapes for roof framing,
the foregoing minimum thicknesses shall apply
to the mean thickness of the flanges regardless
of web thickness.
Roof supports or stiffeners, if used, shall be in
accordance with current specifications of the
American Institute of Steel Construction
covering structural steel for buildings, with the
following exceptions:
1. Roof plates are considered to provide the
necessary lateral support by friction
between roof plates and rafters to
eliminate reduction in the basic allowable
compressive stress, except where trusses
and open web joists are used for rafters, or
rafters having nominal depth greater than
15 in. or rafters having a slope greater
than 2 in 12.
2. The roof, rafter and purlin depth may be

In the interest of economy, upper courses may be of weaker

material than used in the lower courses of shell plates, but in no
instance shall the calculated stress at the bottom of any course be
greater than permitted for the material in that course. A plate
course may be thicker than the course below it provided the extra
thickness is not used in any stress or wind stability calculation.

Compliance with this requirement will

probably result in the course or courses
immediately below the transition point being
somewhat heavier than required by stress.
Using a steel of intermediate strength level as a
transition between A517 steel and carbon steel
may help the situation. In any event, the use of
two or more steels will result in plates of the
same thickness made of different steels. Careful
attention to plain marking for positive
identification
becomes
very
important.
Consideration might be given to varying plate
widths for different materials of the same
thickness to aid in identification in the event
that markings are lost.

less than

fb
600,000

times the span length in inches where f b is

the maximum bending stress in psi,
providing slope of the roof is to 12 or
greater.
3. The maximum slenderness ratio (L / r ) for
roof support columns shall be 175.
4. Roof support columns shall be designed
as secondary members.
5. Roof trusses, if any, shall be placed above
the maximum water level in climates
where ice may form. Roof trusses are not
recommended due to the high degree of
maintenance required over the life of the
tank roof.
6. Roof rafters and connections shall be
placed above maximum water level.

Roofs
Whereas oil tanks are strictly utilitarian, a
pleasing appearance is often an important
consideration in the case of water tanks. Since
the roof line has an important effect on
appearance, this striving for beauty has led to a
wide variety of roof designs.
Often a self-supporting roof, such as an
ellipsoid, will extend a considerable distance
above the cylindrical portion of the shell, and
18

Accessories
AWWA does not provide detailed designs of
tank fittings and accessories, but specifies the
following:
1. Compliance with OSHA and other
regulations.
2. Two manholes shall be provided in the
first ring of the tank shell. Manholes shall
be either a 24" diameter or at least 18" x
22" when elliptical manholes are used.
30-inch diameter manholes are often
recommended for safe recovery of
personnel.
3. The purchaser shall specify pipe
connections, sizes, and locations. Due to
freezing hazard these connections are
normally made through the tank bottom
and as near to the shell as practical. A
concrete valve box may be provided to
permit access to piping. This valve box
must be designed as a part of the
ringwall.
4. If a removable silt stop is required, it shall
be at least 4" high. If not required, then
the connecting pipe shall extend at least
4" above the tank bottom.
5. The purchaser shall specify the overflow
size and type. If an overflow to ground is
required, it should be brought down the
outside of the tank and discharged onto a
splash block or other appropriate
drainage structure with an air break.
Inside overflows are not recommended.
They are easily damaged by ice, and a
failure in the overflow will empty the
tank to the level of the break.
6. An outside vertical ladder shall begin 8
feet (or as specified) above the tank
bottom and afford access to the roof.
Need for access to AWWA tanks is
infrequent and a conscious effort is made
to render access difficult for unauthorized
personnel.
7. The contractor shall provide access to the
roof hatches and vents. The access must
be reached from the outside tank ladder
and fulfill the AWWA D100 requirements
consistent with the roof slope or as
specified by the purchaser.
8. A roof door or hatch whose least
dimensions are 24" x 15", with a curb 4"
high, provided with a hinged door and
clasp for locking shall be placed near the
outside tank ladder. A second opening of

9.

at least 20" in diameter and with a 4" neck

must be provided near the center of the
tank. Additional openings may be
required for ventilation during painting.
Adequate venting shall be provided to
accommodate the maximum filling and
emptying rates. A frost resistance
pressurevacuum
relief
device
is
required. These rates should be specified
by the purchaser.

Venting for outflow (partial vacuum

condition) is based upon the unrestricted vent
area and the pressure differential that can safely
be allowed between the outside and inside of
the tank. This differential is established by
quantifying the strength of the roof and shell
above
and
beyond
other
structural
requirements; for example, the margin of extra
strength of the shell against buckling with
respect to the design wind load.
Venting for inflow (pressure condition) is
again based upon the restricted vent area and
the pressure differential that can safely be
allowed before lifting the roof plates. For
example, if 3/16" roof plates are used, the
pressure differential would be 7.65 psf, 0.053
psi, or 1.47 inches water column. If the
differential is limited to the weight of the roof,
the shell/roof juncture does not become
involved. The overstress in the shell would be
minimal.
The equation for outflow vent capacity is:

P 0.286
Q = 0.5 A 110 T a
1

Pi

1/ 2

(3-6)

where Q =

vent capacity in cubic feet

per second
A = minimum clear vent open
area in square feet
T = air temperature in degrees
Rankine
Pa = atmospheric pressure in psia
Pi = pressure in tank during
withdrawal in psia
The equation for inflow vent capacity is:
1/ 2

0.286

6 Pi
Q = 0.5 A6.25 10
1
Pa

19

(3-7)

APPENDIX A
The following design example covers the
AWWA D100 tank.
Calculate shell thickness using the basic
equation:

Design Example
For typical examples of tank design consider
two tanks 150 feet in diameter by 40 feet
nominal height with flat cone supported roofs.
Consider one tank per AWWA D100 and the
other tank per API 650. See Figure 3A-1 for tank
dimensions.
These examples are for illustration only and
are not to be used for an actual design or
construction. Design of similar tanks should be
accomplished by competent people experienced
in the design of like structures and the use of
applicable standards.
For the AWWA tank consider Section 14,
shell design and a site with mapped seismic
ground motion values per ASCE 7-05 and
AWWA D100-05 values of Ss = 0.5 and S1 = 0.15.
Assume the Seismic Use Group is III. The Site
Classification is C and TL = 8 sec.
For the API 650 tank consider the standard,
shell design by the variable point method, 1/16
inch corrosion allowance on the shell only. The
seismic design procedures for API 650, API 620
and AWWA D100 are similar and are not
repeated.
Consider design metal temperature (DMT) of
20F, standard 100 mph wind loads, standard
25 PSF roof loads, a maximum liquid content
height of 39'-6, and a design specific gravity of
1.0 for both tanks.
The economics of plate selection with respect
to width and grade and structural selection will
differ
with
location
and
construction
capabilities. Factors to consider are plate width
and grade availability in a particular locality
and structural rolling schedules. Also the
availability of plate and structural stock in a
particular locality will sometimes influence the
selection of material. Further discussion of
material selection will be beyond the scope of
this document.

t=

2.6h p DG
sE

= 0.1547"

(3-8)

All nomenclature in the above and following

equations is defined in the AWWA D100
standard. Notice that hp in the above equation is
the full liquid height above the design point
rather than h - 1 as used in API 650. The
calculation for ring five (top ring) is:

t5 =

2.6 7.66 150 1.0

= 0.1547"
19,330 1.0

The thicknesses for the remaining rings

calculate:
hp =15.63
hp= 23.58
hp= 31.54
hp= 39.50

S=19,330 psi
S=23,330 psi
S=23,330 psi
S=23,330 psi

t4=0.3152
t3=0.3942
t2=0.5273
t1=0.6603

using A36 steel for rings 4 and 5 and A573

GR70 for rings 1, 2 and 3. Ring 5 will be
increased to 0.3125 because of minimum
thickness requirements in AWWA D100.
Shell stability is calculated using the basic
equation.
H=

10.625 106 t

(3-9)

Pw (D / t )1.5

The calculation for ring five (top ring) is:

H5 =

10.625 10 6 0.3125
18 (150 / 0.3125 )1.5

= 17.54' > 7.96'

For each ring the h calculated is compared

to the actual height of shell above the design
point. When h calculates less than the height
of shell above, the shell is unstable. This may be
corrected by thickening the shell or adding a
stiffening ring. For this example we will
consider only thickening the shell.
h4 = 17.73 > 15.92
h3 = 21.76 < 23.87
Recalculate the thickness of ring 3 by using a
lower strength steel (A36).
t3 = 0.4758
20

Using an inner support radius of 2.38 ft,

which is dependent upon the method of
supporting the inner rafters, the maximum
design length of the inner rafters is 39.33 ft, as
indicated in Figure 3A-2. The maximum design
moment calculates to be 27,580 ft-Ibs. Using an
AISC allowable stress of 0.66 Fy, a section
modulus of 13.93 in3 is required. A W12 14
section with a section modulus of 14.9 in3 is
chosen. See Figure 3A-3 for a typical rafter
loading.
The maximum design length for the outer
rafters is 35.33 ft, as indicated in Figure 3A-2.
The maximum design moment calculates to be
27,890 ft-lbs. A section modulus of 14.09 in3 is
required and again we will choose a W12 14
section.
The rafter reactions are placed on the girder
at the locations as determined by the roof
framing layout. The outer rafter reactions are
3480 Ibs.; the inner rafter reactions are 2840 lbs.;
and the girder design length is 29.07 ft. The
maximum design moment calculates to be
150,440 ft-lbs. Again using AISC allowable
stresses, a section modulus of 75.98 in3 is
required. AW18 46 sections with a section
modulus of 78.80 in3 is chosen. See Figure 3A-4
for a typical girder loading.
For the center column a design load of 74,900
lbs. is calculated from the accumulated
reactions of the inner rafters. Using AISC
design procedures an allowable compressive
stress is determined based upon the
unsupported column length of 486.5 inches and
a calculated slenderness ratio of 131. A 10"
diameter schedule 20 pipe will meet the design
criteria. See Figure 3A-5 for typical center
column detail.
For the outer columns we have chosen an 8
diameter schedule 20 pipe based upon a design
load of 41,400 lbs, an unsupported column
length of 470.6 inches, and a slenderness ratio of
159; using the same design criteria as the center
column. See Figure 3A-6 for a typical outer
column detail.
Seismic design requirements are given in
Section 13 of AWWA D100-05 and follow ASCE
7-05. The weights of the tank and liquid are
computed to be:
Weight of the product = 43,556,000 lbs
Weight of the tanks shell = 340,000 lbs
Weight of the roof/framing = 354,000 lbs
Weight of roof acting on shell = 205,000 lbs

Recalculate:
h3 = 26.37 > 23.87
The shell is now stable above ring 3;
continuing;
h2 = 34.10 > 31.83
h1 = 45.67 > 39.79
The entire shell is now stable for a design
wind velocity of 100 mph. See Table 3A-1 for
shell thicknesses before and after minimum
thickness and wind stability adjustments.
For 90 mph wind load, minimum design
loads are 18 psf on projected areas of cylindrical
surfaces (shell) and 15 psf on projected areas of
double curved surfaces (roof). Based upon the
tank geometry and the design loading, the wind
shear is calculated:
Shell = 150 40.04 18 =
108,113 lbs
Roof = 150 4.69 0.5 15 = 5,273 lbs
Total =
113,386 lbs
The minimum required coefficient of friction
against sliding is:
Wind Shear = 113,386 = 0.154
Tank Weight
734,250

(3-10)

This coefficient is well below established

values which range as high as 0.4 to 0.5.
The wind moment at the base of the shell is
calculated:
Shell = 108,113 20.02 = 2,164,421 ft-lbs.
Roof = 5,273 41.60 = 219,357
Total =
2,383,778 ft-lbs.
The ratio, C = 2M/dw, calculates to be 0.076 <
0.666; therefore, no anchors are required to
resist overturning due to wind.
Roof framing concepts, layout and detail
vary among tank designers and suppliers.
Rafter spacing is dependent upon roof loading
and plate thickness. For reasons of plate
strength and construction a maximum rafter
spacing of approximately 7.00 feet is desirable.
For this example consider nine girders and
outer columns, 36 inner rafters and 72 outer
rafters (see Figure 3A-2). The outer columns
will be located on a 42'-6" radius. The rafter
spacing is 6.54 feet at the shell and 6.92 feet at
the girder. Consider 25 psf snow load and 7.65
psf (3/16" roof plate) dead load.
21

Sc = 416 psi < 4570 psi

Weight of tank bottom = 181,000 lbs

Center of gravity of roof above shell = 3 ft
Center of gravity of shell above fdn = 16.7 ft
H/D ratio = 0.263

OK

The additional hydrodynamic hoop stresses

are calculated by Section 13 (Equations 13-43
through 13-46). These hydrodynamic hoop
stresses are added directly to the hydrostatic
hoop stresses.

Given Ss = 0.5 and S1 = 0.15. From Table 26

and 27 of AWWA D100 and Site class C, Fa and
Fv can be determined as Fa = 1.2 and Fv = 1.65.
Substituting into the equations using k=1.5, U =
2/3, I=1.5 Ri= 2.5 for self-anchored and Rc = 1,5,
the values for Ai and Ac are computed as:

Hydrodynamic Hoop Stresses

Y

Ai = 0.172
Ac = 0.022 and
Av = 0.056
The sloshing period, Tc is 8.18 sec.
Using Section 13.5, the impulsive weight, Wi,
and convective weight, Wc, are calculated:

Y/D

Eqn
13-43

Eqn
13-44

7.5 .050

790

692

15.5 .103

1450

23.5 .157

4
5

Eqn
Ni
13-45
5388

Eqn
13-46
Nc

790

432

1378

5388 1450

384

1922

2009

5388 1922

351

31.5 .210

2207

2586

5388 2207

332

39.5 .263

2303

3108

5388 2303

326

The addition of the stresses is usually done on

a force/per unit length of circumference basis.
In the longitudinal or vertical membrane (phi)
direction, the forces from dead load, snow load,
live load, and overturning are added together in
the appropriate load combinations and
compared to the applicable allowable stress.
Similarly, in the circumferential (hoop or theta)
membrane direction the hydrostatic and
hydrodynamic forces are added. The results of
this calculation are summarized in the table on
page 23.
Finally, the freeboard must be evaluated per
Section 13. Since this was identified as a SUG
III tank, and the sloshing period, Tc > 4 sec. Af =
0.03g, and the calculated wave height is 2.2 ft.
The freeboard provided is 0 ft (0.5 ft from roof
plate was there for rafter ends to project into the
tank). Thus, the design must be modified.
Either the liquid level must be reduced or the
shell height increased by 2.2 ft to provide the
required freeboard.

Wi = 13,208,000 lbs
Wc = 28,423,000 lbs
Similarly, substituting into the equations for
the moment arms of the lateral forces, Xi, Xc can
be computed:
Xi = 14.81 ft
Xc = 21.16 ft
Substituting into Equation 13-23 of AWWA
D100, the ringwall moment is
Ms = 37,354,000 ft lbs
Using this value of Ms in Equation 13-36,
calculate the J ratio.
Less than 0.785, so there is no net uplift for
the design overturning moment and the tank is
self-anchored
if
the
maximum
shell
compression calculated by Equation 13-39 is
met. Substituting into Equation 13-39,

22

Summary of Shell Stresses (AWWA)

Ring No
Material
Design allowable tensile
Ht of Ring
Thickness of Ring
Y
RoofDL, including equipment
Shell Weight
Cummulative DL, Self Weight
Nphi DL(metal)
Nphi , LL
Nphi HEQ, impulsive
Nphi HEQ, convective
Nphi HEQ (direct sum)
Nphi HEQ (srss)
Nphi VEQ
Ntheta, hydrostatic
Ntheta HEQ, Nimpulsive
Ntheta HEQ, Nconvective
Ntheta, HEQ (direct sum)
Ntheta VEQ, Ntheta x EQ vert factor
Ntheta EQ (srss)
Ntheta total
Check Load Combinations
DL+LL
Nphi
Sphi
Sphi allowable

Roof
psi
ft
inches
ft
lbs/in
kips
lbs/in
lbs/in
lbs/in
lbs/in
lbs/in
lbs/in
lbs/in
lbs/in
lbs/in
lbs/in
lbs/in
lbs/in
lbs/in
lbs/in
lbs/in

Top
A36M
19330
8
0.3125
7.5
36.3
48.1
44.8
44.8
0.0
3.4
2.6
5.9
4.2
2.5
2925
790
432
1221
165
915
3840

36.25
36.3
36.3
0.0
0.5
0.0
0.5
0.5
2.0

Top
lbs/in
psi
psi

Ntheta
Stheta
Stheta allowable

lbs/in
psi
psi

Nphi
Sphi
Sphi allowable

lbs/in
psi
psi

Ntheta
Stheta
Stheta allowable

lbs/in
psi
psi

36.3
116
611
OK

44.8
143
611

2
3
4
5
A36M A573Gr70A573Gr70A573Gr70
19330
23330
23330
23330
8
8
8
8
0.3152
0.3942
0.5273 0.6604
15.5
23.5
31.5
39.5
36.3
36.3
36.3
36.3
48.5
60.6
81.1
101.6
53.3
64.0
78.4
96.4
53.3
64.0
78.4
96.4
0.0
0.0
0.0
0.0
12.9
34.5
73.5
135.4
11.0
25.3
45.5
71.6
23.9
59.9
119.1
207.0
17.0
42.8
86.5
153.2
3.0
3.6
4.4
5.4
6045
9165
12285
15405
1450
1922
2207
2303
384
351
332
326
1834
2274
2539
2629
340
516
691
867
1538
2021
2336
2482
7583
11186
14621
17887
2
3
4
5

OK
2925
9360
19330
OK

53.3
169
617
OK
6045
19178
19330
OK

64.0
162
774
OK
9165
23250
23330
OK

78.4
149
1043
OK
12285
23298
23330
OK

96.4
146
1319
OK
15405
23327
23330
OK

49.0
157
2096
OK
3840
12288
25773
OK

70.4
223
2312
OK
7583
24058
25773
OK

106.9
271
2895
OK
11186
28376
31107
OK

164.9
313
3750
OK
14621
27728
31107
OK

249.6
378
4570
OK
17887
27086
31107
OK

DL+EQ (srss, default AWWA)

36.8
118
815
OK

23

APPENDIX B - TANK FOUNDATIONS

Soils Investigation
The subgrade of a potential tank site must be
capable of supporting the weight of the tank
and contained fluid. A qualified geotechnical
engineer should be retained to conduct the
subsurface exploration and to make specific
recommendations concerning: the type of
foundation required, anticipated settlements,
allowable soil bearing and specific construction
requirements.
The ultimate soil bearing capacity should be
determined using sound principles of
geotechnical engineering. The following
minimum factors of safety should be applied to
the ultimate bearing capacity when determining
the allowable soil bearing:
1. A factor of safety of 3.0 for normal
operating conditions.
2. A factor of safety of 2.25 during
hydrotest.
3. A factor of safety of 2.25 for operating
conditions plus the maximum effect of
wind or seismic forces.
An allowable soil bearing based solely on the
above factors of safety may result in excessive
total settlements. If required, these factors of
safety should be increased in order to limit the
anticipated total settlements to acceptable
values. Factors of safety larger than the above
minimums are also required by certain codes
and standards, such as AWWA D100.

24

3/16 ROOF PLATE

LAP WELDED TOP SIDE ONLY

BOTTOM PLATE LAP WELDED TOP SIDE ONLY

Figure 3A-1 Flat Bottom Tank

Table 3A-1 Shell Plate Thicknesses

a.) CALCULATED SHELL THICKNESSES
FROM STATIC HEAD ONLY
(AWWA DESIGN)
RING #
5
4
3
2
1

THICKNESS
0.1547
0.3152
0.3942
0.5273
0.6603

b.) ADJUSTED FINAL THICKNESSES FOR

STATIC HEAD AND WIND STABILITY
(AWWA DESIGN)
RING#
5
4
3
2
1

MATERIAL
A36
A36
A573GR70
A573GR70
A573GR70

25

THICKNESS
0.3125
0.3152
0.4758
0.5273
0.6603

MATERIAL
A36
A36
A36
A573GR70
A573GR70

Figure 3A-2 Framing Layout -- AWWA

Figure 3A-3 Typical Rafter Loading

Figure 3A-4 Typical Girder Loading

26

Figure 3A-5 Typical Center Column

27

Figure 3A-6 Typical Outer Column

VARIABLE POINT DESIGN: API 650

RING NO. 1
DESIGN: D = 150.000 H = 39.500
G = 1.000
S = 28000.
CA = 0.0625
Td = 2.6*D*(H - 1) * G/S + CA = 0.5362 + CA = 0.5987
T1d = [1.06- (0.463*D/H)* SQRT(H*G/S)] *2.6*D*H*G/S + CA =
T1d = 0.5469 + CA = 0.6094
HYDROTEST: D = 150.000 H = 39.500 G = 1.000
S = 30000.
TT= 2.6*D* (H-1) * G/S = 0.5005
T1T = [1.06-(0.463* D/H) * SQRT(H* G/S)] * 2.6* D* H* G/S= 0.5115
USE: 0.599 IN. A573 70
L/H = SQRT(6.0* D* T)/H3 = 0.5929 <= 2.0 OK
RING NO. 2
DESIGN: D = 150.000

H = 31.542

G = 1.000

S = 28000.

CA = 0.0625

Td = 2.6* D* (H - 1) * G/S + CA = 0.4254 + CA = 0.4879

TX = 2.6* D* (H -X/12) *G/S
TU
0.4254
0.4116
0.4121

TL
K
C
0.5362
1.2606
0.1211
0.5362
1.3028
0.1390
0.5362
1.3013
0.1384
TX = 0.4121 + CA = 0.4746

X2
45.847
52.595
52.369

X1
26.607
28.571
28.506

X3
23.872
23.482
23.495

X
23.872
23.482
23.495

TX
0.4116
0.4121
0.4121

X3
23.062
22.711
22.723

X
23.062
22.711
22.723

TX
0.3851
0.3854
0.3854

DESIGN: PARA. 3.6.4.5

RATIO = 95.500/[SQRT (6 * D * 0.5362) ] = 4.3471
T2 = Tx + (T1 - Tx) * (2.1 - 4.3471/1.25) = 0.2410
T2D = 0.4121 + CA
HYDROTEST:
D = 150.000
H = 31.542 G = 1.000
TT = 2.6 * D * (H - l) * G/S = 0.3970
TX = 2.6 * D * (H - X/l2) * G/S
TU
0.3970
0.3851
0.3854

TL
K
0.5005
1.2606
0.5005
1.2998
0.5005
1.2985
TX = 0.3854

C
0.1211
0.1377
0.1372

X2
45.846
52.127
51.924

S = 30000.

X1
26.202
28.036
27.977

HYDROSTATIC: PARA. 3.6.4.5

RATIO = 95.500/[SQRT (6 * D * 0.5005) ] = 4.4997
T2 = TX + (T1 - TX) * (2.1 4.4997/1.25) = 0.2128
T2T = 0.3854
MINIMUM DESIGN THICKNESS = 0.4746
USE: 0.475 IN. A573 70
Figure 3A-7 API 650 VARIABLE POINT CALCULATIONS
28

TABLE 3A-2 ADJUSTED FINAL THICKNESSES

(API 650 DESIGN)
RING #
5
4
3
2
1

THICKNESS
0.3125
0.3125"
0.3750
0.4750
0.5990

MATERIAL
A36
A36
A573GR70
A573GR70
A573G R70

Wtr (ring 5)
Wtr (ring 4)
Wtr (ring 3)
Wtr (ring 2)
Wtr (ring 1)
SUM OF Wtr

=
95.50 inches
=
95.50
=
60.54
=
33.61
=
18.79
=
303.94 inches
=
25.33 feet
Figure 3A-8 TRANSPOSED SHELL HEIGHT
(API 650 DESIGN)
top. If this is not done, the sand will percolate
Factors of safety lower than the above
down through the voids in the coarser rock.
minimums may be considered when actual
An excellent tank grade can also be obtained
experience with similar tanks and foundations
at a particular site indicates that satisfactory
by substituting about 1 inches of asphalt
performance can be expected.
road paving mix for the sand cushion. This
material is available from ready mix plants in
Tank Grade
many sections of the country. It is very
The tank grade (surface which supports the
important that the paved tank grade be
tank bottom) can be constructed of earth
constructed level and to the proper profile,
materials provided the subgrade beneath the
particularly near the shell. Once the asphalt
tank bottom is capable of supporting the
has set up, it is extremely difficult for the tank
weight of the contained fluid. The tank grade
builder to correct inaccuracies by taking down
usually consists of a 4" sand cushion placed
the high and filling in the low spots.
over properly compacted fill or soil. It is
Drainage is important both from the
recommended that the finished tank grade be
standpoint
of soil stability and bottom
constructed at least 6 inches above the
corrosion.
Good
drainage should be provided
surrounding ground surface and be crowned
under
the
tank
itself
and in the general area
from its outer periphery to its center. A slope
around
the
tank.
Where
the terrain does not
of 1 inch to 10 feet is suggested. The sand
afford
natural
drainage,
proper ditching
should be clean and free of corrosive elements.
around
the
tank
may
help
to correct the
Care should be taken to exclude lumps of
deficiency.
earth or other deleterious materials from
coming into contact with the bottom. These
materials can cause electrolytic action that will
result in pitting of the bottom plate.
If the sand cushion is placed on top of
crushed rock fill, the rock should be carefully
graded from coarse at the bottom to fine at the

Foundations
The shell of a flat bottom tank can be
supported on a compacted granular berm,
concrete ringwall or concrete slab foundation.
Local soil conditions, tank loads and the
intended use of the tank will determine which
29

of these foundations is suitable for a particular

site. Tanks that require anchor bolts must be
supported by ringwall or slab foundations.

on the inside horizontal projection of the

ringwall.
3. Bending, torsion and shear resulting
from lateral, wind or seismic, loads. A
rational analysis, which includes the
effect of the foundation stiffness, should
be used to determine the soil-bearing
stress distribution and the above internal
design forces.
The area of reinforcement provided must be
sufficient to resist the above forces and should
not be less than the following minimums.
These minimums are intended to prevent
excessive cracking due to shrinkage and
temperature.
1. For wall-like ringwalls, the area of
vertical reinforcement provided should
not be less than 0.0015 times the
horizontal cross-sectional area of the
ringwall.
2. The area of hoop reinforcement provided
should not be less than 0.0025 times the
vertical cross-sectional area of the
ringwall.
The detailed design of the ringwall and
requirements for the materials, construction
and testing should be in accordance with the
American Concrete Institute's Building Code
Requirements for Reinforced Concrete (ANSI/ACI
318).
Recesses shall be provided in the concrete
ringwall for flush type cleanouts, drain off
sumps and any other appurtenances that
require recessing. Refer to API 650 for details
of recesses at flush type cleanouts.
The top of the ringwall should be smooth
and level to within 1/8 inch in any 30 feet of
circumference. No point on the total
circumference should vary more than l/4
inch from the specified finish elevation.
Slab Foundation - When the subgrade
beneath the tank bottom cannot adequately
support the weight of the contained fluid, a
slab foundation is required. The area of the
slab must be sufficient to produce a soil
bearing (due to the total weight of the tank,
foundation and contained product) less than
the allowable soil bearing. The depth to the
bottom of the slab will depend on local
conditions and must be sufficient to place the
bottom of the slab below anticipated frost
penetration and within the specified bearing
strata.

Granular Berm Foundation - When a

qualified geotechnical evaluation concludes
that it is unnecessary to construct a ringwall or
slab foundation, the shell can be supported by
a granular berm foundation. The berm should
be constructed of well-graded and properly
compacted stone or gravel. The berm should
extend a minimum of 3 feet beyond and 2 feet
inside the tank shell as shown in Figure 3B-1.
The berm should be level to within 1/8 inch
in any 10 feet of circumference and to within
1/2 inch in the total circumference. Adequate
drainage away from the berm must be
provided to prevent erosion of the berm under
the shell. Alternatively, a welded or bolted
steel grade band can be used to retain the
outer portion of the berm.
Concrete Ringwall Foundation - When
suitable bearing is not available at the surface,
but is available at a reasonable depth below
the surface, a ringwall foundation should be
considered. The depth of the ringwall will
depend on local conditions and must be
sufficient to place the bottom of the ringwall
below anticipated frost penetration and within
the specified bearing strata. As a minimum,
the bottom of the ringwall should be located 2
feet below the lowest adjacent finish grade.
The width of the ringwall must be sufficient to
produce a soil bearing less than the specified
allowable soil bearing. As a minimum, the
ringwall width should be 1 foot. The inside
horizontal projection (inside the tank shell)
should be no less than 4 inches.
The ringwall must be reinforced to resist the
following forces:
1. Direct hoop tension resulting from the
lateral earth pressure on the inside face
of the ringwall. Unless substantiated by
proper geotechnical analysis, the lateral
earth pressure should be assumed to be
30% of the vertical pressure due to the
contained fluid and the soil weight.
2. Bending moment resulting from the
uniform moment load. The uniform
moment load is due to the eccentricities
of the shell and pressure loads relative to
the centroid of the soil bearing stress. The
pressure load is due to the fluid pressure
30

The detailed design of the slab and

requirements for the materials, construction
and testing should be in accordance with the
American Concrete Institute's Building Code
Requirements for Reinforced Concrete (ANSI/ACI
318).

References, Part Ill

1. API Standard 650 Welded Steel Tanks for

Oil Storage, Division of Refining,
American Petroleum Institute.
2. AWWA Standard D100 Welded Steel
Tanks for Water Storage, American Water
Works Association.
3. Manual of Steel Construction, American
Institute of Steel Construction, Inc.
4. "Oil Storage Tanks," The Prince William
Sound, Alaska, Earthquake of 1964 and
Aftershocks, Volume II, Part A, U.S. Dept.
of Commerce, 1967.
5. "Fluid Mechanics," Dodge and Thompson.

31

Figure 3B-1 Example of Foundation with Crushed Stone Ringwall from API 650

Note:

Bottom of excavation should be level. Remove any unsuitable material and replace with suitable fill,
thoroughly compacted.

Figure 3B-2 Example of Concrete Ringwall Foundation

32

This page has been left blank intentionally.

Part IV
Stainless Steel Tanks For Liquid Storage
Type 304
Possessing corrosion resistance, strength and
fabricability, this is the general purpose
stainless steel, long known as "18-8. Type 304
is the most widely used type and is extensively
specified for food handling and storage, dairy
equipment, nuclear fluids, and in general most
applications where even small amounts of
corrosion product would be intolerable.

Introduction
t the present time, the only rules for
stainless steel storage tanks are given
in Appendix Q of API Standard 620(l)
which covers low-pressure tanks for liquefied
hydrocarbon gases, particularly liquefied
ethane, ethylene, and methane, at a minimum
temperature of 270F. Rules for the design and
construction of, pressure vessels including
stainless steel vessels are given in the ASME
(2) Boiler and Pressure Vessel Code, Section
VIII, Division 1 and Division 2, Pressure
Vessels.
In the following discussion rules are
presented for design and construction of
stainless steel tanks at atmospheric pressures.
These rules are not intended to cover storage
tanks which are to be erected in areas subject to
regulations more stringent than specified in the
following pages. These rules are recommended
only insofar as they do not conflict with local
requirements.

Type 316
Containing higher nickel than Type 304, and
2-3% molybdenum, Type 316 possesses greatly
improved resistance to corrosion by pitting. It is
used under conditions too severe for Type 304,
such as mineral acids (phosphoric acid, sulfuric
acid), strong organic acids (oxalic, formic, etc.)
and halides in various dilutions.
Types 304L and 316L
Containing 0.03% maximum carbon, these are
the low carbon counterparts of Types 304 and
316. The lower the carbon content, the less the
chromium carbide that can be formed.
Chromium-nickel stainless steels form a grain
boundary chromium-carbide precipitate when
heated in the 800-1650F temperature range for
sufficient time (see Figure 4-1) (5). If the degree
of precipitation is severe i.e. the grains are
completely surrounded there may be a loss
of corrosion resistance in aggressive media such
as hot, oxidizing acids (e.g. strong nitric acid),
iron or copper sulfates in hot dilute sulfuric
acid, and air-saturated hot sulfuric acid. Such
aggressive corrosion conditions do not
normally exist in storage tanks.
Intergranular corrosion attack used to be a
common occurrence when the stainless steels
contained up to 0.12% carbon (as in Type 302,
for example). This was enough carbon to
remove considerable chromium from solution
during welding cycles, causing mild to heavy
carbide precipitation in the weld heat-affected
zone. Corrosive attack would be evident in this
zone, if the environment was severe. This
situation resulted in widespread specifying of
low carbon (0.03% maximum) stainless steels,
but it should be understood that there are
relatively few situations where the L grades are
actually required for storage vessels. Even these
should be carefully investigated to establish

FACTORS AFFECTING SELECTION OF STAINLESS

STEEL
There are a total of 62 stainless steel
compositions that are recognized as standard
by the American Iron and Steel Institute (3), as
well as commercially available proprietary
compositions.
The five stainless steels most generally used
as plate material for construction of liquid
storage tanks are Types 304, 304L, 316, 316L and
410S. The last is not recognized as standard by
the American Iron and Steel Institute. The
chemical compositions of these types are listed
in Table 4-1 and their mechanical properties are
listed in Table 4-2.
The selection of a particular type of stainless
steel for a given corrosive environment often
follows extensive study of comparative data,
and sometimes even pilot or service testing.
However, a general understanding of the
corrosion resistance capabilities of the five
stainless steels, in terms of their relative
resistance to various common media, is shown
in Table 4-3. The five types fall within two
categories: namely, Types 304, 304L, 316 and
316L are in the chromium-nickel group, while
Type 410S is in the straight chromium group.
33

such a need before the additional expense of the

L grades is incurred. Types 304 and 316 (0.08%
maximum carbon) can, in many cases, be
welded free of carbide precipitation. If a small
amount does develop, it may be unaffected by
the liquid being stored, except possibly as
indicated above.
It should be noted here that galvanized
material or other zinc products welded to
stainless steel will cause intergranular cracking.
In general, the L grades should be used when
and only when it is ascertained that
conditions will be present, which are conducive
to intergranular attack on as-welded 0.08%
maximum carbon stainless steel (see Figure 4-1).
The general corrosion and pitting resistance of
the L grades is not better than their higher
carbon counterparts in the annealed condition;
nor is there any advantage in weldability or
fabricability.

2.
3.

4.

5.

Type 410S
This straight-chromium stainless steel is not
subject to the above form of carbide
precipitation. It finds use where moderate
corrosion resistance is needed, and slight
product contamination is not critical (see Table
4-3). The low carbon (0.08% maximum) results
in a tough plate product which avoids
formation of the less-tough metallurgical
structures possible in high strength, low alloy
steels.

related to exposure to hot chloridecontaining corrosive media.

At temperature much below 160F, stress
corrosion failures are not very likely to
occur.
At temperatures exceeding 160F, the
time to failure of life of the vessel is very
dependent on temperature, concentration
of chloride and stress. Increased values in
each case will shorten the life of the
vessel.
While Types 304 and 316 are both
susceptible to hot chloride stress
corrosion cracking, Type 316 under
similar service conditions tends to give
better life than Type 304.
Because of variation in fabrication and
service stresses, it is frequently difficult to
predict the life of an austenitic stainless
steel vessel in hot chloride-containing
media.

STAINLESS STEEL PLATE

Manufacture
Stainless steel plate is defined as a flat rolled
or forged product, 3/16 inch (4.76 mm) and
over in thickness, and 10 inches (254 mm) and
over in width (3). It is formed in the same type
of equipment as utilized for carbon steel plate,
although
production
allowances
must
accommodate the much greater hot strength of
stainless steel over carbon and low alloy steels.
In producing plate, care is taken to attain the
good surface condition that is essential to
corrosion resistance. The first of several surface
cleaning operations occurs at the slab stage,
where the 4 to 10-inch-thick bloomed slab is
ground or scarfed overall to remove not only
the scale but some of the underlying base metal
as well. The slab is then cut to size, yielding the
ordered plate size, reheated and hot rolled. The
plate is then annealed, and again cleaned of
scale by either a chemical solution or
mechanical means, or both.
During the hot rolling, high pressure water
jets and other mechanical devices are employed
to assure that the refractory scale is not rolled
into the surface.
Light gauge plate (3/16 inch and 1/4 inch
thick) can be rolled in coils up to 60 inches wide
on continuous mills. This product normally has

Stress Corrosion Cracking

Another phenomenon associated with the
chromium nickel stainless steels Types 304,
304L, 316 and 316L, is stress corrosion cracking.
By definition, stress corrosion cracking
involves the combined action of a tensile stress
and a corrosive medium. Aside from some
ultra-pure metals, most commercial metals are
subject to this phenomenon in certain specific
environments. While the initial reaction may be
one of great concern, it should be emphasized
that throughout industry there are numerous
applications of stainless steel in environments
where stress corrosion cracking does not occur
or which have been engineered to avoid stress
corrosion cracking.
Although stress corrosion cracking is not
fully understood, there are some general
guidelines that can be related to stainless steels:
1. The form of stress corrosion usually
found in the AISI 300 stainless steels is
34

is clean edges (and near-edge areas) prior to

welding. Contaminants, whether organic or
inorganic, can cause problems when they
decompose in the arc heat. The oxide film on
stainless steel surfaces is very refractory and
reduces the wetting action between parent and
weld metal.
The filler metals for welding stainless steels
are listed in Table 4-4.
When stainless steel is welded to carbon or
low alloy steels, Type 309 electrodes are
normally used.

improved surface, gauge accuracy and offers

greater flexibility in length.
Forming
Press brake flanging or bending, and roll
bending are the most widely used cold forming
operations performed on stainless steel plate.
Type 410S has cold forming characteristics
similar to carbon and low alloy steels in the
35,000-50,000 psi yield strength range.
The chromium-nickel stainless steels (Types
304, 304L, 316, 316L), on the other hand, workharden quite rapidly with increasing plastic
deformation. As the steel increases in strength
with increased deformation, the bending forces
exerted by the forming equipment rise
commensurately.
The most noticeable effect of work hardening
is the greater degree of springback, compared
with carbon steel. Dies for brake bending and
rolls for roll bending must overbend the
stainless steel to compensate for the springback.
In brake bending, bending to a smaller radius
can compensate for the greater springback.

Cleaning and Passivation

These two terms are actually synonymous if
the word clean is strictly interpreted to mean
"the complete removal of all contaminating
materials from the stainless steel surface. If the
surface is clean, it will self passivate.
Iron particles and/or oil are the contaminants
most generally encountered, and to the extent
that they are present some surface staining or
rusting may occur. Whether such an occurrence
is serious or not depends on the requirements
of the application which can range from "food
quality surfaces" to simply "retention of
structural strength."
Prevention of contamination is to be
preferred over removal of contaminants. A
main source of contamination due to iron
particles results from comingling plate
fabrication operations involving carbon steel
and stainless steel. Another source is the
existence of weld scale or weld spatter, which
can only be removed by energetic chemical or
mechanical means. In extremely demanding
requirements, finishing may include a final
polishing to produce the brightest surface
possible.
Various cleaning practices are summarized in
ASTM A380 (6) and more comprehensively
described in ASTM Special Technical
Publication 538 (7).

Cutting
Most stainless steel plates are cut by tank
fabricators with the plasma arc process. Thin
plates can be sheared. Thicker plates can be cut
by saw cutting or abrasive wheel cutting. Gasoxygen (oxy-gas) is also applicable if used in
conjunction with iron powder. Stainless steel
cannot be cut by conventional flame cutting, but
in some cases may be cut and beveled with the
carbon arc gouge if the cut edges are ground to
remove oxides.
Welding
Gas metal arc and submerged arc welding are
high-production methods and are usually used
in the downhand position, fully automated.
Both give deep penetration and, for high
volume welding that can be positioned, are the
lowest cost methods for joining plates. A
modification of gas metal arc called interrupted
(or pulse) arc welding is useful for butt, fillet,
and lap welding.
Shielded metal arc welding is widely used for
all types of stainless steel welding, particularly
where automatic welding is impractical.
Advantages are low cost equipment and
mobility. Disadvantages are slow speeds and
high labor cost for skilled operators. An
essential requirement for any welding method

FACTORS AFFECTING TANK DESIGN

The design rules and details of API Standards
650 (8) and 620 (1) are applicable for stainless
steel tanks at atmospheric pressure with certain
exceptions. In the following discussion, the
design stresses recognize the increased
toughness of stainless steels over carbon steels
and the low yield/tensile ratios of Types 304,
304L, 316 and 316L. The increased toughness
permits designing to a higher proportion of
35

yield strength, but the lower yield strength

introduces the problem of permanent strain.
Section VIII of the ASME Code (2) recognizes
the strain by stating two allowable design
stresses for the austenitic stainless steels. The
higher stresses are related to the following
footnote to the ASME table of stresses:
"Due to the relatively low yield strength
of these materials, these higher stress
values were established at temperatures
where the short time tensile properties
govern to permit the use of these alloys
where slightly greater deformation is
acceptable. These higher stress values
exceed 62X percent but do not exceed 90
percent of the yield strength at
temperature. Use of these stress values
may result in dimensional changes due
to permanent strain. These stress values
are not recommended for flanges of
gasketed joints or other applications
where slight amounts of distortion can
cause leakage or malfunction.
After consideration of the allowable design
stresses of the ASME Code and the API
Standards, the following stress basis is
suggested for stainless steel tanks at
atmospheric pressure:
a. The design basis for shells where
permanent strain .05% is acceptable is
the lesser of:
Sd = .8 x Fy or
.3 x Ft
see Table 4-5(a)
b. The design basis for shells where
permanent strain .1% is acceptable is the
lesser of:
Sd = .9 x Fy or
.3 x Ft
see Table 4-5(b)
where Fy = min. yield strength
Ft
= min. tensile strength
Sd = design stress
c. Because the lower carbon grades (Types
304L, 316L) usually exhibit yield strengths
at room temperature greater than the
specified minimum value, the allowable
stress for 100oF has been based only on
the tensile factor.
d. Where a lower level of permanent strain
is desirable such as mentioned above for
gasketed joints or other applications
where slight amounts of distortion can
cause leakage or mechanical malfunction
see Table 4-5(c) for values.

e. The yield values at temperature can be

obtained from table AHA2 of ASME
Section VIII, Div. 2.
The allowable stresses listed in Table 4-5, a, b,
and c, result from these bases, with the higher
stresses for the austenitic grades taking into
account the greater deformation of item b
above. Table 4-5(b) should be the default basis
unless directed otherwise for shells and Table 45(c) should be used for flanges.
The design thickness for each shell course can
be calculated by the formula in API Standard
650, modified by the allowable stress and joint
efficiency:

t=
where:
t
=
C =

(2.6)(D )(H 1)(G ) + C

(E )(1000 S d )

minimum thickness, in inches

an additional thickness required for
corrosion allowance (rarely if ever
required)
D =
nominal diameter of tank, in feet.
This shall be the centerline diameter
of the shell plates, unless otherwise
specified by the purchaser
H =
height, in feet, from bottom of
course under consideration to top
of top angle, or to bottom of any
overflow which limits tank filling
height
G =
specific gravity of liquid to be
stored, but in no case less than 1.0
maximum allowable stress in ksi.
Sd =
Values tabulated in Table 4-5, a or b
E =
1.0 if tank is radiographed in
accordance with section 6
=
0.85 if tank is radiographed in
accordance with A.5.3
=
0.7 if tank is not radiographed. This
value shall be given as part of the
buyers' specifications
Stainless steel separation pads (or poison
pads) should be considered at points where
carbon steel is welded to the stainless steel tank
to avoid carbide precipitation. Typical areas for
this would be anchor bolts and support
brackets.
The recommended nominal thickness of shell
plates should not be less than the following
based on construction minimums:

36

Nominal Tank
Diameter
Smaller than 50'
50' to 120' excl.
120' to 200' incl.
Over 200'

FACTORS AFFECTING FABRICATION AND

CONSTRUCTION

Nominal Plate
Thickness
3/16"
1/4"
5/16"
3/8"

Before proceeding with any fabrication or

construction of a stainless steel tank,
satisfactory weld procedure qualifications
should be performed in accordance with all the
essential variables of Section IX, Welding
Qualifications, of ASME Boiler and Pressure
Vessel Code, including conditions of postweld
heat treatment or the omission of postweld heat
treatment. Requirements and restrictions for
postweld heat treatment are described in ASME
Section VIII. Materials that would require
postweld heat treatment should not be used for
storage tanks designed by the rules in this
manual.
All austenitic chromium-nickel alloy steel
welds, both butt and fillet, between plates
exceeding 3/4 inch nominal thickness, shall be
examined for detection of cracks by the liquid
penetrant method, before the hydrostatic test of
the tank. All cracks shall be eliminated.
Butt-welded joints in Type 410S welded with
electrodes
that
produce
an
austenitic
chromium-nickel weld deposit shall be
radiographed when the thinner plate at the
welded joint exceeds 1 inches.
Referring to Part AM of ASME Section VIII
Division 2 it will be noted that both values of
thermal conductivity (TC) and thermal
diffusivity (TD) (given in Btu/hr ft F and
ft2/hr respectively) are considerably lower
(about 2 to 1) for stainless compared to carbon
steel. This indicates that heat (from welding) is
not conducted away or diffused as rapidly with
stainless steels and therefore distortion is likely
unless design steps are taken to ensure that
nozzle location (with respect to vertical and
horizontal seams) should be thought out. Also,
the tolerances given in API 650 for banding and
peaking may not be achievable for stainless
steel shells.
For the design of stainless steel tanks at
refrigerated temperatures, the designer is
referred to Appendix Q, "Low-Pressure Storage
Tanks for Liquefied Hydrocarbon Gases" of API
Standard 620. This subject is beyond the
intended scope of this discussion.

Throughout this design procedure it shall be

remembered that Young's Modulus for stainless
steel is less than that of carbon steel. Therefore
designs for compression and stability should
consider this fact.
Normally the tank shell should be designed
to resist the design wind velocity given in the
customer's specifications. API Standard 650
provides rules for stiffening tank shells. The
background for the API rules was given in a
paper, "Stability of API Standard 650 Tank
Shells, presented in May 1963, to the API
Division of Refining.
For the design of shells under external load
(small negative pressures) the designer should
refer to ASME Section VIII or Part III of "Design
of Plate Structures" (11). For the design of
structural members, the designer should refer
to the Stainless Steel Cold-Formed Structural
Design Manual (10).
NOTE: Roof designs for stainless steel tanks
may be done in a similar manner as that
outlined in Part III for carbon steel tanks, but
normally all structural units will need to be
fabricated in custom shapes. A lighter gauge
lighter than .17 allowed in API 650 for carbon
steel structural units, may be used to
accommodate forming.

37

Figure 4-1

Time required for formation of carbide precipitation in stainless steels with

various carbon contents. Carbide precipitation forms in the areas to the right of
the various carbon-content curves. Within time-periods applicable to welding,
chromium-nickel stainless steels with 0.05% carbon would be quite free from
grain boundary precipitation. (5)

38

Table 4-1 STAINLESS STEELS COMMONLY USED

FOR CONSTRUCTION OF LIQUID STORAGE TANKS (4)
COMPOSITION, PERCENT
Phosphorus
Sulfur Silicon
max.
max.
max. Chromium

ASTM
Type

UNS
No.a

304

(S30400)

0.08

2.00

0.045

0.030

1.00

18.0020.00

8.0010.50

N 0.10 Max

304L

(S30403)

0.03

2.00

0.045

0.030

1.00

18.0020.00

8.0012.00

N 0.10 Max.

316

(S31600)

0.08

2.00

0.045

0.030

1.00

16.0018.00

10.0014.00

2.00-3.00
Molybdenum
N 0.10 Max.

316L

(S31603)

0.03

2.00

0.045

0.030

1.00

16.0018.00

10.0014.00

2.00-3.00
Molybdenum
N 0.10 Max.

410S

(S41008)

0.08

1.00

0.040

0.030

1.00

11.5013.50

0.60
(max)

Carbon Manganese
max.
max.

Nickel

Other
Elements

Unified Numbering System, originated by ASTM and SAE, developed to provide a single orderly system for designating
commercial metals and alloys.
a

Table 4-2 MECHANICAL PROPERTIES OF STAINLESS STEELS

COMMONLY USED FOR CONSTRUCTION OF LIQUID STORAGE TANKS (4)

ASTM
Type

UNS
No.

Tensile Strength,
min
ksi
MPa

Yield Strength,
min
ksi
MPa

Elongation, a
min
Percent

Hardness,
max
Bhnb
RBc

304

(S30400)

75

515

30

205

40

202

92

304L

(S30403)

70

485

25

170

40

183

88

316

(S31600)

75

515

30

205

40

217

95

316L

(S31603)

70

485

25

170

40

217

95

410S

(S41008)

60

415

30

205

22.0

183

88

Elongation in 2 inches (50.8 min)

Brinell
c Rockwell-B
a

39

Table 4-3 RELATIVE CORROSION RESISTANCE OF STAINLESS STEELS

COMMONLY USED FOR CONSTRUCTION OF LIQUID STORAGE TANKS
ASTM

UNS

Mild Atmospheric

Atmospheric

Salt

Chemical

Type

No.

and Fresh Water

Industrial Marine

Water

Mild

Oxidizing

Reducing

304

(S30400)

304L

(S30403)

316

(S31600)

316L

(S31603)

410S

(S41008)

Note: Xs indicate environments to which the various stainless steels may be considered resistant.

Table 4-4 TYPICAL FILLER METALS FOR WELDING STAINLESS STEELS

Base Metal

Electrodes (AWS)

Type 304

E308-15 or 16; ER308; E308T-2

Type 304L

E308L-15 or 16; ER308L; E308T-2

Type 316

E316-15 or 16; ER316; E316T-2

Type 316L

E316L-15 or 16; ER316L; E316T-2

Type 410S**

E410*-15; ER410*; E410T*-2

* Type 410 electrodes must be specified to 0.08% maximum carbon in all cases.
** It is permissible (and often desirable) to weld Type 410S with austenitic (chromium-nickel) electrodes.

40

Table 4-5 ALLOWABLE STRESSES FOR TANK SHELLS OF STAINLESS STEEL

Minimum
Yield,
KSI

Minimum
Tensile,
KSI

304a

30.0

304b

Type

100F

For Metal Temperatures Not Exceeding

200F
300F
400F
500F

600F

75.0

22.5

20.0

18.0

16.6

15.5

14.6

30.0

75.0

22.5

22.5

20.3

18.6

17.5

16.4

304La

25.0

70.0

21.0c

17.0

15.3

14.0

13.0

12.4

304Lb

25.0

70.0

21.0c

19.2

17.2

15.8

14.7

14.0

316a

30.0

75.0

22.5

20.6

18.6

17.1

15.9

15.0

316b

30.0

75.0

22.5

22.5

21.0

19.3

17.9

16.8

316La

25.0

70.0

21.0c

16.9

15.1

13.8

12.7

12.0

316Lb

25.0

70.0

21.0c

19.0

17.0

15.5

14.3

13.5

410Sa

30.0

60.0

18.0

18.0

18.0

18.0

18.0

18.0

Note: a, b, and c explained on page 39.

41

Table 4-5(a) ALLOWABLE STRESSES FOR TANK SHELLS OF STAINLESS STEEL

Limiting % Strain = 0.05%
Minimum
Yield,
KSI

Minimum
Tensile,
KSI

100F

304

30

75

22.5

20

18

16.6

15.5

14.6

304L

25

70

21

17

15.3

14

13

12.4

316

30

75

22.5

20.6

18.6

17.1

15.9

15

316L

25

70

21

16.9

15.1

13.8

12.7

12

410S

30

60

18

18

18

18

18

18

Type

For Metal Temperatures Not Exceeding

200F
300F
400F
500F

600F

Table 4-5(b) ALLOWABLE STRESSES FOR TANK SHELLS OF STAINLESS STEEL

Limiting % Strain = 0.1%
Minimum

Minimum

Yield,

Tensile,

KSI

KSI

100F

200F

300F

400F

500F

600F

304

30

75

22.5

22.5

20.3

18.6

17.5

16.4

304L

25

70

21.0

19.2

17.2

15.8

14.7

14.0

316

30

75

22.5

22.5

21.0

19.3

17.9

16.8

316L

25

70

21.0

19.0

17.0

15.5

14.3

13.5

Type

For Metal Temperatures Not Exceeding

Table 4-5(c) ALLOWABLE STRESSES FOR FLANGES OR GASKETED JOINTS OF STAINLESS STEEL
Limiting % Strain = 0.01% per 62.5% Fy Yield Strength Limit
Minimum

Minimum

Yield,

Tensile,

KSI

KSI

304

30

75

304L

25

316

Type

For Metal Temperatures Not Exceeding

100F

200F

300F

400F

500F

600F

20.0

16.7

15.0

13.9

12.9

11.5

70

16.7

14.3

12.8

11.7

10.9

10.3

30

75

20.0

16.7

15.0

13.9

12.9

11.5

316L

25

70

16.7

14.3

12.8

11.7

10.9

10.3

410S

30

60

18

18

18

18

18

18

42

Table 4-6
FACTORS FOR LIMITING PERMANENT STRAIN
IN HIGH-ALLOY STEELS1
Limiting Permanent
Strain, %
0.10
0.09
0.08
0.07
0.06
0.05
0.04
0.03
0.02
0.01

5. Svetsaren English edition 1-2; 1969, p. 5.

6. ASTM Designation A380-78 (ANS G81.16)
Standard Recommended Practice for
Cleaning and Descaling Stainless Steel
Parts, Equipment and Systems.
7. ASTM S.T.P. 538 "Cleaning Stainless Steel"
includes ASTM A380 and 22 papers
presented at a symposium.
8. API Standard 650 Welded Steel Tanks for
Oil Storage; Division of Refining,
American Petroleum Institute, Eighth
Edition.
9. ASTM Designation A370-77 (ANS G60.1)
Standard Methods and Definitions for
Mechanical Testing of Steel Products.
10. "Stainless Steel Cold-Formed Structural
Design Manual," American Iron and Steel
Institute.
11. "Design of Plate Structures," Vol. 2,
AISI/SPFA.
12. Steel Products Manual - Plates; Rolled
Floor Plates: Carbon, High Strength Low
Alloy, and Alloy Steel.

Factors
0.90
0.89
0.88
0.86
0.83
0.80
0.77
0.73
0.69
0.63

NOTE:
(1) Table 4-6 lists multiplying factors which, when
applied to the yield strength values shown on
Table AHA-2, will give a value that will result in
lower levels of permanent strain. If this value is
less than the design stress intensity value listed in
Table AHA-1, the lower value shall be used.

References, Part IV
1. API Standard 620 Recommended Rules
for Design and Construction of Large,
Welded, Low-Pressure Storage Tanks;
Division of Refining, American Petroleum
Institute.
2. American Society of Mechanical Engineers.
3. "Steel Products Manual Stainless and
Heat Resisting Steels," American Iron and
Steel Institute.
4. ASTM Designation A240-80b (ANS G81.4)
Standard Specification for HeatResisting Chromium and ChromiumNickel Stainless Steel Plate, Sheet and Strip
for Fusion-Welded Unfired Pressure
Vessels.

Other Information on Corrosion

"Corrosion Resistance of the Austenitic
Chromium-Nickel Stainless Steels in Chemical
Environments," The International Nickel Co.,
April 1970, 16 pages.
"Selection of Stainless Steels," American
Society for Metals, 1968, 82 pages.
Corrosion Engineering," G. Fontana and
N.D. Greene, McGraw-Hill Book Co., 1967.
"The Possibility of Service Failure of
Stainless Steels by Stress Corrosion Cracking,"
J.E. Truman and H.W. Kirkby, Metallurgia,
August 1965.

43

This page has been left blank intentionally.

Steel Plate Engineering Data-Volume 2

Useful Information on the

Design of Plate Structures
Revised Edition 2011

The material presented in this publication is for general information

only and should not be used without first securing competent advice
with respect to its suitability for any given application. The publication
of the material contained herein is not intended as a representation or
warranty on the part of the Steel Market Development Instituteor of
any other person named hereinthat this information is suitable for
any general or particular use or of freedom from infringement of any
patents. Anyone making use of this information assumes all liability
arising from such use.

Published by
STEEL MARKET DEVELOPMENT INSTITUTE,
A business unit of the American Iron and Steel Institute

In cooperation with and editorial collaboration by

STEEL PLATE FABRICATORS ASSOCIATION, Div. of STI/SPFA

Copyright Steel Market Development Institute 2011

This page has been left blank intentionally.

Introduction

he purpose of this publication is to provide a design reference for the usual design of tanks
for liquid storage. Volume 1, "Steel Tanks for Liquid Storage, deals with the design of flatbottom, cylindrical tanks for storage of liquids at essentially atmospheric pressure. Volume 2,
"Useful Information on the Design of Plate Structures, provides information to aid in design of such
structures.
Scope
Volume 2, "Useful Information on the Design of Plate Structures, covers many facets of plate
design that are generally applicable to many types of structures. Information on these is now
conveniently collected in one source.
Drawing on many sources, this volume offers such information and discusses some of the more
commonly encountered problems. Included is an outline of membrane theory, data for weld design,
commonly used details, plus data and mathematical tables useful in design of steel plate structures.
The intent is to include information principally pertinent to plate structures. For convenience of
users of this volume, some data readily available elsewhere, particularly in mathematical tables, has
been incorporated.

Inquiries for further information on design of steel tanks should be directed to:
Steel Plate Fabricators Association
Division of STI/SPFA
944 Donata Court
Lake Zurich, IL 60047
www.steeltank.com

ii

Contents

Part I
Part II
Part III
Part IV
Part V
Part VI
Part VII
Part VIII

Flat Plates ....................................................................................1

Membrane Theory .....................................................................7
Self-Supported Stacks ..............................................................17
Supports for Horizontal Tanks and Pipe Lines ...................25
Anchor Bolt Chairs ..................................................................42
Design of Fillet Welds .............................................................45
Inspection and Testing of Welded Vessels ...........................57
Appendices ...............................................................................59

Part I
Flat Plates

elastic limit. In general it must not be expected

that these formulas will yield stresses accurate
to better than 5%.
The mode of support and manner of loading
specified must be complied with if the stresses
are to be realized. No commercial edge
fastening will correspond exactly with the
theoretical conditions. The exact restraint of the
edge, where bending is of prime importance,
will depend on the rigidity of the support, the
flexibility of any gaskets used, the position of
the bolting circle and the spacing of the bolts
therein, as well as the tightness with which the
joint is bolted up. When membrane action is of
importance, the degree of bolting up and the
ability of the reinforced opening to resist slight
deformations under radial tensions will largely
determine the exact stress in the plate and the
corresponding deformation. The bending
moment at the edge is of less importance than
at points where plate resistance depends
primarily on bending. In view of these remarks,
the conditions "Fixed" and "Supported" serve as
guides to the possible range of stress and
deflection.
Notation
a = length, in. of semi-minor axis of
supporting ellipse for elliptical plates
A = length, in., of semi-major axis of
supporting ellipse for elliptical plates
b = length, in., of short side of rectangular
plate at supports
B = length, in., of long side of rectangular
plate or side of square at supports
B1 = factor for stress in uniformly loaded,
fixed edge, rectangular plates (Tables
1-1A and 1-1B)
B2 = factor for stress in uniformly loaded,
simply supported, rectangular plate (see
Tables 1-1A and 1-1B)
E = modulus of elasticity, psi
f = maximum fiber stress in bending, psi
Fy = specified minimum yield strength, psi
H = uniform load, ft. of water
LS = stiffener spacing, in.
n = a/A or b/B
p = uniform load or pressure, psi

lat plates are used in many conventional

structural forms, such as plate girders,
built-up columns, or component parts of
trusses. Such uses are well covered in standard
texts or handbooks and are not discussed in this
volume. Instead, Part I will cover applications
in steel tanks.
Bending Stresses and Deflections
Used as a membrane, as in the shell of a tank,
a steel plate is a very efficient member. In
contrast, a flat plate in bending normal to its
plane
is
inefficient.
Circumstances,
nevertheless, sometimes dictate the use of a flatwalled tank because of space limitations, or the
storage of a corrosive liquid may dictate use of
a grillage-supported bottom to facilitate
inspection. In such cases, a stiffened flat surface
is indicated.
On the next page, formulas are given for
calculating the maximum bending stresses and
maximum center deflections of certain flat
plates. These formulas have been derived from
various sources, the most important being
based on an analytical derivation from elastic
theory. However, those relating to three classes
of elliptical plates and to certain others with a
central applied load are less rigid in their
derivation though sufficiently reliable for the
use of the designer. It must be remembered that
all formulas apply to materials such as steel, for
which Poisson's ratio is 0.30.
The inherent limitations of these formulas
must be kept in mind. It is assumed that
tensions in the plane of the plate appropriate to
membrane action are small or negligible
compared with the stresses due to bending. In
general, the deflection must be small compared
with the plate thickness if this is to be true. For
greater deflections, other more complicated
formulas must be used in whose derivation
both membrane and bending action are
considered. The formulas given may yield
reliable working stresses yet be absolutely
unreliable in calculating the load at failure and
the corresponding deflection, particularly in the
case of materials which elongate materially
before failure, or which assume a dished form
under load through initial stressing beyond the
1

P =
r =
r =

concentrated load, lb
radius, in., of central loaded area
inside knuckle radius, in., for flat,
unstayed, circular plates
R = radius, in., to support for circular plates
S = spacing, in., of adjacent staybolts at
corners of square plates
t = plate thickness, in.
= center deflection, in., of plate relative to
supports
= factor for stress in circular flanged plate
(see Table 1-1A)
1 = factor for deflection of uniformly
loaded, fixed-edge, rectangular plates
(see Tables 1-1A and 1-1B)
2 = factor for deflection of uniformly
loaded, simply supported rectangular
plates (see Tables 1-1A and 1-1B)
3 = factor for deflection of fixed-edge,
rectangular plates subjected to central
concentrated load (see Tables 1-1A and
1-1B)
One of the most commonly encountered
conditions is a uniformly loaded flat plate
supported on uniformly spaced parallel
stiffeners. In the absence of any code or
specification requirement, assume an allowable
bending stress equal to 3/4 of the specified
minimum yield stress value in the plate for
determination of stiffener spacing Ls, in.
The plate stress can be obtained from the
formula in Table 1-1A for the case of a rectangle
b x B, where B = and b is taken as Ls. Thus, for
the fixed condition (continuous over the
supports), the maximum permissible spacing of
stiffeners becomes:

taken into account. An alternative solution,

therefore, is to assume that yielding does occur
at the support and the plate acts as a catenary
between supports. At intermediate supports,
the tension in the plate will be balanced; but at
the outside support, restraint must be provided
to resist that tension. This is not always easily
accomplished.
When the span is such that the profile of the
plate approaches a catenary between supports,
the support spacing is given approximately by
the following formula:
1/ 2

Ls =

Ls = 900

t
t
= 2.076
p
H

(1-4)

Figure 1-2 gives graphical solutions for

Equations 1-3 and 1-4.
For the catenary approach, it is essential that
a lateral force of 10,000t be resisted at the
peripheral support. Since this is not always
feasible, application of the catenary approach is
limited. Similarly, it should not be used where
pressure is reversible or where deflection is
objectionable.
In the above discussion, only plate stresses
have been considered, and it is assumed that
any welded plate joints will develop the full
strength of the plate including appropriate joint
efficiencies. Also, the stiffener system should be
in accordance with accepted structural design
principles.
Protection against brittle failure of a structure
should be considered at the time of design.
Since environmental extremes, design detail,
material selection, fabrication methods and
inspection adequacy are all interrelated in
protecting a structure from such failure, these
factors should be evaluated.

(1-1)

For convenience in connection with tank

bottoms, the load can be expressed in feet of
water, rather than psi, in which case:
1/ 2

124,615t 2

Ls =

(1-3)

Because of the approximate nature of the

solution, a conservative value for f is indicated.
Assuming f = 10,000t and E = 29,000,000 psi for
mild carbon steel, the equation becomes:

1/ 2

54,000t 2

Ls =

tf 24 f

p F

(1-2)

Figure 1-1 gives graphically stiffener spacing

determined from Equations 1-1 and 1-2 for an
allowable bending stress of 27,000 psi (i.e. Fy =
36,000 psi).
If deflection exceeds t /2, the plate will tend
to act as a membrane in tension and exert a
lateral pull on the outside support that must be
2

Figure 1-1. Stiffener Spacing for Flat Plate Acting as Continuous Beam.

Figure 1-2. Stiffener Spacing for Flat Plate with Catenary Action.
3

Table 1-1A. Flat Plate Formulas

Poissons Ratio = 0.30
SHAPE

Loading

Uniform

Circle
Radius
R

Edge
Fixation

0.75 p

Supported

1.24 p

Fixed

on r

Supported

Uniform

p
Ellipse

Fixed

Supported

2A X 2a
a<A

Uniform

Central
concentrated

Uniform

Square
BXB
Central
concentrated

Flat
Stayed
Plate

Circular
Flanged
*

R2
t

R2
t

R
r
1 . 43 log 10
P /t2
+ 0 . 11

r
R

1 . 43 log

10

R
r
+ 0 . 334 + 0 . 06

3n + 2n + 3
4

Fixed

Uniform

Uniform

Supported

0.42n + n + 1

B2 p

Fixed
Supported

t2

t2
P

P
2

1 + 2.4n t
2

Fixed

1.32

Supported

1.58

Staybolts spaced
at corners of
square of side S

0.228 p

Fits n=0 and n=1

n=a/A Approximate
Fits n=1, load over
0.01% of area

5.3

B2
t2
B2
t

n=a/A Approximate

p b4

E t3

1 and B1 depend on B/b.

p b4

E t3

2 and B2 depend on B/b.

p b2

E t3

b/B = n Approximate
Fits n =1 and n = 0

2
P

s2
t2

r
r
R 1+
r
R
p + 2
t

2t

See Table 1B.

Fits n =1 and n = 0

p B4
0.0138
E t3
p B4

E t3

0.0443

max. center of side

max. of center
As above.
Deflection nearly exact.

t2
2

See Table 1B.

b/B = n Approximate

n=a/A Exact Solution

n=a/A Approximate
Fits n=0 and n=1
Load over 0.01% of area

1 + 2n t

0.287 p

Supported

As above
Center Stress

4.00

0.308 p

p R2

E t3

b2
t

P uniform over
circle, radius r.,
Center Stress

0.55

b2

Fixed

Fastened
to shell

a2

0.42n + n + 2.5 t

Supported

p R2

E t3

p a4

3n + 2n + 3 E t 3

max. at center

1.365

13.1

B1 p

3n 4 + 2n 2 + 12.5 t 2

Fixed

p R4

E t3

0.22

a2

3
4

max. at edge

0.695

Remarks

p R4

E t3

0.17

50
Central
Concentrated

Rectangle
BXb
b<B

Fixed

Central
concentrated

Center Deflection

In.

Maximum Fiber Stress psi

p B2

E t3

Approximately for ;
Area of contact not too
small.

p S4

E t3

If plate as a whole de-forms,

superimpose the stresses
and deflections on those for
plate flat when loaded.

0.0125

0.0284

varies with shell and joint

stiffness from 0.33 to 0.38
Knuckle Radius, r

Formula of proper form to fit circle and infinite rectangle as n varies from 1 to 0
Formulas for load distributed over 0.0001 plate area to match circle when n = 1. They give reasonable values for stress when n = 0.
Stress is lower for larger area subject to load.
Formulas of empirical form to fit Hutte values for square when n = 1. They give reasonable values when n = 0. Assume load on 0.01 of area.
Apparent stresses only considered.
These formulas are not to be used in determining failure.

Table 1-1 B. Flat-Plate Coefficients

Stress Coefficients Circle with Concentrated Center Load

r/R

1.0

0.10

0.09

0.08

0.07

0.06

0.05

0.04

0.03

0.02

0.01

Fixed1

0.157

1.43

1.50

1.57

1.65

1.75

1.86

2.00

2.18

2.43

2.86

Supported2

0.563

1.91

1.97

2.05

2.13

2.23

2.34

2.48

2.66

2.91

3.34

4.0

5.0

2.00
0.455

Stress and Deflection Coefficients - Ellipse

A/a

1.0

1.2

1.4

1.6

1.8

2.0

2.5

3.0

Uniform Load
Fixed.
Stress3
Deflection4

0.75
0.171

1.03
0.234

1.25
1.284

1.42
0.322

1.54
0.350

1.63
0.370

1.77
0.402

1.84
0.419

1.91
0.435

1.95
0.442

Uniform Load
Supported5

1.24

1.58

1.85

2.06

2.22

2.35

2.56

2.69

2.82

2.88

Central Load
Fixed6
Supported7

2.86
3.34

3.26
3.86

3.50
4.20

3.64
4.43

3.73
4.60

3.79
4.72

3.88
4.90

3.92
5.01

3.96
5.11

3.97
5.16

3.00
4.00
5.24

Stress and Deflection Coefficients - Rectangle

B/b

1.0

1.25

1.5

Stress B1

0.308

0.399

0.454

Stress B 2

1.75

2.0

0.490

0.497

0.287

0.376

0.452

0.517

0.569

0.610

1.33

1.75

2.12

2.25

2.42

2.67

1.56

2.09

2.56

2.74

2.97

3.31

Deflection 1

0.0138

0.0199

0.0240

0.0264

0.0277

Deflection 2

0.0443

0.0616

0.0770

0.1017

0.1106

0.1336

0.1400

Deflection 3

0.1261

0.1802

0.1843

0.1848

4
1 + 2n 2
5.3
1 + 2.4n 2

1.6

0.0906

0.1671

Values of 1.43 [ log10 R /r + 0.11 (r/R)2 ]

2.5

3.0

0.741

0.748

0.750

3.03

3.27

3.56

3.70

4.00

3.83

4.18

4.61

4.84

5.30
0.0284

5Values

of 3/[0.42n4 + n 2 + 1)

6Values

of 50/(3n4+ 2n2 + 12.5)

3Values

of 6/[ 3n4 + 2n2 + 3)

7Values

of 13.1/[0.42n4 + n 2 + 2.5)

Values of 1.365/[ 3n4 + 2n2 + 3)

0.713

of 1.43 [ log10 R / r + 0.334 + 0.06 (r/R)2 ]

5.0

0.500

2Values

4.0

0.1416

0.1422
0.1849

This page has been left blank intentionally.

Part II
Membrane Theory

ost vessels storing liquid or gas are

surfaces of revolution, formed by
rotation of one or more continuous
plane curves about a straight line in their
plane. The line is called the axis of revolution.
All sections of a shell of revolution
perpendicular to the axis of revolution are
circles. Usually the axis of revolution of a
storage vessel is vertical, in which case all
horizontal sections are circles.

Note:
Radii R1 and R2 lie in the same line, but have
different lengths except for a sphere where R1
= R2.
T1 and T2 are loads per inch and will give the
membrane stress in the plate when divided by
the thickness of the plate.
General Equation for Membrane Forces
Consider an element of a spherical section of
unit length in each direction.
Figure 2-1 indicates the radii and forces T1
and T2 acting on the element.
Figures 2-2 and 2-3 indicate the pressure on
the element and the components of the
membrane unit forces in the latitudinal and
meridional planes.
For equilibrium, the summation of forces
must be equal to zero.

Notation
P = The internal pressure on shell. It may be
due to gas alone (PO), liquid alone (PL), or
both together (P + P) (psi).
T1 = The meridional force (sometimes called
longitudinal force). This is force in
vertical planes, but on horizontal sections
(pounds per inch).
T1 is positive when in tension.
T2 = The latitudinal force (sometimes called
hoop or ring force). This is force in
horizontal planes, but on vertical section
(pounds per inch).
T2 is positive when in tension.
R = Horizontal radius at plane under consideration from axis of revolution (in).
R1 = Radius of curvature in vertical
(meridional) plane at level under
consideration (in). Generally Rl is
negative if it is on the opposite side of the
shell from RP.
R2 = Length of the normal to the shell at the
plane under consideration, measured
from the shell to its axis of revolution
(in). Generally R2 is positive unless the
plane results in more than one circle.
W = Total weight of that portion of the vessel
and its content, either above or below the
plane under consideration, which is
treated as a free body in computations
for such plane (pounds). W has the same
sign as P when acting in the same
direction as the pressure on the plane of
the free body, and the opposite sign from
P when acting in the opposite direction.
AT = Cross-sectional area of the interior of the
vessel at the plane under consideration
(square inches).
= Density of product (pounds per cubic
inch).

Outward force = PR2 2 R1 1

Inward Force = 2T1 2 R 2 2 + 2T2 2 R1 1

2
2
Equating the two:
PR2 2 R1 1 =
1

2T1
R 2 2 + 2T2 2 R1 1
2
2
PR1 R 2 = T1 R 2 + T2 R1
P =

T1 T2
+
R1 R 2

(2-1)

Equation 2-1 is the general equation for

membrane forces. This equation considers
membrane forces primarily produced by the
product contained within the vessel. The
weight of the vessel itself may add to these
forces and should be considered in the
analysis.
Modified Equations for Membrane Forces
In general, the meridional force is the unit
force in the wall of the vessel required to
support the weight of the product, internal
pressure and plate weights at the plane under
consideration.
In the free body diagram (Figure 2-5),
consider the forces acting at Plane 1-1.
The total forces acting at Plane 1-1 from
above the Plane = P R2.
7

General Equation for Membrane Forces

Figure 2-1

Figure 2-2

Figure 2-3
8

Modified Equations for Membrane Forces

Figure 2-4

Figure 2-5
9

Total forces acting at Plane 1-1 from below

the plane = W.
Total vertical downward force= P..R2 + W
Vertical force required along circumference at
Plane 1-1 to support the downward forces:
TVI =

T
T2 = R2 P 1
R1

For Figures 2-10, 2-11, 2-12, and 2-13 where

R1 = , the equations for membrane forces
reduce to:

P..R 2 + W
2R

T1 =

Membrane force
T1 =

TVI
P..R 2 + W
=
Sin 2 R Sin

T2 = PR2

The sign of R1, R2, P, W, and AT are shown

in Table 2-1 and must be included in computing
the forces.
For any other vessel configuration, a free
body diagram can be drawn and the forces T1
and T2 calculated in a similar way.

PR
W
T1 =
+
2 Sin 2 R Sin

or

R
2 Sin

P +
R 2

Further Simplifications

R
= R2 and R2 = AT
Sin

Since
T1 =

R2
2

P +
AT

R2
W
P +

2
AT

The equations for membrane forces can be

further simplified for some of the shapes.
(2-2)

a. Spheres
For spheres with no product (gas pressure
only), the equations reduce to:

From Equation 2-1

T
T2 = R 2 P 1
R1

PG R2
2

PR2
T2 = R 2 P

2R1

T1 =

These are the equations used in API 620.

Simplified Equations for Commonly Used Shapes
Figures 2-6 to 2-14 show the common vessel
shapes used and the direction and magnitude of
the radii, pressure, and weights acting on the
free body diagram. Table 2-1 indicates the sign
for each variable. The figures use the following
notations:

Since R1 = R2 = R
T1 = T2 =

PR
2

where R = radius of sphere.

b. Cylinders
If the weight of the plate is neglected and
there is no internal pressure in the vessel and
since R2 = R :

Level of product in the vessel.

Volume of product to be used in
calculating the weight of product above
or below the free body diagram.

T1 =

Area of plate to be used in calculating

the weight of plate above or below the
free body diagram.
For all figures:

R
R 2H

PL
2
R 2

Since H = PL
T1 = 0

P = PG + H

T2 = PL.R
where R = radius of cylinder.

AT = R2
For Figures 2-6, 2-7, 2-8, 2-9, and 2-14, the
equations for membrane forces are:
10

Figure 2-6
Spherical Vessel or Segment. Plane below line of support.

Figure 2-7
Spherical Vessel or Segment. Plane above line of support.
11

Figure 2-8
Spheroidal Vessel or Segment. Plane below line of support.

Figure 2-9
Spheroidal Vessel or Segment. Plane above line of support.

12

Figure 2-10
Conical Vessel or Segment. Plane below line of support.

Figure 2-11
Conical Vessel or Segment. Plane above line of support.
13

Figure 2-12
Conical Vessel or Segment. Pressure on convex side. Conical Vessel
or Segment. Pressure on convex side. Plane above line of support.

Figure 2-13
Cylindrical Vessel. Plane above line of support.
14

Figure 2-14
Curved Segment. Pressure on convex side.
Plane above line of support.

TABLE 2-1

Figure
2-6
2-7
2-8
2-9
2-10
2-11
2-12
2-13
2-14

R1
+
+
+
+

R2
+
+
+
+
+
+
+
+
+

P
+
+
+
+
+
+
+
+

15

W
+
+
+
+
-

AT
+
+
+
+
+
+
+
+
+

This page has been left blank intentionally.

Part III
Self-Supported Stacks

Scope

If investigation shows that stresses due to

vibrations are not within safe limits, the only
solutions are to change the stack diameter or to
add a damping device. Such devices might
consist of a gunite or similar lining or so-called
wind spoilers on the exterior of the stack.
The subject is quite complex. To attempt a
brief summarization could be dangerously
misleading. Instead, a bibliography of
references is appended at the end of this part
for the benefit of those who wish to explore
the subject more thoroughly.

he scope defined for this Volume

stated that stacks would not be
discussed in detail because of the
complicated problem of resonant vibrations.
Apart from this phase, however, there are
purely structural facets that may be of interest.
For the benefit of those not familiar with the
problem, a brief explanation of stack vibration
follows:
Wind-Induced Vibrations
When a steady wind blows on an
unsheltered, unguyed stack, formation and
shedding of air vortices on each side of the
stack can apply alternating lateral forces that
cause movement of the stack perpendicular to
the direction of the wind.
The frequency of vortex shedding is a
function of wind velocity and stack diameter.
The term critical velocity denotes the wind
velocity at which the frequency of vortex
shedding equals the natural frequency of the
stack. Under such conditions, resonance
occurs.
Excessive lateral dynamic deflection and
vibration of the stack from vortex shedding
may occur at wind velocities considerably
below the maximum wind velocity expected in
the area.
One way to avoid resonance and consequent
damage to the stack is to proportion the stack
so that the critical wind velocity exceeds the
highest sustained wind velocity that is likely to
occur. In most areas, for example, it is unlikely
that a steady wind of more than 75 mph will
occur. Hence, a stack having a critical velocity
of 75 mph is probably safe in those regions,
though gusts of greater velocity might occur.
There may be reasons, however, why a stack
of such proportions will not serve the purpose.
If so, the effects of dynamic vibrations must be
thoroughly investigated. If the critical wind
velocity is low enough, it may be that the
stresses due to dynamic deflections are within
design limits. In that case, the stack is
structurally adequate if noticeable movement
of the stack is not objectionable.

Minimum Thickness and Corrosion

In view of the corrosive nature innate to
stack operation, it is wise to add a corrosion
allowance to the calculated shell thickness. The
nature of the flue gasses and moisture content
in the area are some important parameters in
determining the amount of corrosion for
which to allow.
Erection requirements usually dictate
minimum plate thicknesses and the stress
formulae in this part are not considered valid
for thicknesses less than 1/4". Therefore, the
minimum thickness for shell plate is taken to
be 1/4" nominal.

Notation
A

As

C
Cc
CL
D
Do
E
EL
Fa

17

= Cross-sectional area of base ring, in.2

= Vertical angle of cone to cylinder, deg.
= Required area for stack stiffeners, in.2
= Critical damping ratio of stack
= See Figure 3-4 (Sec. A-A)
= Euler Factor
= Lift coefficient (0.2 for circular cylinder)
= Outside diameter of stack, in.
= Outside diameter of cylindrical portion
of stack, ft.
= Modulus of elasticity, psi at design
temperature
= Joint efficiency for base plate design
= Allowable compressive stress for
circumferential stiffeners, 12,000 psi
(unless otherwise noted)

Fb
Fc
Fcr
FL
Fs
Fy
FS
H
H1
Is
K
K
Ls
L
Ls1
M
N
Pd
R1
Ro
S
Ss
T
V
Vcr1
Vcr2
Vo
W
W
Ws
do
fc
fo
ft
g
h
p
qcr
r
t

= Allowable bending stress, 0.6 F4, psi for

stiffeners
= Allowable compressive stress, ksi
= Critical buckling stress, ksi
= Equivalent static force, lb/ft of height
= Allowable compressive stress, psi (in
cone cylinder junction area)
= Yield point of stack material, ksi
= Factor of safety
= Overall height of stack, ft.
= Overall height of stack, in.
= Required moment of inertia for stack
stiffeners, in.4
= Effective length factor
= Slenderness reduction factor
= Stiffener spacing, ft.
= Length for KL/r
= Stiffener spacing, in.
= Moment at any design point, inchpounds
= Number of anchor bolts
= Wind load, psi
= Outside conical radius, in.
= Outside radius of cylinder portion of
stack, in.
= Strouhal number (0.2 for steel stack)
= Required section modulus for stack
stiffeners, in.3
= Load per bolt, lb.
= Total direct load at any design point, lb.
= Critical wind velocity, mph
= Critical wind velocity, ft/sec.
= Resonance velocity, ft/sec.
= Chord for arc W, in.
= Arc length of breeching opening, in.
= Unit weight of stack shell, lb./in.3
= Outside diameter of belled stack base,
ft.
= Compression stress, ksi
= Frequency of the lowest mode of
ovaling vibration, cps
= Natural frequency, cps
= Acceleration of gravity, 386 in./sec.
= Height of stack bell, ft.
= Wind load, psf
= Dynamic wind pressure, psf
= Radius of gyration, in.
= Thickness of stack, in.

= Uniform load over breeching opening,

lb./in.

Static Design Criteria

In the suggested static design criteria below,
the proportions indicated are those desirable
from a structural standpoint. Independent
calculations are needed to determine sizes to
satisfy draft or capacity requirements. In
general, stacks proportioned as suggested will
probably have a high critical wind velocity,
but a dynamic check should be made to verify
this.
Short stacks (less than 100 ft. high) may be
straight cylinders without a belled base.
Minimum base
diameter
(3-1)
do = H / 10
Minimum bell height
h = 0.3H
(3-2)
Minimum diameter of
cylinder,
(3-3)
do = H / 13

Figure 3-1. Cylindrical Stack with Belled Base.

Stacks are likely to be subjected at least to
the following loads:
1. Metal Weight.
2. Lining Weight.
3. Wind: Wind load provisions may be found
in ASCE 7-95. Local building codes should
also be consulted.
4. Icing (if required).
5. Seismic (if required).
6. Thermal cycling (vertical and
circumferential).
7. Possible negative pressures.
8. Other requirements of local building codes.
Dynamic Wind Criteria
The dynamic influence of wind may be
approximated by assuming an equivalent
static force, FL, in pounds per foot of height,
acting in the direction of oscillations, given by:
FL = CL Do qcr /2
(3-4)
18

NOTE: = Critical damping factor which

varies from 1% for an unlined steel
stack of small diameter to 5% for
concrete.

Stack Stresses
The stresses associated with buckling have
four ranges into which they can fall depending
on the t/R ratio. They in turn may be affected
by the Euler effect or slenderness ratio
reduction factor. The stresses calculated in this
manner are not to be increased for wind or
earthquake stresses.

The dynamic wind pressure, qCR, in psf, is

given by: *qCR = 0.00119 VCR2. The critical wind
velocity, VCR2 in fps, for resonant transverse
vibration is given by:
VCR 2 ( ft / sec) =

f t Do
S

(3-5)

Fcr

The natural frequency, ft (cps), of vibration of

a stack of constant diameter and thickness is
given by:
3.52 D Eg
ft =

4H12 2Ws

(3-6)

Critical velocity for a steel stack with an S

value of 0.2 is given by:
Vcr1 (mph ) = 3.41Do f t

(3-7)

Do 2

f o Do
= ( ft / sec )
2S

Fy [0.8 + 5t /Ro]

0.01 t/Ro .04

Fy

t/Ro > .04

2 2 E
Fcr

FS = 2.0

(3-10)
2

KL / r
If Cc KL/r K = 1 0.5

C'c

Tables 3-1, 3-2 and 3-3 have been developed

using ASTM A36 steel with a yield of 36 ksi.
The value of K is taken as 2 in view of the fact
that a stack is normally a cantilever.
These allowable stresses will also be used
for tapered or belled base stacks using the
equivalent cylindrical radius approach as
shown below. In order to arrive at allowable
stresses in the conical section one would
substitute R1 into the above formulae for Ro.

(3-8)

Resonance occurs when frequency of the

lowest mode of ovaling vibration is twice the
vortex shedding frequency; thus, the critical
wind velocity for ovaling frequency is:
Vo =

Fy/11600 t/R, 0.01

C'c
If Cc < KL/r K = 0.5

KL / r

Critical Wind Velocity for Ovaling Vibrations

In addition to transverse swaying
oscillations, stacks experience flexural
vibration in the cross-sectional plan as a result
of vortex shedding. This frequency of the
lowest mode of ovaling vibration in a circular
shell is:
678.5t

Fy [0.35 + 50t / Ro]

Fc = KFcr/FS

* Reference number 14(b)

fo =

0.017 < t/Ro Fy/11600

C'c =

Values of effective diameters and effective

height for stacks of varying diameter and
thickness may be determined by methods
found in reference number 19.

t/Ro Range

5.8 x 103 t/Ro

(3-9)

Unlined stacks are subject to ovaling

vibrations. In order to prevent this
phenomenon, the thickness of the stack should
not be less than D/250 or intermediate
stiffeners are required to raise the resonant
velocity above 60 mph. Care should be
exercised in coastal areas to give special
attention to high winds as outlined in the
aforementioned ASCE 7-95.
Figure 3-2. Loads on Cylinder-Cone Junction
19

Cylinder-Cone Junction
In many applications of tubular columns, it
is desirable to use a base cone to provide a
broader base for anchorage. At the junction of
the cone and cylinder (Figure 3-2), it is
necessary to provide reinforcement to resist
the maximum vertical force.
V=

P
M
+
2 Ro Ro 2

resistance to negative draft. Spacing of

intermediate stiffener Ls is:
Ls = 60

and the radial thrust

The ring compression to be resisted is

C = HRo = VRo tan
The area of reinforcement required is
As =

VR o tan

(3-12)

Ss =

(3-13)

Is =

HRo
E

(3-18)

(in4)

(3-19)

(in2)

(3-20)

Breeching Opening
The breeching opening should be as small as
consistent with operating requirements with a
maximum width of 2Do/3.
The opening must be reinforced vertically to
replace the area of material removed increased
by the radio of Do/C. Therefore, each vertical
stiffener on each side of the opening should
have a cross-sectional area of:
As =

(3-15)

Wd Ls1 D
2C

(3-21)

Each vertical stiffener in conjunction with a

portion of the liner shell would be designed as
a column. Each stiffener should extend far
enough above and below the opening to
develop its strength.
Horizontal reinforcement should be
provided by a ring girder above and below the
opening. These girders should be designed as
fixed-end beams to carry the load across the
opening above and below. The span in
bending is the width W between the side
column, but the girders should encircle the
stack to preserve circularity at the opening. To
form each ring girder, stiffener rings should be
placed to act as top and bottom flanges. The
shell of the stack will serve as the web. Each
ring girder must be capable of carrying a
uniform distributed load, in terms of pounds
per inch of arch W, of:

based on a factor of safety of 3 for critical

buckling.
The area of reinforcement and computation
of Is provided by a stiffener may include an
area of cylinder and cone plate equal to:
0.78 t Rot + t1 R1t
(3-16)

(in3)

To satisfy the requirements of the above

intermediate stiffener design formulae a
portion of the stack equal to 1.1t Dot may be
included.

Under load, the junction reinforcement, or

stiffener, will move elastically inward. This
will induce secondary vertical bending stresses
on each side of the junction. For that reason, it
is desirable to keep allowable stress Fs
relatively low. If Fs is in the range of 8,000 psi,
the secondary stresses can usually be ignored if
Do is not greater than about 15 ft. For greater
diameters or higher values of Fs it would be
advisable to evaluate the secondary stresses.
Note that V is the maximum value resulting
from both vertical load and bending moment
in the cylinder at the junction level.
The moment of inertia Is of the stiffener
section should not be less than:
3

pLs1D 2
1100 Fb

Pd Ls1 D 3
8E
Pd Ls1 D
As =
2 Fa

(3-14)

Fs

(3-17)

To insure a nominal size of intermediate

stiffener, the spacing is limited within 1.5
times the stack diameter.
Intermediate stiffeners should meet the
following minimum requirements:

(3-11)

H = V tan

Do t
p

Where R1 = Ro /cos
This approach can be used in designing the
junction of two cones having different slopes,
except that H would be the difference between
the horizontal components of the axial loads in
the two cones.
Circumferential Stiffeners
A stiffener is required at the top of the stack;
also intermediate ring stiffeners are required to
prevent deformation of the stack shell under
wind pressure and to provide structural
20

w=

4M
V
+
Do Do 2

(3-22)

The bending moment in the girder is:

Mq =

mW '2
12

(3-23)

Allowable bending stresses may be chosen

using AISC rules.
Base Plates
In addition to bending stresses due to
bending loads, the stack base plate must resist
ring tension due to the horizontal component
of the base cone if one is used. Maximum ring
tension should be limited to 10,000 psi to
account for secondary bending stresses in the
base cone. This value may be varied upward
depending upon the extent of secondary stress
evaluation. Tension should be checked at the
minimum cross-section occurring at the anchor
bolt holes or at a weld joint where 85% or 100%
efficiency may be assumed.
A base plate area may be calculated by the
following equation:
VDo tan
A=
(3-24)

Figure 3-4.
Horizontal Section Through Opening.
(Section A-A, Figure 3-3)
Base plate thickness may be determined by
using AISC formulae and allowable bending
stresses.
Anchor Bolts
Minimum diameter = 1
Maximum spacing of anchor bolts = 5-6
Maximum tension at root of threads = 15,000
psi.
Each bolt should be made to resist a total
tension in pounds of:

20,000 E1

T =

4M V

ND N

(3-25)

A suggested design procedure for anchor

bolt brackets is covered in Part V.

Figure 3-3.
Elevation of Stack.
(See Figure 3-4 for Section A-A)

21

Table 3-1
Fc Allowable Compressive Stress (Fy = 36 ksi)
For t/Ro from .0017 through Fy/11600
KL/r
0
17.5
35
52.5
70
87.5
105
122.5
140
157.5
175

t/Ro

.0017

.00192

.00214

.00236

.00258

.0028

.00302

4930
4917
4878
4813
4722
4605
4462
4293
4097
3877
3630

5568
5551
5502
5419
5303
5154
4971
4755
4507
4225
3909

6206
6185
6124
6071
5876
5691
5414
5196
4887
4537
4145

6844
6819
6744
6618
6443
6217
5942
5616
5240
4814
4338

7482
7452
7362
7212
7003
6733
6404
6015
5565
5056
4487

8120
8085
7979
7803
7556
7238
6850
6392
5862
5263
4593

8758
8717
8594
8389
8101
7732
7281
6747
6132
5434
4655

Table 3-2
Fc Allowable Compressive Stress (Fy = 36 ksi)
For t / Ro from Fy /11600 to .01
KL/r
0
17.5
35
52.5
70
87.5
105
122.5
140
157.5
175

t/Ro

.003104

.00425

.0054

.00655

.0077

.00885

.00999

9094
9049
8917
8695
8386
7988
7501
6926
6262
5510
4670

10128
10073
9908
9634
9250
8756
8152
7439
6616
5683
4673

11162
11095
10895
10562
10095
9496
8762
7896
6896
5763
4673

12196
12116
11888
11480
10928
10207
9331
8297
7103
5769
4673

13230
13136
12855
12387
11732
10889
9859
8642
7237
5769
4673

14264
14155
13829
13284
12523
11543
10345
8930
7298
5769
4673

15298
15173
14797
14171
13295
12168
10791
9163
7301
5769
4673

Table 3-3.
Fc Allowable Compressive Stress (Fy = 36 ksi)
For t / Ro from .01 to .04
KL/r
0
17.5
35
52.5
70
87.5
105
122.5
140
157.5
175

t/Ro

.01

.015

.02

.025

.03

.035

.04

15300
15175
14798
14173
13296
12169
10792
9163
7302
5769
4673

15750
15617
15219
14556
13627
12432
10972
9247
7302
5769
4673

16200
16060
15638
14936
13954
12690
11146
9320
7302
5769
4673

16650
16502
16057
15315
14277
12942
11311
9383
7302
5769
4673

17100
16944
16474
15692
14597
13189
11468
9435
7302
5769
4673

17550
17385
16891
16067
14914
13431
11618
9476
7302
5769
4673

18000
17827
17307
16440
15227
13666
11760
9305
7302
5769
4673

If t/Ro > .04

Fc = .5 x Fy x K

Underlined numbers are an indicator at which point Cc > KL/r

22

14. F.B. Farquaharson, "Wind Forces Structures:

Structures
Subject
Oscillations,
Proceedings of ASCE, Vol. 84, ST 4, Paper
1712, 1958, p.13. ASCE Transaction Paper
#3269 ("Wind Forces on Structure").
15. C.F. Cowdrey and J.A. Lewes, "Drag
Measurements at High Reynolds Numbers
of a Circular Cylinder Fitted with Three
Helical Strakes, NPLIAeroI384, July 1959.
16. L. Woodgate and J. Maybrey, "Further
Experiments on the Use of Helical Strakes
for Avoiding Wind-Excited Oscillations of
Structures with Circular or Near Circular
Cross-Section," NPL/Aero/381, July 1959.
17. A. Roshko, "On the Wake and Drag Bluff
Bodies," presented at Aerodynamics
Sessions, Twenty- Second Annual Meeting,
IAS, New York, NY, January, 1954.
18. J.O. Smith and J.H. McCarthy, "Wind Versus
Tall Stacks," Mechanical Engineering, Vol.
87, January, 1965, pp. 38-41.
19. Gaylord
and
Gaylord,
"Structural
Engineering Handbook." 2nd Edition,
Chapter 26.
20. R. Stuart III, A.R. Fugini, A. DeVaul,
Pittsburgh-Des Moines Corp. Research
Report #98528, "Design of Allowable
Compressive Stress Cylindrical or Conical
Plates, AWWA DlOO," May 1981.
21. Roger L. Brockenbrough, Pittsburgh-Des
Moines Corp. Research Report 98030,
"Determination of The Critical Buckling
Stress of Cylindrical Plates Having Low t/R
Values," October 5, 1960.
22. Tom Buckwalter, Pittsburgh-Des Moines
Corp.
Supplement
to
RP
98030,
"Determination of the Critical Buckling
Stress in a Cylinder Having a t/R of
0.00426," December 20, 1960.
23. AlSC "Specification for Structural Steel
Buildings Allowable Stress Design and
Plastic Design."
24. ASCE 7 Minimum Design Loads for
Buildings and Other Structures.

References
1. M.S. Ozker and J.O. Smith, "Factors
Influencing the Dynamic Behavior of Tall
Stacks Under the Action of Winds, Trans.
ASME Vol. 78, 1956, pp. 1381-1391.
2. P. Price, "Suppression of the Fluid-Induced
Vibration
of
Circular
Cylinders,"
Proceedings of ASCE, Vol. 82, EM3, Paper
No. 1030, 1956, p. 22.
3. W.L. Dickey and G.B. Woodruff, "The
Vibration of Steel Stacks, Proceedings of
ASCE, Vol. 80, 1954, p. 20.
4. T. Sarpkaya and C.J. Garison, "Vortex
Formation and Resistance in Unsteady
Flow, Journal of Applied Mechanics, Vol.
30, Trans. ASME, Vol.85, Series E, 1963, pp.
16-24.
5. A.W. Marris, "A Review on Vortex Streets,
Periodic Wakes, and Induced Vibration
Phenomena,"
Journal
of
Basic
Engineering,Trans. ASME, Series D, Vol. 86,
1964, pp. 185-l 96.
6. J. Penzien, "Wind Induced Vibration of
Cylindrical Structures, Proceedings of
ASCE,Vol. 83, EM 1 Paper No. 1141,
January, 1957, p. 17.
7. W. Weaver,"Wind-Induced Vibrations in
Antenna Members, Transactions of ASCE,
Vol. 127, Part 1, 1962, pp. 679-704.
8. C. Scruton and D. Walshe, "A Means of
Avoiding Wind-Excited Oscillations of
Structures with Circular or Nearly Circular
Cross-Section," NPL/Aero/335, October
1957.
9. C. Scruton, D. Walshe and L. Woodgate,
"The Aerodynamic Investigation for the East
Chimney Stack of the Rugeley Generating
Station," NPL/Aero/352.
10. A. Roshko, "On the Development of
Turbulent Wakes from Vortex Streets,
NACA Report 1191, 1954.
11. A. Roshko, On The Drag and Shedding
Frequency of Two-Dimensional Bluff
Bodies, NACA Technical Note 3169, July
1954.
12. N. Delany and N. Sorensen, Low-Speed
Drag of Cylinders of Various Shapes, NCA
Technical Note 3038, November, 1953.
13. G.B. Woodruff and J. Kozok, "Wind Forces
on
Structures:
Fundamental
Considerations," Proceedings of ASCE, Vol.
84, ST 4, Paper No. 1709, 1958, p. 13.
23

This page has been left blank intentionally.

Part IV
Supports for Horizontal Tanks and Pipe Lines

the saddle, which result in a different

distribution of stress in the pipe or vessel wall
from that encountered with a full ring support,
are discussed in the following paper by L.P.
Zick. It includes some revisions of and
additions to the original paper published in
The Welding Journal Research Supplement,
September, 1951, and reprinted in Pressure
Vessel and Piping Design Collected Papers
1927-1959, published by ASME in 1960.

here is considerable information

available on design of supports for
horizontal cylindrical shells where a
complete ring girder is used.
There are many installations where a
horizontal tank, pressure vessel, or pipeline is
supported by a saddle extending less than 180
around the lower part of the cylinder. The
effects of vertical deflection of the cylinder and
the concentration of stress around the horn of

Original paper published in September 1951 THE WELDING JOURNAL RESEARCH SUPPLEMENT. This paper contains
revisions and additions to the original paper based upon questions raised as to intent and coverage.

Stresses in Large Horizontal Cylindrical

Pressure Vessels on Two Saddle Supports
Approximate stresses that exist in cylindrical vessels supported on two saddles at various conditions
and design of stiffening for vessels which require it.
By L.P. Zick*
pressure vessels made of mild steel for storage
of liquid weighing 42 lb. per cu. ft

INTRODUCTION
The design of horizontal cylindrical vessels
with dished heads to resist internal pressure is
covered by existing codes. However, the
method of support is left pretty much up to the
designer. In general the cylindrical shell is
made a uniform thickness which is determined
by the maximum circumferential stress due to
the internal pressure. Since the longitudinal
stress is only one-half of this circumferential
stress, these vessels have available a beam
strength which makes the two-saddle support
system ideal for a wide range of proportions.
However certain limitations are necessary to
make designs consistent with the intent of the
code.
The purpose of this paper is to indicate the
approximate stresses that exist in cylindrical
vessels supported on two saddles at various
locations. Knowing these stresses, it is possible
to determine which vessels may be designed for
internal pressure alone, and to design
structurally adequate and economical stiffening
for the vessels which require it. Formulas are
developed to cover various conditions, and a
chart is given which covers support designs for

HISTORY
In a paper1 published in 1933, Herman
Schorer pointed out that a length of cylindrical
shell supported by tangential end shears
varying proportionately to the sine of the
central angle measured from the top of the
vessel can support its own metal weight and the
full
contained
liquid
weight
without
circumferential bending moments in the shell.
To complete this analysis, rings around the
entire circumference are required at the
supporting points to transfer these shears to the
foundation without distorting the cylindrical
shell. Discussions of Schorers paper by H.C.
Boardman and others gave approximate
solutions for the half-full condition. When a
ring of uniform cross-section is supported on
two vertical posts, the full condition governs
the design of the ring if the central angle
* L.P. Zick is a former Chief Engineer for the
Chicago Bridge and Iron Co., Oak Brook,
Illinois.
25

Figure 4-1. Strain gauge test set up on 30,000 gal. propane tank.
............................................................................

had actually slumped down over the horns of

the saddles while being filled with liquid, but
had rounded up again when internal pressure
was applied.
Testing done by others,2,3 gave very useful
results in the ranges of their respective tests,
but the investigators concluded that analysis
was highly indeterminate. In recent years the
author has participated in strain gauge
surveys of several large vessels.4 A typical test
setup is shown in Figure 4-1.
In this paper an attempt has been made to
produce an approximate analysis involving
certain empirical assumptions which make the
theoretical analysis closely approximate the
test results.

between the post intersections with the ring is

less than 126, and the half-full condition
governs if this angle is more than 126.
However, the full condition governs the
design of rings supported directly in or
adjacent to saddles.
Mr. Boardmans discussion also pointed out
that the heads may substitute for the rings
provided the supports are near the heads. His
unpublished paper has been used successfully
since 1941 for vessels supported on saddles
near the heads. His method of analysis
covering supports near the heads is included
in this paper in a slightly modified form.
Discussions of Mr. Schorers paper also gave
successful and semi-successful examples of
unstiffened cylindrical shells supported on
saddles, but an analysis is lacking. The semisuccessful examples indicated that the shells

Table 4-1 Values of Coefficients in Formula for Various Support Conditions

Saddle
Angle,

Maximum long.
bending stress,
Min. K1

Tangent
shear,
K2

120
150

0.63 (A/L = 0.11)

0.55 (A/L = 0.11)

1.171
0.799

120
150

1.0 (A/L = 0)
1.0 (A/L = 0)

0.880
0.485

120
150

0.23 (A/L = 0.193)

0.23 (A/L = 0.193)

0.319
0.319

120
150

0.23 (A/L = 0.193)

0.23 (A/L = 0.193)

1.171
0.799

Circumf.
stress top
of saddle,
K3 j

Additional
Ring
head
compres.
stress,
in shell,
K4
K5
Shell unstiffened
0.0528

0.760
0.0316

0.673
Shell stiffened by head, A R/2
0.0132
0.401
0.760
0.0079
0.297
0.673
Shell stiffened by ring in plane of saddle

Shell stiffened by rings adjacent of saddle

0.0132

0.760
0.0079

0.673

Ring Stiffeners
Circumf.
Direct
bending,
stress,
K6
K7

0.204
0.260

0.204
0.260

0.0528
0.0316

0.340
0.303

0.204
0.260

0.0577
0.0353

0.263
0.228

0.204
0.260

*See Figure 4-5, which plots K1 against A/L, for values of K1, corresponding to values of A/L not listed in table.
jSee Figure 4-7

26

Tension
across
saddle,
K8

Figure 4-2. Location and type of support for horizontal pressure vessels on two supports.

27

liquids usually associated with pressure

vessels.
Where liquids of different weights are to be
stored or where different materials are to be
used, a rough design may be obtained from
the chart and this design should be checked by
the applicable formulas outlined in the
following sections. Table 4-1 outlines the
coefficients to be used with the applicable
formulas for various support types and
locations. The notation used is listed at the end
of the paper under the heading Nomenclature.

SELECTION OF SUPPORTS
When a cylindrical vessel acts as its own
carrying beam across two symmetrically
placed saddle supports, one-half of the total
load will be carried by each support. This
would be true even if one support should
settle more than the other. This would also be
true if a differential in temperature or if the
axial restraint of the supports should cause the
vessel acting as a beam to bow up or down at
the center. This fact alone gives the twosupport system preference over a multiplesupporting system.
The most economical location and type of
support generally depend upon the strength of
the vessel to be supported and the cost of the
supports, or of the supports and additional
stiffening, if required. In a few cases the
advantage of placing fittings and piping in the
bottom of the vessel beyond the saddle will
govern the location of the saddle.
The pressure-vessel codes limit the contact
angle of each saddle to a minimum of 120
except for very small vessels. In certain cases a
larger contact angle should be used. Generally
the saddle width is not a controlling factor, so
a nominal width of 12 in. for steel or 15 in. for
concrete may be used. This width should be
increased for extremely heavy vessels, and in
certain cases it may be desirable to reduce this
width for small vessels.
Thin-wall vessels of large diameter are best
supported near the heads provided they can
support their own weight and contents
between supports and provided the heads are
stiff enough to transfer the load to the saddles.
Thick-wall vessels too long to act as simple
beams are best supported where the maximum
longitudinal bending stress in the shell at the
saddles is nearly equal to the maximum
longitudinal bending stress at mid-span,
provided the shell is stiff enough to resist this
bending and to transfer the load to the saddles.
Where the stiffness required is not available in
the shell alone, ring stiffeners must be added
at or near the saddles. Vessels must also be
rigid enough to support normal external loads
such as wind.
Figure 4-2 indicates the most economical
locations and types of supports for large steel
horizontal pressure vessels on two supports. A
liquid weight of 42 lb. per cu. ft. was used
because it is representative of the volatile

MAXIMUM LONGITUDINAL STRESS

The cylindrical shell acts as a beam over the
two supports to resist by bending the uniform
load of the vessel and its contents. The
equivalent length of the vessel (see Figures 4-2
and 4-3) equals L + 4H/3, closely, and the total
weight of the vessel and its contents equals 2Q.
However, it can be shown that the liquid
weight in a hemispherical head adds only a
shear load at its junction with the cylinder.
This can be approximated for heads where H
R by representing the pressure on the head
and the longitudinal stress as a clockwise
couple on the head shown at the left of Figure
4-3. Therefore the vessel may be taken as a
beam loaded as shown in Figure 4-3; the
moment diagram determined by statics is also
shown. Maximum moments occur at the
midspan and over the supports.
Tests have shown that, except near the
saddles, a cylindrical shell just full of liquid
has practically no circumferential bending
moments and therefore behaves as a beam
with a section modulus I/c = r2t.
However, in the region above each saddle,
circumferential
bending
moments
are
introduced, allowing the unstiffened upper
portion of the shell to deflect, thus making it
ineffective as a beam.

28

Figure 4-3. Cylindrical shell acting as beam over supports.

Figure 4-4. Load transfer to saddle by tangential shear stresses in cylindrical shell.
29

This reduces the effective cross section acting

as a beam just as though the shell were split
along a horizontal line at a level above the
saddle [See Figure 4-4 (a)]. If this effective arc
is represented by 2A (A in radians) it can be
shown that the section modulus becomes:
2

+ sin cos 2 sin

I / c = r 2 t
sin

cos

Strain gauge studies indicate that this

effective arc is approximately equal to the
contact angle plus one-sixth of the unstiffened
shell as indicated in Section A-A of Figure 4-4.
Of course, if the shell is stiffened by a head or
complete ring stiffener near the saddle the
effective arc, 2, equals the entire cross section,
and I/c = r 2 t.
Since most vessels are of uniform shell
thickness, the design formula involves only the
maximum value of the longitudinal bending
stress. Dividing the maximum moment by the
section modulus gives the maximum axial
stress in lb. per sq. in. in the shell due to
bending as a beam, or:

S1 =

Figure 4-5. Plot of longitudinal bendingmoment constant, K1.

This maximum bending stress, S1, may be
either tension or compression. The tension
stress when combined with the axial stress due
to internal pressure should not exceed the
allowable tension stress of the material times
the efficiency of the girth joints. The
compression stress should not exceed one half
of the compression yield point of the material
or the value given by:

3K1 QL

E
S1 (t / r )[2 (2 / 3) (100) (t / r )]
29

r2 t

K1 is a constant for a given set of conditions,

but actually varies with the ratios A/L and H/L
R/L for different saddle angles. For
convenience, K1 is plotted in Figure 4-5 against
A/L for various types of saddle supports,
assuming conservative values of H = 0 when
the mid- span governs and H = R when the
shell section at the saddle governs. A
maximum value of R/L = 0.09 was assumed
because other factors govern the design for
larger values of this ratio. As in a beam the
mid-span governs for the smaller values of A/L
and the shell section at the saddle governs for
the larger values of A/L; however, the point
where the bending stress in the shell is equal at
mid-span and at the saddle varies with the
saddle angle because of the reduced effective
cross section. Figure 4-10 on page 39 gives
acceptable values of K1.

which is based upon the accepted formula for

buckling of short steel cylindrical columns.*
The compression stress is not a factor in a steel
vessel where t/r >0.005 and the vessel is
designed to be fully stressed under internal
pressure.
*See also par UG-23(b) ASME Code Section
VIII Div. 1.
TANGENTIAL SHEAR STRESS
Figure 4-4(d) shows the total shear diagram
for vessels supported in saddles away from the
heads.
Where the shell is held round, the tangential
shearing stresses vary directly with the sine of
the central angle , as shown in Section B-B of
Figure 4-4, and the maximum occurs at the
equator.
However, if the shell is free to deform above
the saddle, the tangential shearing stresses act
on a reduced effective cross section and the
maximum occurs at the horn of the saddle.
This is approximated by assuming the shears
30

continue to vary as the sin but only act on

twice the arc given by(/2 + (/20) or (-) as
shown in Section A-A of Figure 4-4. The
summation of the vertical components of these
assumed shears must equal the maximum total
shear.
The maximum tangential shear stress will
occur on the center side of the saddle provided
the saddle is beyond the influence of the head
but not past the quarter point of the vessel.
Then with saddles away from the heads, the
maximum shear stress in lb. per sq. in. is given
by:

CIRCUMFERENTIAL STRESS AT HORN OF

SADDLE
In the plane of the saddle the load must be
transferred from the cylindrical shell to the
saddle. As was pointed out in the previous
section the tangential shears adjust their
distribution in order to make this transfer with
a minimum amount of circumferential
bending and distortion. The evaluation of
these shears was quite empirical except for the
case of the ring stiffener in the plane of the
saddle. Evaluation of the circumferential
bending stresses is even more difficult.
Starting with a ring in the plane of the
saddle, the shear distribution is known. The
bending moment at spy point above the saddle
may be computed by any of the methods of
indeterminate structures.

K Q L 2A
S 2 = 2
rt L + 4 H
3

values of K2 listed in Table 4-1 for various

types of supports are obtained from the
expressions given for the maximum shears in
Figure 4-4, and the Appendix.
Figure 4-4(f) indicates the total shear
diagram for vessels supported on saddles near
the heads. In this case the head stiffens the
shell in the region of the saddle. This causes
most of the tangential shearing stress to be
carried across the saddle to the head, and then
the load is transferred back to the head side of
the saddle by tangential shearing stresses
applied to an arc slightly larger than the
contact angle of the saddle. Section C-C of
Figure 4-4 indicates this shear distribution;
that is, the shears vary as the sin and act
downward above angle and act upward
below angle . The summation of the
downward vertical components must balance
the summation of the upward vertical
components. Then with saddles at the heads
the maximum shear stress in lb. per sq. in. is
given by:
S2 =

K 2Q
rt

S2 =

K 2Q
rt h

Figure 4-6. Circumferential bending-moment

diagram, ring in plane of saddle.
If the ring is assumed uniform in cross
section and fixed at the horns of the saddles,
the moment, M, in in.-lb. at any point A is
given by:

3 sin cos 1

cos + sin
2
2
2
4

Qr
sin
M =
+ 2 cos2
4 6

9
2
sin
sin


cos + 1 2

in the shell, or

in the head.
Values of K2 given in Table 4-1 for different
size saddles at the heads are obtained from the
expression given for the maximum shear stress
in Section C-C of Figure 4-4 and the Appendix.
The tangential shear stress should not
exceed 0.8 of the allowable tension stress.

sin
cos

This is shown schematically in Figure 4-6.

Note that must be in radians in the formula.
The maximum moment occurs when = .
Substituting for and K6 for the expression
in the brackets divided by , the maximum
circumferential bending moment in in.-lb. is:
M= K6 Qr
31

When the shell is supported on a saddle and

there is no ring stiffener the shears tend to
bunch up near the horn of the saddle, so that
the actual maximum circumferential bending
moment in the shell is considerably less than
M, as calculated above for a ring stiffener in
the plane of the saddle. The exact analysis is
not known; however, stresses calculated on the
assumption that a wide width of shell is
effective in resisting the hypothetical moment,
M, agree conservatively with the results of
strain gauge surveys.

assumption that this moment is divided by

four when A/R is less than 0.5.
The change in shear distribution also
reduces the direct load at the horns of the
saddle; this is assumed to be Q/4 for shells
without added stiffeners. However, since this
load exists, the effective width of the shell
which resists this direct load is limited to that
portion which is stiffened by the contact of the
saddle. It is assumed that 5t each side of the
saddle acts with the portion directly over the
saddle. See Appendix B.
Internal pressure stresses do not add directly
to the local bending stresses, because the shell
rounds up under pressure. Therefore the
maximum circumferential combined stress in
the shell is compressive, occurs at the horn of
the saddle, and is due to local bending and
direct stress. This maximum combined stress
in lb. per sq. in. is given by:
S3 =

3K 3 Q
Q

if L 8R
4t (b +10t )
2t 2

S3 =

Q
12 K3 QR

if L * < 8 R
4t (b + 10t )
Lt2

or

* Note: For multiple supports:

L = Twice the length of portion of shell carried
by saddle.
If L 8R use 1st formula.

Figure 4-7. Plot of circumferential bendingmoment constant, K3.

It was found that this effective width of shell
should be equal to 4 times the shell radius or
equal to one-half the length of the vessel,
whichever is smaller. It should be kept in mind
that use of this seemingly excessive width of
shell is an artifice whereby the hypothetical
moment M, is made to render calculated
stresses in reasonable accord with actual
stresses.
When the saddles are near the heads, the
shears carry to the head and are then
transferred back to the saddle. Again the
shears tend to concentrate near the horn of the
saddle. Because of the relatively short stiff
members
this
transfer
reduces
the
circumferential bending moment still more.
To introduce the effect of the head the
maximum moment is taken as:

It seems reasonable to allow this combined

stress to be equal to 1.50 times the tension
allowable provided the compressive strength
of the material equals the tensile strength. In
the first place when the region at the horn of
the saddle yields, it acts as a hinge, and the
upper portion of the shell continues to resist
the loads as a two-hinged arch. There would
be little distortion until a second point near the
equator started to yield. Secondly, if rings are
added to reduce this local stress, a local
longitudinal bending stress occurs at the edge
of the ring under pressure.5 This local stress
would be 1.8 times the design ring stress if the
rings were infinitely rigid. Weld seams in the
shell should not be located near the horn of the
saddle where the maximum moment occurs.
EXTERNAL LOADS
Long vessels with very small f/r values are
susceptible to distortion from unsymmetrical
external loads such as wind. It is assumed that
vacuum relief valves will be provided where
required, so it is not necessary to design

M = K3 Qr
where K3 equals Ks when A/R is greater than 1.
Values of K3 are plotted in Figure 4-7 using the
32

combined stress be allowed to be 25% greater

than the allowable tension stress because of
the nature of the stress and because of the
method of analysis.

against a full vacuum. However, experience

indicates that vessels designed to withstand 1
lb. per sq. in. external pressure can
successfully resist external loads encountered
in normal service.
Assume the external pressure is 1 lb. per sq.
in. in the formulas used to determine the
sloping portion of the external pressure chart
in the current A.S.M.E. Unfired Pressure
Vessel Code. Then when the vessel is
unstiffened between the heads, the maximum
length in feet between stiffeners (the heads) is
given approximately by:
L + 2 / 3H =

E rt t

52.2 r

Figure 4-8. Loads and reactions on saddles.

When ring stiffeners are added to the vessel

at the supports, the maximum length in feet
between stiffeners is given by:
E rt t
L 2A=

52.2 r

RING COMPRESSION IN SHELL OVER SADDLE

Figure 4-8 indicates the saddle reactions,
assuming the surfaces of the shell and saddle
are in frictionless contact without attachment.
The sum of the assumed tangential shears on
both edges of the saddle at any point A is also
shown in Figure 4-8. These forces acting on the
shell band directly over the saddle cause ring
compression in the shell band. Since the saddle
reactions are radial, they pass through the
center O. Taking moments about point O
indicates that the ring compression at any
point A is given by the summation of the
tangential shears between and .
This ring compression is maximum at the
bottom, where
= . Again, a width of shell
equal to 5t each side of the saddle plus the
width of the saddle is assumed to resist this
force. See Appendix B. Then the stress in lb.
per sq. in. due to ring compression is given by:

ADDITIONAL STRESS IN HEAD USED AS A

STIFFENER
When the head stiffness is utilized by
placing the saddle close to the heads, the
tangential shear stresses cause an additional
stress in the head which is additive to the
pressure stress. Referring to Section C-C of
Figure 4-4, it can be seen that the tangential
shearing stresses have horizontal components
which would cause varying horizontal tension
stresses across the entire height of the head if
the head were a flat disk. The real action in a
dished head would be a combination of ring
action and direct stress; however, for
simplicity the action on a flat disk is
considered reasonable for design purposes.
Assume that the summation of the
horizontal components of the tangential shears
is resisted by the vertical cross section of the
flat head at the center line, and assume that the
maximum stress is 1.5 times the average stress.
Then the maximum additional stress in the
head in lb. per sq. in. is given by:
3Q
sin 2
S4 =

8r th + sin cos
S4 =

S5 =

Q
1 + cos

t (b + 10t ) + sin cos

or
S5 =

K5 Q
t (b +10t )

The ring compression stress should not

exceed one-half of the compression yield point
of the material.

WEAR PLATES
The stress may be reduced by attaching a
wear plate somewhat larger than the surface of
the saddle to the shell directly over the saddle.
The thickness t used in the formulas for the
assumed cylindrical shell thickness may be

K4 Q
r th

This stress should be combined with the

stress in the head due to internal pressure.
However, it is recommended that this
33

taken as (t1 + t2) for S5 (where t1 = shell

thickness and t 2 = wear plate thickness),
provided the width of the added plate equals
at least (b + 10 t1) (see Appendix B).
The thickness t may be taken as (t1 + t2) in
the formula for S2 provided the plate extends
r/10 inches above the horn of the saddle near
the head, and provided the plate extends
between the saddle and an adjacent stiffener
ring. (Also check for S 2 stress in the shell at the
equator.)
The thickness t may be taken as (t1 + t2) in
the first term of the formula for S 3 , provided
the plate extends r/10 inches above the horn of
the saddle near the head. However, (t 1 2 + t 2 2 )
should be substituted for t 2 in the second term.
The combined circumferential stress ( S 3 ) at
the top edge of the wear plate should also be
checked using the shell plate thickness t 1 and
the width of the wear plate. When checking at
this point, the value of K 3 should be reduced
by extrapolation in Figure 4-7 assuming
equal to the central angle of the wear plate but
not more than the saddle angle plus 12.

found by statics. Then the direct load at the

horn of the saddle is given in pounds by
n P =

or

P = K 7

Q
n

If n stiffeners are added adjacent to the

saddle as shown in Figure 4-4 (b), the rings
will act together and each will be loaded with
shears distributed as in Section B-B on one side
but will be supported on the saddle side by a
shear distribution similar to that shown in
Section A-A. Conservatively, the support may
be assumed to be tangential and concentrated
at the horn of the saddle.

DESIGN OF RING STIFFENERS

When the saddles must be located away
from the heads and when the shell alone
cannot resist the circumferential bending, ring
stiffeners should be added at or near the
supports. Because the size of rings involved
does not warrant further refinement, the
formulas developed in this paper assume that
the added rings are continuous with a uniform
cross-section. The ring stiffener must be
attached to the shell, and the portion of the
shell reinforced by the stiffener plus a width of
shell equal to 5t each side may be assumed to
act with each stiffener. The ring radius is
assumed equal to r.
When n stiffeners are added directly over
the saddle as shown in Figure 4-4 ( c ) , the
tangential shear distribution is known. The
equation for the resulting bending moment at
any point was developed previously, and the
resulting moment diagram is shown in Figure
4-6. The maximum moment occurs at the horn
of the saddle and is given in in.-lb. for each
stiffener by:
M = K6

Q sin
cos
cos +
M M1

2 (1 cos )
r (1 cos )

Figure 4-9. Circumferential bending-moment

diagram stiffeners adjacent to saddle.
This is shown schematically in Figure 4-9;
the resulting bending moment diagram is also
indicated. This bending moment in in.-lb at
any point A above the horn of the saddle is
given by:
M =

Q r
sin cos [3 / 2 + ( ) cot ]

2 n sin

For the range of saddle angles considered,

M is maximum near the equator where = .
This moment and the direct stress may be
found using a procedure similar to that used
for the stiffener in the plane of the saddle.
Substituting for and K6 for the expression
in the brackets divided by 2, the maximum
moment in each ring adjacent to the saddle is
given in in.-lb. by:

Qr
n

Knowing the maximum moment M and the

moment at the top of the vessel, Mt, the direct
load at the point of maximum moment may be

M = K 6

34

Qr
n

Knowing the moments M and Mt, the

direct load at may be found by statics and is
given by:
P =

cos
Q sin
cos
M + Mt

(
r
cos )
n 2(1 cos )
1

circumferential bending in the shell at the horn

of the saddle has been neglected.
The upper and lower flanges of a steel
saddle should be designed to resist bending
over the web(s), and the web(s) should be
stiffened according to the AISC. Specifications
against buckling. The contact area between the
shell and concrete saddle or between the metal
saddle and the concrete foundation should be
adequate to support the bearing loads.
Where extreme movements are anticipated
or where the saddles are welded to the shell,
bearings or rockers should be provided at one
saddle. Under normal conditions a sheet of
elastic waterproof material at least in. thick
between the shell and a concrete saddle will
suffice.

Then the maximum combined stress due to

liquid load in each ring used to stiffen the shell
at or near the saddle is given in lb. per sq. in.
by:
S6 =

K 7Q K 6QR

na
nl / c

where a = the area and l/c = the section

modulus of the cross section of the composite
ring stiffener. When a ring is attached to the
inside surface of the shell directly over the
saddle or to the outside surface of the shell
adjacent to the saddle, the maximum
combined stress is compression at the shell.
However, if the ring is attached to the opposite
surface, the maximum combined stress may be
either compression in the outer flange due to
liquid or tension at the shell due to liquid and
internal pressure.
The maximum combined compression stress
due to liquid should not exceed one-half of the
compression yield point of the material. The
maximum combined tension stress due to
liquid and pressure should not exceed the
allowable tension stress of the material.

Nomenclature
Q = load on one saddle, lb. Total load = 2Q.
L = tangent length of the vessel, ft.
A = distance from center line of saddle to
tangent line, ft.
H = depth of head, ft.
R = radius of cylindrical shell, ft.
r = radius of cylindrical shell, in.
t = thickness of cylindrical shell, in.
th = thickness of head, in.
b = width of saddle, in.
F = force across bottom of saddle, lb.
S1, S2, etc. = calculated stresses, lb. per sq. in.
constants
for
K1, K2, etc. = dimensionless
various support conditions
M, M, etc. = circumferential bending moment
due to tangential shears, in.-lb.
= angle of contact of saddle with shell,
degrees.
= (180 - /2) = central angle from vertical
to horn of saddle, in degrees (except as
noted)
= /180 (/2 + /6) = /180 (50/12 + 30)
2 = arc, in radians, of unstiffened shell
in plane of saddle effective against
bending.
= - /180 (/2 + /20) = the central angle,
in radians, from the vertical to the
assumed point of maximum shear in
unstiffened shell at saddle.
= any central angle measured from the
vertical, in radians.

DESIGN OF SADDLES
Each saddle should be rigid enough to
prevent the separation of the horns of the
saddle; therefore, the saddle should be
designed for a full water load. The horn of the
saddle should be taken at the intersection of
the outer edge of the web with the top flange
of a steel saddle. The minimum section at the
low point of either a steel or concrete saddle
must resist a total force, F, in pounds, equal to
the summation of the horizontal components
of the reactions on one-half of the saddle.
Then:
1+ cos 1 / 2 sin 2
F =Q
= K8Q
+ sin cos

the effective section resisting this load should

be limited to the metal cross section within a
distance equal to r/3 below the shell. This
cross section should be limited to the
reinforcing steel within the distance r/3 in
concrete saddles. The average stress should
not exceed two-thirds of the tension allowable
of the material. A low allowable stress is
recommended because the effect of the
35

= central angle from the upper vertical to

the point of maximum moment in ring
located adjacent to saddle, in radians.
= modulus of elasticity of material, lb. per
sq. in.
l/c = section modulus, in.3
n = number of stiffeners at each saddle.
a = cross-sectional area of each composite
stiffener, sq. in.
P, P = the direct load in lb. at the point of
maximum moment in a stiffening
ring.

any central angle measured from the bottom,

the moment of inertia is:

sin sin 2
+
2r 3 t cos 2 2 cos
d =

2 sin sin sin 2

+
=
2r 3t 1 sin cos +
2

2
2

sin 2
r 3t sin cos + 2

The section modulus for the tension side of

the equivalent beam is:

sin 2

+ sin cos 2

r 2t
sin

cos

Bibliography
1. Schorer, Herman, "Design of Large Pipe
Lines," A.S.C.E. Trans., 98, 101 (1933) and
discussions of this paper by Boardman,
H.C., and others.
2. Wilson, Wilbur M., and Olson, Emery D.,
"Test of Cylindrical Shells," Univ. 111.
Bull. No. 331.
3. Hartenberg, R.S., "The Strength and
Stiffness of Thin Cylindrical Shells on
Saddle Supports," Doctorate Thesis,
University of Wisconsin, 1941.
4. Zick, L.P., and Carlson, C.E., Strain
Gauge Technique Employed in Studying
Propane Tank Stresses Under Service
Conditions, Steel, 86-88 (Apr. 12, 1948).
5. U.S. Bureau of Reclamation, Penstock
Analysis and Stiffener Design, Boulder
Canyon Project Final Reports, Part V.
Technical Investigations, Bulletin 5.

Then the stress in the shell at the saddle in

lb. per sq. in. is given by:
2
2

1 A + R H

3Q L 4 A
2AL
L
S1 = 2
1
4H
r t L
1+

3L

or
S1 =

3K1QL

r 2t

where
2
2

sin

cos
1 A + R H

4
A

L
2 AL

K1 =
1
4H

sin 2 L
+
1

+
sin

cos

3L

The bending moment in ft.-lb. at the mid-span

is:

APPENDIX A
The formulas developed by outline in the
text are developed mathematically here under
headings corresponding to those of the text.
The pertinent assumptions and statements
appearing in the text have not been repeated.

2
2

1+ 2 R H

2
A
2Q (L 2 A)2 2 HA A2 R 2 H 2 QL

L
4

4H
4H
L
8
3
2
4

4 1+
L+

3
3L

The section modulus = r2 t, and

S1 =

Maximum Longitudinal Stress

Referring to Figure 4-3, the bending moment
in ft.-lb. at the saddle is:

A R2 H 2
1 +

2
2
2

2Q
2 HA A
R H
2 AL
L
+

= QA 1
4H 3
4H

2
4

1
L+
+

3
3L

sin

cos

2
+ sin cos 2 sin

3K1 Q L

r 2t

where

2
2

1+ 2 R H

2
A

L
K1 =
4
4H
L
1+

3L

Referring to Section A-A of Figure 4-4 the

centroid of the effective arc = r sin . If equals

36

Tangential Shear Stress

Section B-B of Figure 4-4 indicates the plot of
the shears adjacent to a stiffener. The
summation of the vertical components of the
shears on each side of the stiffener must equal
the load on the saddle Q. Referring to Figure
4-4 (d), the sum of the shears on both sides of
the stiffener at any point is Q sin /r. Then
the summation of the vertical components is
given by:
2 o

Circumferential Stress at Horn of Saddle

See under the heading Design of Ring
Stiffeners.
Additional Stress In Head Used as Stiffener
Referring to Section C-C of Figure 4-4, the
tangential shears have horizontal components
which cause tension across the head. The
summation of these components on the
vertical axis is:

Q sin 2
2Q sin cos
r d =

=Q
r
2
2
o

the maximum shear stress occurs at the

equator when sin = 1 and K2 = l/ = 0.319
Section A-A of Figure 4-4 indicates the plot
of the shears in an unstiffened shell. Again this
summation of the vertical components of the
shears on each side of the saddle must equal
the load on the saddle. Then the total shear at
any point is:

S4 =

K2 =

the maximum shear occurs where = and

sin 2

+ sin cos

Q sin 2 sin cos

r + sin cos

Q sin 1
r d1 =
r

r d 2

Q sin 2 d 2
=
( + sin cos )

[cos 2 ] =

sin

cos

Q sin 2 sin cos

Q sin 2 1
r d1 = 2
r d

r
r + sin cos

cos + cos
Q

+ sin cos

or

2Q 1 sin1 cos 1

=
2
2
o

the ring compression becomes a maximum in

the shell at the bottom of the saddle. Or if =
this expression becomes:

2Q sin cos 2 sin 2 cos 2

+ sin cos 2
2

1+ cos
Q

sin

cos

Finally:

( sin cos )= Q ( sin cos )

then

The maximum shear occurs when 2 = and

K2 =

3
8

Wear Plates
The ring compression at any point in the
shell over the saddle is given by the
summation of the tangential shears over the
arc = ( - ) shown in Section A-A or C-C of
Figure 4-4 or in Figure 4-8. Then

sin
+ sin cos

Section C-C of Figure 4-4 indicates the shear

transfer across the saddle to the head and back
to the head side of the saddle. Here the
summation of the vertical components of the
shears on arc acting downward must equal
the summation of the vertical component of
the shears on the lower arc ( - ) acting
upward. Then

K 4Q
rth

where

Then assuming this load is resisted by 2r th

and that the maximum stress is 1.5 times the
average:

sin cos
Q sin 2
r d = Q
=Q
r ( + sin cos )
+ sin cos

2 o

sin cos
sin 2 cos 2
r d 2
+ sin cos

sin 2
Q

2 + sin cos

and the summation of the vertical components

is given by:

K2 =

sin 1 cos 1 r d 2

sin cos sin 2

Q sin 2 1

=

r 2
+ sin cos r
o

Q sin
r ( + sin cos )

sin sin cos

+ sin cos

37

1+ cos
K5 =

sin

cos

Design of Ring Stiffeners;

Stiffener in Plane of Saddle
Referring to Figure 4-6, the arch above the
horns of the saddle resists the tangential shear
load. Assuming this arch fixed at the top of the
saddles, the bending moment may be found
using column analogy.
If the arch is cut at the top, the static moment
at any point A is:
Q

Ms =

sin cos

+
2 sin cos sin

4
4
Q r3

El sin sin
(2 2 sin sin + cos )

o
=

o (sin 1 sin 1 cos 1 cos - sin 1 sin )

cos 1

1 cos
r

sin
cos 2
1
sin 1 + sin sin 1 cos 1 1

2
2
2
o

Y
4r

sin
2

2 r
r
d=
El
El

sin r 3
2r 3

d=
El
El

2 sin 2
r3
sin cos +

El
2

The load on the analogous column is:

Ms
2Q r 2

r d =
o 1 cos sin d
El
2
El

q=

2Q r 2
sin cos
sin
+
=
El
2
2
o

Q r2
[2 3sin + cos ]
El

The moment about the horizontal axis is:

Mh = 2

Ms
El

sin
cos

9 sin cos + 3 2 12 sin 2 + 2 2 sin 2

sin cos + 2 2 sin 2

1
1
2 sin sin sin 2
+
sin cos +
=
2

2 o
2

q = 2 o

M h y Q r 2 3 sin + cos
=

Ih

2

cos

Y = cos

The centroid is sin/r, and the moment of

inertia about the horizontal axis is:
lh = 2

The distance from the neutral axis to point A

is given by:

then the Ms/El diagram is the load on the

analogous column.
The area of this analogous column is:
a1 = 2 o

9
3
3 sin 2 sin 2
+
sin cos +

4
2
4

M1 =

Q r2
El

Then the indeterminate moment is:

d1
=

2 cos 2 cos 2 sin cos

Q r3

o sin
d =
El
(
)
2

2
cos

sin

2
r d =

38

sin
r

Figure 4-10. Plot of longitudinal bending-moment constant K1

Finally, the combined moment is given by:

Finally:

3 sin cos
cos + sin
+

2
2
2

Qr 1
sin

M = M s + Ml =
4 cos

sin

+
4
6
2
cos

9
2

sin cos +1 2 sin

Because of the symmetry the shear stress is

zero at the top of the vessel; therefore, the
direct load in the ring at the top of the vessel,
Pt, may be found by taking moments on the
arc about the horn of the saddle. Then

(1 cos )r Pt = Q r 1 cos

Pt =

sin M M t

sin
1
Q
M Mt
1

2 (1 cos ) r (1 cos )

The direct load, P , at = , the point of

maximum moment may be found by taking
moments about the center. Then
Qr
(1 cos ) (M M t )
r (P + Pt )=

This is the maximum when = ; then:

sin

cos

sin 3 cos + 3 sin +

2
4
4
4

Qr
M =
2

sin

2

4 6
+ 2 cos

2
sin cos
sin

1
2
+

M = K6Qr

Substituting the value above for Pt, and

solving for P, gives:
P =

39

cos
Q sin
M Mt
cos +

2 (1 cos )
(
1
r
cos )

or

P = K7 Q

APPENDIX B
After the article had been published, certain
refinements seemed desirable; therefore, the
following has been added to take greater
advantage of the inherent stiffness of these
vessels. The methods outlined in the paper
will give conservative results.
The effective width of shell has been limited
to lot in order to prepare the chart of Figure
4-2. It has been shown5 that this effective
width may be taken as 1.56 rt . That is, where
5t each side of the saddle or stiffener has been
used, the more liberal value of 0.78 rt each
side could be used.
The values plotted in Figure 4-5 for K1 cover
conservatively all types of heads between H =
O and H = R. More liberal values are given in
Figure 4-10 for hemispherical and 2 to 1
ellipsoidal heads for values of H/L between 0
and 0.1. The minimum values of K1 given in
Table 4-1 have not been listed for specific
values of R/L and H/L; so they are
conservative. Specific minimum values of K1
may be read from Figure 4-10.

where
K7 =

1 sin
cos
cos +
M Mt

2 (1 cos )
(
1 cos )
Q
r

If the rings are adjacent to the saddle, K6 and

K7 may be found in a similar manner, except
that the static structure would become the
entire ring split at the top and loaded as
indicated in Figure 4-9.
Design of Saddles
The
summation
of
the
horizontal
components of the radial reactions on one-half
of the saddle shown in Figure 4-8 must be
resisted by the saddle at = . Then this
horizontal force is given by:
F=

Q ( cos sin + cos sin

r d =
r ( + sin cos )

1 sin 2 cos cos

2

+ sin cos

2
1+ cos 1 sin
2

Q
+ sin cos

Then :

1+ cos 1 sin 2
2
K8 =
+ sin cos

The bending at the horn would change the

saddle reaction distribution, and increase this
horizontal force.

40

This page has been left blank intentionally.

Part V
Anchor Bolt Chairs

hen anchor bolts are required at

supports for a shell, chairs are
necessary to distribute the load to
the shell. Small tubular columns (less than 4 ft
in diameter) may be an exception if the base
plate is adequate to resist bending. Otherwise,
chairs are always needed to minimize
secondary bending in the shell.
For flat-bottom tanks, choose a bolt circle to
just barely clear the bottom without notching
it. For other structures, follow the minimum
clearances shown in Figure 5-1(a). The
designer must evaluate anchor bolt location
for interference with base or bottom plate.

W = total load on weld, kips per lin. in. of

weld
WH = horizontal load, kips per lin. in. of weld
WV = vertical load, kips per lin. in. of weld
= cone angle, degrees, measured from axis
of cone
Z = reduction factor
Top Plate
Critical stress in the top plate occurs
between the hole and the free edge of the
plate. For convenience we can consider this
portion of the top plate as a beam with
partially fixed ends, with a portion of the total
anchor bolt load distributed along part of the
span. See Figure 5-2.
P
(5-1)
S = 2 (0.375 g 0.22d )

Notation
a = top-plate width, in., along shell
b = top-plate length, in., in radial direction
c = top-plate thickness, in.
d = anchor-bolt diameter, in.
e = anchor-bolt eccentricity, in.
emin = 0.8864 + 0.572, based on heavy hex nut
clearing shell by in. See Table 5-1.
f = distance, in., from outside of top plate to
edge of hole
fmin = d/2 + 1/8
g = distance, in., between vertical plates
(preferred g = d + 1) [Additional distance
may be required for maintenance.]
h = chair height, in.
j = vertical-plate thickness, in.
k = vertical-plate width, in. (average width
for tapered plate)
L = column length, in.
m = bottom or base plate thickness, in.
P = design load, kips; or maximum allowable
anchor-bolt load or 1.5 times actual bolt
load, whichever is less
r = least radius of gyration, in.
R = nominal shell radius, in., either to inside
or centerline of plate (radius normal to
cone at bottom end for conical shells)
S = stress at point, ksi
t = shell or column thickness, in.
w = weld size (leg dimension), in.

fc

or
P
c=
Sf

(0.375g

0.22d )

(5-2)

Top plate may project radially beyond

vertical plates as in Figure 5-1(d), but no more
than .
Chair Height
Chair must be high enough to distribute
anchor bolt load to shell or column without
overstressing it. If the anchor bolt were in line
with the shell the problem would be simple
the difficulty lies in the bending caused by
eccentricity of the anchor bolt with respect to
the shell. Except for the case where a
continuous ring is used at the top of chairs,
maximum stress occurs in the vertical
direction and is a combination of bending plus
direct stress. Formulas which follow are
approximations, based on the work of Bjilaard.

1.32 Z
Pe
.031
S= 2
+
t 1.43 ah 2
Rt
2 ..333
4
ah
+

Rt

( )

42

(5-3)

Minimum chair height h = 6, except use h =

12 when base plate or bottom plate is 3/8 or
thinner and where earthquake or winds over
100 mph must be considered.
Maximum recommended chair height h = 3a.
Table 5-1. Top-Plate Dimensions
Based on anchor-bolt stresses up to 12 ksi for
1 -in.-diameter bolts and 15 ksi for bolts 1
in. in diameter or larger; higher anchor bolt
stresses may be used subject to designers
decision.

g=d+1

e min

c min

Bolt
Load,
kips
P

1.87

0.734

19.4

Top Plate Dimensions, in.

d

Figure 5-2. Assumed Top-Plate Beam.

Z=

1.0
2

.177 am m
+ 1. 0
Rt t

2.09

0.919

32.7

2.30

1.025

43.1

2.52

1.145

56.6

If chair height calculated is excessive, reduce

eccentricity e, if possible, or use more anchor
bolts of a smaller diameter. Another solution is
to use a continuous ring at top of chairs.
If continuous ring is used, check for
maximum stress in circumferential direction,
considering the ring as though it were loaded
with equally spaced concentrated loads equal
to Pe/h. Portion of shell within 16t either side
of the attachment may be counted as part of
the ring. (Refer to Figure 5-3)
Note that the base plate or bottom is also
subjected to this same horizontal force, except
inward instead of outward. This is true even if
a continuous ring is not used around the top of
the chairs but it should never cause any very
high stresses in the base, so we do not
normally check it. However, it is a good thing
to keep in mind in case you have a very light
base ring.

Figure 5-1. Anchor-Bolt Chairs.

Where:

Vertical Side Plates

Be sure top plate does not overhang side
plate (as in Figure 5-1d) by more than
radially.
Vertical-plate thickness should be at least jmin
= or 0.04 (h c ), whichever is greater.
Another requirement is jk P/25, where k is
the average width if plate is tapered.
These limits assure a maximum L/r of 86.6
and a maximum average stress in the side

(5-4)

Maximum recommended stress is 25 ksi.

This is a local stress occurring just above the
top of the chair. Since it diminishes rapidly
away from the chair, a higher than normal
stress is justified but an increase for temporary
loads, such as earthquake or wind is not
recommended.
The
following
general
guidelines are recommended.
43

plates of 12.5 ksi, even assuming no load was

transmitted into the shell through the welds.

Figure 5-5. Typical Welding,

Base or Bottom Field Attached.

Figure 5-3. Chair with Continuous Ring at Top.

Assembly of Chair
For field erected structures, ship either the
top plate or the entire chair loose for
installation after the structure is sitting over
the anchor bolts.
Where base plate is welded to skirt or
column in shop, attach side plates in the shop
and ship top plate loose for field assembly.
(See Figure 5-4)
Where base or bottom plate is not welded to
shell in the shop, as for flat-bottom tanks and
single pedestal tanks, shop attach side plates
to top plates and then ship the assembly for
field installation. When you do this, weld both
sides at top of side plates so shrinkage will not
pull side plate out of square. (See Figure 5-5)
Welds between chair and shell must be
strong enough to transmit load to shell.
minimum fillet welds as shown in Figures 5-4
and 5-5 are nearly always adequate, but you
should check them if you have a large anchor
bolt with a low chair height. Seal welding may
be desired for application in corrosive
environments.
Assume a stress distribution as shown in
Figure 5-6 as though there were a hinge at
bottom of chair. For the purpose of figuring
weld size, the base or bottom plate is assumed
to take horizontal thrust only, not moment.
Note that loads are in terms of kips per inch
of weld length, not in terms of kips per square
inch stress. Critical stress occurs across the top
of the chair. The total load per inch on the
weld is the resultant of the vertical and
horizontal loads.

Figure 5-6. Loads on Welds.

Formulas may also be used for cones,
although this underrates the vertical welds
some.
WV =
WV =

P
a + 2h
Pe

ah + 0.667h 2

W = W 2v + W 2 H

(5-5)
(5-6)
(5-7)

For an allowable stress of 13.6 ksi on a fillet

weld, the allowable load per lin. in. is 13.6 x
0.707 = 9.6 kips per in. of weld size. For weld
size w, in., the allowable load therefore is:

9.5w > W

(5-8)

Design References
H.
Bednar,
Pressure
Vessel
Design
Handbook, 1981, pp. 72-93.
MS. Troitsky, Tubular Steel Structures, 1982,
pp. 5-10 to 5-16.
P.P. Bjilaard, Stresses From Local Loadings in
Cylindrical Pressure Vessels, ASME
Transactions, Vol. 77, No. 6, 1955.
P. Buthod, Pressure Vessel Handbook, 7th
Edition, pp. 75-82.

Figure 5-4. Typical Welding,

Base Plate Shop Attached.
44

This page has been left blank intentionally.

Part VI
Design of Fillet Welds

should check to see what code, if any, applies

to the work.
In these papers, however, the fillet weld
throat dimension for an equal-leg fillet is
assumed to be the leg length times 0.707 (i.e.
cos 45).

esign of butt welds is closely

controlled by weld details and joint
efficiencies clearly specified in
various codes and specifications.
Design of fillet welds, however, is not so
clearly outlined. The following pages are
intended to fill the gap. While referring to the
following pages and designing fillet welds, the
designer is encouraged to keep in mind actual
shop and field welding practice and the
quality of fillet welds that can consistently be
expected. The size and length of the weld as
well as the allowable stresses used in their
design should reflect the actual shop and field
welding and not necessarily the value used
here.
Size of an equal-leg fillet weld is the leg
width W of the largest 45 right triangle which
fits in its cross section. They are referred to by
their leg sizes, such as a in. fillet weld.

Economy of Welding
Economical design of fillet welds includes
the following:
1. Use of 450 (equal leg) fillet welds
whenever possible
2. Minimum size of fillet
3. Lower cost of down welding position
4. Locate weld to eliminate eccentricity
5. Balanced welds to control distortion
6. Avoid locating welds in highly stressed
areas
7. Readily accessible
Use the smallest size of fillet permitted (see
Fillet Weld Limitations). Flat fillets 5/16 and
smaller are normally made in one pass and are
more economical than larger fillets. Generally,
the fillet with the least cross-sectional area is
the most economical. Increasing the size of a
fillet weld from 1/4 to 3/8 more than
doubles the amount of filler metal, but the
strength only increases 50%. A gap also
requires additional filler metal.

Figure 6-1. Fillet-Weld Sizes (Leg Dimensions).

Size of an unequal length fillet weld is
described by the leg lengths of the largest right
triangle which fits in its cross section, such as a
3/8 by 1/2 fillet weld.
The strength of a fillet weld is assumed to
equal the allowable shearing stress times the
throat area of the weld. The throat area of a
weld is the length of weld times the theoretical
throat distance, which is the shortest distance
from the root of the weld to the theoretical
welds surface.
Some codes, however, define the throat
distance differently.
AWWA defines the throat as .707 times the
length of the shorter leg of the fillet weld.
AISC distinguishes between welding
processes to be used when determining throat
distances (eg. AISC ASD J2.2). The designer

Figure 6-2. Volumes of 1-in. Long Welds.

45

Parallel welds have forces applied parallel

to their axis. Fillet weld throat is stressed only
in shear. Parallel welds may also be called
longitudinal welds.

Flat welding position is the most economical

and overhead the least. For example, the
relative costs of 3/8 fillets for different
positions are:
lap flat
flat fillet
vertical fillet
overhead fillet

100%
110%
240%
250%

Figure 6-6. Parallel Weld.

The costs can vary according to weld

procedure used.
Specify shop welding whenever practical.
The fitted-up material can normally be
repositioned easier in the shop.

Transverse welds have forces applied at

right angles to their axis. Fillet weld throat has
both shear and normal (tensile or compressive)
stresses. Transverse welds are about 33%
stronger than parallel welds.

Type of Fillet Welded Joints

Single-fillet welded joints
Strength depends on size of fillet.
Do not use when tension due to bending is
concentrated at root of weld.
Do not use for fatigue or impact loading.
Difficult to control distortion.

Figure 6-7. Transverse Weld

Allowable Loads on Fillet Welds
Stress in a fillet weld is assumed as shear on
the throat area, for any direction of applied
load.
Many codes express the allowable shear
stress for fillet welds in psi on the throat area.
It is more convenient, however, to express the
strength of fillet welds as allowable load f, kips
per lin. in. for 1 fillet.
The following formula may be used to
convert allowable shear stress on throat area to
allowable load for 1 fillet with equal leg
lengths.

Figure 6-3. Types of Single Fillet Welds.

Double-fillet welded joints
Used for static loads.
Economical when fillet size is or less. Lap
joint maximum strength in tension when
length of lap equals at least 5 times the
thickness of thinner material.

F = 0.707 x allowable shear stress, ksi

Figure 6-4. Types of Double Fillet Welds.

(6-1)

Since transverse welds are stronger than

parallel (or longitudinal) welds, some codes
permit different allowable stresses for them.
API 620 9th Edition and AWWA D100-96 are
two codes that have different allowable
stresses for the two types of welds. API 650 9th
Edition and AISC 9th Edition, however, make
no distinction between transverse welds and
parallel welds and use the same allowable
stress for both. The designer is cautioned to
check which code applies to the work at hand
as well as the most recent edition of the code

Double-fillet welded corner joint - Complete

penetration and fusion. Used for all types of
loads. Economical on moderate thickness.

Figure 6-5. Corner Joint.

Welds transmit forces from one member to
another. They may be named according to the
direction of the applied forces.
46

= polar moment of inertia, in.4, of member

transmitting load to weld
Jw = polar moment of inertia, in.3, of weld
lines subjected to torque
L = column length, in.
M = bending moment, in.-kips
n = number of plate sides welded or number
of welds loaded
P = allowable concentrated axial load, kips
Q = statical moment of area, in.3, above or
below a point in cross section, about
neutral axis
r = least radius of gyration, in.
S = section modulus, in.3, of member
transmitting load to weld or of weld
subjected to moment
Sw = section modulus, in.2, of weld lines
subjected to bending moment
t = plate thickness, in., or thickness, in., of
thinnest plate at weld
T = torque, in.-kips
V = vertical shear, kips
w = fillet weld size (leg dimension), in.
W = total load on fillet weld, kips per lin. in.
of weld
Wb = bending force on weld, kips per lin. In. of
weld
Wh = horizontal component of torsional force
on weld kips per lin. in of weld
Wq = longitudinal shear on fillet weld, kips per
lin. in. of weld
Ws = average vertical shear on fillet weld, kips
per lin. in. of weld
Wsa = actual shear on fillet weld, kips per lin.
in. of weld
Wt = torsional load on fillet weld, kips per lin.
in. of weld
Wv = vertical component of torsional force on
weld, kips per lin. in.
x = distance from y axis to vertical weld
y = distance from x axis to horizontal weld

to see if their approach to these types of

stresses has changed.
In the following pages, however, for the sake
of completeness, a distinction will be made
between the two types of stresses, fp and ft.
When a joint has only transverse forces
applied to the weld, use the allowable
transverse load ft. If only parallel forces are
applied to the weld, use the allowable parallel
load fp. If one of the forces is parallel and the
other forces are transverse, use the allowable
transverse load when the resultant force is
found from Equation 6-1.
Specifications on allowable stress for fillet
welds are given in Section 2.4 of AWS
Structural Welding Code, D1.1-94.
Current AISC specifications also refer to:
1. allowable stress at weld for both weld
metal and base metal
2. minimum length of fillet weld
3. minimum size of fillet weld
4. maximum size of fillet weld
5. end returns or boxing of welds
6. spacing of welds
7. fatigue loading of welds
Notation
A = cross-section area, sq. in., of member
transmitting load to weld
Aw = length, in., of weld
b = length, in., of horizontal weld
c = distance, in., from neutral axis to outer
parallel surface or outer point
ch = horizontal component of c, in.
cv = vertical component of c, in.
d = depth, in., of vertical weld
f = allowable load on fillet weld, kips per lin.
in. per in. of weld size
fb = bending stress, ksi
fp = allowable parallel load on fillet weld,
kips per lin. in. per in. of weld size
ft = allowable transverse load on fillet weld,
kips per lin. in. per in. of weld size
fto = torsional stress, ksi
I = moment of inertia, in.4, of member
transmitting load to weld or of weld
subjected to torque
Io = moment of inertia about o axis, in.4
Ix = moment of inertia about x axis, in.4
Iy = moment of inertia about y axis, in.4

Combined Loads on Welds

It is necessary to designate the size and
length of fillet welds. Since neither are known,
it is usually simpler to assume the length and
then calculate the size.
Fillet weld size w, in., is found by dividing
the force W, kips per lineal inch, on the weld
by the allowable load f (kips per lin. in. for 1
fillet) for the weld.
w=

47

W
f

(6-2)

Table 6-1. Formulas for Force on Weld

Type of Loading

Tension or
Compression

Vertical
Shear

Bending

Torsion

Common
Design
Formulas
for Stress,
psi

P
A

V
A

Formula
for Force
on Weld
K/Kips per
in

W=

Ws =

M
S

Wb =

Tc
J

Wt =

P
Aw

Figure 6-8. Forces on Weld Combined.

To determine the resultant force for
combined forces, use Equation 6-3. If only two
forces exist, use 0 for one force.

V
Aw

[ (

W = W12 + W2 2 + W3 ft / f p

)]

(6-3)

Refer to Figure 6-8 for explanation of W1, W2,

and W3.
The total force shall be determined in
accordance with the applicable code.

M
Sw

Simple tension or compression loads. The force

W, kips per inch of weld, is the load P divided
by the length Aw of weld. As shown in Table
6-1 the tensile or compressive force on a weld
is:
P
(6-4)
W=

Tc
Jw

Aw

With this force W, the required fillet weld is

calculated from Equation 6-1.
Longitudinal
Shear

VQ
tl

Wq =

Example: Find size of fillet welds for the

connection shown in Figure 6-9. Assume Aw =
5 + 5 + 2 = 12 .

VQ
nl

Force W on a weld depends on the loading

and shape of the weld outline. Table 6-1 shows
the basic formulas for determining weld forces
for various types of loads.

30,000 lbs

Combining forces: There may be more than

one force on the weld, such as bending force
and shear force. It is usually easier to
determine each force independently and then
combine vectorially to obtain a resultant force.
All forces which are vectorially added must
occur at the same position in the weld. Be sure
to find the position on the welded connection
where the combination of forces will be
maximum.
To simplify calculations, increase parallel
forces by the ratio ft/fp before combining to
account for the lower allowable parallel shear
stress specified by some codes.

Figure 6-9. Tension-Member Connection1.

Referring to API 650, the allowable basic
shearing stress of an E60 electrode fillet weld is
13.6 ksi.
f = (.707) (13.6 ksi) (1 inch weld) = 9.6
kips/inch/1 inch weld.

AISC for E60 electrodes would give f = (.707) (.3)

(60) = 12.7 ksi shear stress with max shear stress on
base metal of .4 yield of base metal.
1

48

W=

P
30
=
= 2.4 kips
Aw 12.5

w=

per lin. in.

W 2.4
=
= 0.25"
f 9.6

Wsa at 2 =

VQ V (0)
=
=0
n
n

(6-8)

Bending or torsional load may be applied to

the same weld outline.

Use fillet

2
Weld volume = (1/ 4) 12.5 = 0.39 cu. in.

Try Aw2 = 5 + 5 = 10
W2 =

P 30
= = 3.0
Aw2 10

W2 =

W2 3.0
=
= 0.312
f 9.6

kips per lin. in.

Use 5/16 fillet

Bending and
Shear Load
on a Weld

2
Weld volume = (5 / 16) 10 = 0.49 cu. in.

Use fillet on three sides because of less

weld volume. Check fillet size (see Fillet Weld
Limitations).
Shear load is considered uniformly
distributed over the length of weld. Force
formula Ws = V/Aw from Table 6-1 gives
average shear force. Use average shear force
when combining with bending force or
torsional force. However, if the average shear
force about equals or exceeds the bending or
torsional force, determine the actual shear
force distribution to aid in locating the
maximum combined force.
The actual shear force per weld at any point
can be determined from:
VQ
(6-5)
Wsa =

In the figure with the bending load, the weld

must transfer the same stress as in the member
at the connection. This stress can be
determined using the common formula for
bending stress.
M
fb =
(6-9)
S

In the connection with the torsional load, the

weld wants to rotate or twist about the center
of gravity of the weld group. The stress in the
weld can be found from:
Tc
(6-10)
fto =
J

However, before using these formulas, it is

necessary to determine the section modulus S
or polar moment of inertia J of the weld
without knowing its width (size). A simple
way to determine the section modulus or polar
moment of inertia of the weld is to treat the
weld as a line. The property, such as section
modulus S, of any thin area is equal to the
property of the section when treated as a line
Sw times its thickness w.
S = Sww
(6-11)

For example, the average shear force and

actual shear force distribution are compared
for a rectangular member in Figure 6-10.

The common formula for bending stress can

now be used to find the bending force on the
weld.
M
M
Wb = w =
(6-12)

Figure 6-10. Shear Distribution at Welds.

Average shear force

d d
V V
VQ
3V
Wsa at 1=
= 2 2 =
=1.5Ws
n
td 3 4d

2
12

Weld
Outline

Figure 6-11. Moment and Torque on Weld.

V
V
Ws =
=
Aw 2d

Torsional
and Shear Load
on a Weld

Sw

Properties of sections treated as lines for

typical weld outlines are shown in Table 6-2.
The method for determining these properties
is given later.
When designing welds using the line
method, select the weld outline with care.

(6-6)
(6-7)

49

Several combinations of line welds will

produce the required property Sw or Jw.
However, select the weld outline where the
weld distribution is consistent with the load
distribution in the member at the connection.
For non-circular members (such as beams,
channels, etc.) resisting torsion loads,
transverse forces on the weld are present in
addition to parallel forces computed from
Tc/Jw. These transverse forces are the result of
the non-circular cross section warping and
should not be neglected.

Bending and shear forces on a welded

connection are combined vectorially after
determining each force independently from
Equations 6-11 and 6-5. Determine the
combined force W on the weld using Equation
6-2. Make sure you have found the position on
the welded connection where the combination
of forces will be maximum. See Figure 6-10 for
shear force distribution.
Calculate the required weld size from
Equation 6-2.

Table 6-2. Properties of Weld Outlines

(Treated as a Line)
Outline of Welded
Joint

Bending
(about x x axis)

Sw =

Torsion

d2 2
in.
6

d3 3
in.
12

Jw =

d2
3

Jw =

d 3b 2 + d 2
6

S w = bd

Jw =

b b 2 + 3d 2
6

Sw =

S w (top )=

d (4b + d )
6

d 2 (4b + d )
6 (2b + d )
(max force at bott )
S w (bott )=

S w = bd +

d
6

Jw =

Jw =

d 2 (2b + d )
3 (b + d )
(max force at bott )

S w = bd +

d2
3

Example:
Find size of
fillet weld on clip loaded as
shown in Figure 6-13. Use ft =
8.9 kips per lin. in. and fp = 6.4
kips per lin. in. from API 620.
Assume length of fillet = 10 (5
each side)

(b + d )4 6b2d 2
12 (b + d )

Sw from Table 6-2

(2b + d )3 b2 (b + d )2

Jw =

= 1.44 kips per lin. in.

(b + 2d )3 d 2 (b + d )2

Avg shear force

b + 2d

12

d 2 52
=
= 8.33 sq.in
3
3
force Wb = M = 4 3
S w 8.33

2b + d

12

Bending

d (2b + d )
S w (top )=
3
S w (bott )=

Figure 6-12. Bending and Vertical Shear

on Welds.

Ws =

V
4
=
Aw 10

= .40 kips per lin. in.

Jw =

Figure 6-13. Loaded Clip.

(b + d )3

ft = (.707) (12.6 ksi) (1 inch weld) = 8.9 kips/inch/1 inch weld

fp = (.707) (9.0 ksi) (1 inch weld) = 6.4 kips/inch/1 inch weld

Sw =

d2
4

Jw

d3
4

Revised and expanded outline properties given in Lincoln Electric

publication D8 10.17. Solutions to Design of Weldments, p. 3.

50

Resultant force W =

f
Wb 2 + Ws t
f p

8.9
1.442 + 0.40

6.4

=
Fillet size w =

Tch
Jw

Wv =

(6-15)

Equation 6-2 can now be used to find the

resultant force on the weld. Increase the forces
parallel to the weld at the point considered by
ft/fp before combining.
The required fillet size is calculated from
Equation 6-2.

1.544 kips per lin. in.

W 1.544
=
= .173"
Ft
8.9

Use 3/16 fillet

Example: Find fillet

size for connection
shown in Figure 6-15.
Use ft = fp = 9.6 kips
per lin. in.

Note that the designer is still cautioned to

check the shear capacity of the plate.
Torsional and shear forces on a welded
connection are combined vectorially after
determining each force independently from
Equation 6-3 and the torsional force formula.
Tc
(6-13)
Wt =
Jw

Maximum torsional force occurs at the most

distant weld fiber measured from the center of
gravity of the weld outline. This distance to
the outer fiber is c in Equation 6-12.
The direction of the torsional force Wt may
be other than horizontal or vertical. By
resolving the torsional force into vertical and
horizontal components, the problem of
combining forces is simplified. Resolve the
torsional force into components by using the
horizontal and vertical components of
dimension c as indicated by Equations 6-13
and 6-14.

Figure 6-15. Loaded Bracket.

From Table 6-2.
x=

32
b2
=
= 0.75"
2b + d 2 3 + 6

Jw =

(2b + d )3
12

= (2 3 + 6)

12

b 2 (b + d )2
2b + d

32 (3 + 6)2
= 83.25 in.3
23+ 6

Find components of maximum torsional

force at 1.
cv = 3"

ch = 3 x = 2.25"

By Equation 6-13, the horizontal component

of torsional force is:
Wh =
=

Tcv
Jw

5 (3.75 + 2.25 ) (3)

83.25

= 1.08 kips per lin. in.

Figure 6-14. Torque and Shear on Welds

NOTE: (3.75 + 2.25) is the distance from the

point load to the centroid of the weld.

The horizontal torsional force component is:

Tc
(6-14)
Wh = v
Jw

The vertical torsional force component is:

51

Longitudinal shear force may vary along the

length of the beam. The vertical shear diagram
for the beam can be used as a picture of the
amount and location of welds between flange
and web.

By Equation 6-14, the vertical component of

torsional force is:
Tcv
Jw

Wv =

5 (3.75 + 2.25) (2.25)

83.25

= 0.810 kips per lin. in.

Find average vertical shear force:
Ws =

5
V
=
Aw 3 + 6 + 3

= 0.416 kips per lin. in.

Combine forces using Equation 6-2
W=

Figure 6-17. Shear in Beams.

*
9.6

(0.810 + 0.416)2 + 1.08 9.6

Notice there is no shear in the middle

portion of beams 1 and 2; therefore, little or no
welding is required in this portion. When
there is a difference in shear along the length
of beam, as in beam 3, the welding could vary
in this same ratio along the length of beam.
This is why continuous welding is sometimes
used at the ends of beams and reduced size or
intermittent fillet welds used throughout the
rest of the beam.

= 1.635 kips per lin. in.

Calculate weld size using Equation 6-1.
w=

W 1.635
=
= 0.17"
f
9.6

Use 3/16 fillet:

Check fillet size (see Fillet Weld Limitations).

Built-up
members
subject
to
axial
compression: Welds joining the component
parts of a built-up compression member, such
as a cone roof tank column, are also stressed in
longitudinal
shear.
Determine
this
longitudinal shear force Wq from Equation
6-15 using the shear V at any position along
the member as given by Equations 6-16 or
6-17.

Welds in Built-up Members

Built up members subject to bending:
Welds attaching the flange to the web are
stressed in longitudinal shear and must be
adequate
to
transfer
the
calculated
longitudinal shear force.

Longitudinal shear force at any position

along the length of beam is calculated from
VQ
n

(6-16)

V = 0.01P for L/r < 60

(6-18)

w=

Note that if we had been using API 620 where ft = 8.9 kips per
lin. in. and fp = 6.4 kips per lin. in. this equation would be

W =

(6-17)

Also at each end of a built-up compression

member, use a total length of continuous fillet
weld equal to the maximum width or depth of
the member or 4, whichever is greater.
Fillet weld size at any position along the
beam or column is determined from Equation
6-1 with the longitudinal shear force W, at the
same position.

Figure 6-16. Examples of Built-up Members.

Wq =

V = 0.02P for L/r > 60

6. 4

(.810 + .416)2 + 1.08 8.9

52

W Wq
=
f
fq

(6-19)

Table 6-3. Length and Spacing

of Intermittent Welds
Continuous
Welds %

Length of Intermittent Welds and

60

3-5

Distance Between Centers, In.

57
50

3-6

plate flange
Compression

3-7
2-5

37

4-10

2-6

3-9

4-12

3-10

25

2-8

20

2-10

16

2-12

rolled shape
flange

24t (12 max)*

24
22t (12 max)*
24

Maximum clear space between intermittent

fillet welds depends on the component parts of
the built-up member. The clear space between
welds must be close enough to prevent local
buckling of the component parts when the
loading develops the full strength of the builtup member.

3-8

30

rolled shape
flange

4-8
4-9

43

33

plate flange
Tension

4-7
2-4

44
40

Table 6-4. Maximum Clear Space

Between Intermittent Fillet Welds
(Carbon Steel Built-up Members)

3-12

Example: Find size and spacing of fillet weld

joining plate and angle of built-up member
shown in Figure 6-19. Use ft = 8.9 kips per lin.
in., fp = 6.4 kips per lin. in.

Use intermittent fillet welds when the

calculated leg size is smaller than the
minimum specified in Table 6-5. The
calculated size divided by the actual size used,
expressed in percent, gives the length of weld
to use per unit length:
calculated leg size (continuous )
(6-20)
%=
100
actual leg size (int ermittent )

Intermittent weld lengths and distances

between centers for given percentages of
continuous welds are shown in Table 6-3.
Figure 6-19. Plate Grinder.
Shear diagram for beam shows that welding
for longitudinal shear could be reduced in
center portion of beam. Because the vertical
shear is small, design the welds for maximum
shear throughout the length of beam.

Figure 6-18. Spacing of Intermittent Welds.

The longitudinal shear force is:

Minimum length of fillets for intermittent

welds is 2 or 4w, whichever is greater.
Selecting the longest fillet possible is usually
the most economical. However, do not exceed
the maximum clear space between fillets in
Table 6-4.

Wq =

VQ
0.623 (0.1875) 6 (0.518)
=
n
1 (1.094)

= 0.332 kips per lin. in.

* Many of the built-up members we use have an assumed flange.

This flange, usually part of a roof, bottom or shell, may be
partially restrained from local buckling when the maximum
load is applied. When the built-up member has restraint on the
flange, the clear space between fillet welds could be increased to
about 32t maximum.

53

Minimum length of fillets for strength welds:

1 or 4w, whichever is greater
(Use 2 or 4w for intermittent welds)

The continuous weld size required is:

w=

Wq
fp

0.332
=
= .052
6.4

Spacing of Fillet Welds:

1. When bars or plates are connected only by
a set of parallel longitudinal fillets, the
length of those welds should not be less
than the perpendicular distance between
those two welds.
2. When fillet welds are used for end
connections, the distance between them
must not be greater than 8 inches unless
transverse bending is otherwise prevented.

(Use fp because longitudinal shear force is

parallel to weld.)
Minimum size fillet from Table 6-5 is 3/16.
% continuous weld =

0.052
100 = 27.7%
0.1875

Minimum length fillet permitted for

intermittent welds is 2.
Maximum clear space between fillets is,
from Table 6-4, 22 x 3/16 = 4.1.
Maximum spacing with 2 fillet = 2 + 4.1 =
6.1.
Use 2 6 intermittent fillet on one side.
This provides 33% (Table 6-3) continuous weld
which is more than adequate to transfer the
calculated longitudinal shear.

Determining Weld Outline Properties

Properties Sw and Jw of a weld outline when
treated as a line are nearly equal to the section
modulus or polar moment of inertia divided
by the width w of the weld. When w is small,
say 10% of d, the error is usually less than 1%.
The properties Sw and Jw in Table 6-2 are
determined as follows:
From handbook

Fillet Weld Limitations

Minimum size fillet: The calculated weld size
may be small. To eliminate cracks resulting
from rapid cooling, it is best not to put too
small a fillet on a thick plate. Follow Table 6-5
for minimum sizes.

Ix =

S = Ix

Table 6-5. Minimum Size Fllets

Thickness3
1/2
3/4
> 3/4

w3 d
12
d wd 2
=
2
6

Treated as a line
dividing by w, then

Minimum Leg Size

of Fillet 2
3/16
1/43
1/43, 4

Sw =

S d2
=
6
w

by

about x axis (6-21)

Figure 6-21. Vertical Weld.

From handbook,

Maximum size fillet for strength welds:

Iw =

w3d
12

When w is small, let Iy = 0

J = Ix + I y =

w3d
+0
12

Treated as a line, then

Jw =

Figure 6-20. Weld Size Limited to Plate

Thickness.
3

Thickness of thicker part to be joined.

Leg size of fillet need not exceed thickness of thinner part to be

joined.

3
A minimum fillet of 3/16 is acceptable provided 200F
preheat or surface examination of the weld (PT,MT) is
performed.
4

AWS D1.1-82 or AISC require a minimum 5/6 fillet.

54

J d3
=
w 12

(6-22)

From handbook, for a horizontal weld

Io =

I
Sw = x
w

w d
12

I y = I o + Ay 2 = 0 + w b y 2
I y = wb y2

Jw =

S = I y y = wb y

Treated as a line, then

S
= by about x axis
W

(6-23)

From handbook
Iy =

wb3
12

J = I x + I y = wby 2 +

wb3
12

Treated as a line, then

Jw =

J
b3
= by 2 +
12
w

(6-24)

By adding the properties of the two basic

lines in Figures 6-21 and 6-22, properties for
other straight line outlines may be determined.
For example, find Sw and Jw for the outline in
Figure 6-23.

Figure 6-23. Combination of Welds.

wd3
wd3
+ 2 wb y2 =
+ 2w b y 2
12
6
d
When y =
,
2
Ix = 2

Ix =

w d 3 w b d 3 w d 2 (d + 3b )
+
=
6
2
6

I y =0+ 2

d2
+ bd about x axis
3

(6-25)

I x + I y w d 2 (d + 3b )+ w b3
J
=
=
w
w
6w
b3 + 3bd 2 + d 3
6

(6-26)

Cautionary Note
Some designers and engineers are not aware
of a form of cracking called lamellar tearing,
which can occur beneath highly stressed Tjoints in steel plate.
Plate forced to deform plastically in the thruthickness direction by welds which are large,
multi-passed, and highly restrained can
decohere at a plane of microscopic inclusions.
A crack may then progress from plane-toplane in a terrace-like fashion.
While lamellar tearing is not frequent, even
one incident has the potential of becoming a
serious problem. Since there are means to
minimize the hazard, it behooves the engineer
to take every precaution by optimizing joint
design and welding procedure selection.
Where these factors cannot be controlled, it
may be necessary to use special steels.
The reader is referred to the following
sources for guidance in designing against
lamellar tearing:
1. Engineering Journal, Third Quarter, 1973,
Vol. 10, No. 3, pages 61-73. American
Institute of Steel Construction, Inc., 1221
Avenue of the Americas, New York,
New York 10020.
2. Bibliography on Lamellar Tearing,
Welding Research Council Bulletin 232.
Welding Research Council, 345 East
Forty-Seventh Street, New York, New
York 10017.

Figure 6-22. Horizontal Weld.

Sw =

d 2 2 wd 2 (d + 3b )
=

6 wd
3

wd 3 wd 3
=
12
12

55

This page has been left blank intentionally.

Part VII
Inspection and Testing of Welded Vessels

the leak. Since the bottom of a tank is

inaccessible
from the underside, the
differential pressure necessary for the test is
accomplished by means of a vacuum box
placed on the top side. This box has a glass top
and is open on the bottom. The portion of the
weld to be inspected is brushed with a soapy
solution, the box is fitted over it, and a vacuum
created in the box. The weld is inspected
through the glass top for leak-indicating
bubbles.

treatise on the subject of defects in

welded vessels and their detection is
beyond the scope of this work. But an
acquaintance with some of the available
inspection and testing tools may serve to
dispel the mystery of unfamiliar terms.
In the interest of economy, the refinement of
inspection and testing must be in tune with the
degree of perfection necessary for various
classes of work. For example, a pressure vessel
storing a lethal substance, or one constructed
of a special material known to be crack
sensitive, may require as a minimum that
100% of all main joints be radiographed. On
the other hand, simple structures such as oil
and water tanks, constructed of readily
weldable materials, usually require only spot
examination. In general, it is safe and wise to
follow the inspection requirements of the
applicable codes.
First, let us distinguish between hydrostatic
or overload testing to demonstrate strength or
liquid tightness, and inspection to determine
weld quality.

Inspection for Weld Quality

Prior to the beginning of any welding, weld
qualification and welder certification tests
should be performed. These tests insure that
the type of welds proposed are adequate for
the application and that the welders are
capable of applying the required welds.
VISUAL INSPECTION is usually the first
stage in the inspection of a finished weld,
regardless of any other tool that may be
employed. Visual inspection can determine
conformity
with
specifications
as
to
dimensional accuracy, extent, etc. It can also
reveal noticeable surface flaws, such as
obvious cracks, surface porosity, undercutting
of parent metal, etc. In some types of work,
visual inspection is the only inspection
performed; e.g., welds subjected only to
compression as in a tubular column, or lowstressed fillet welds. But for most important
structures, further inspection is usually
required for the main joints, on which the
strength of the structure depends. Visual
inspection should also verify proper contour,
weld preparation and removal of weld spatter
and burrs per NACE standards for application
of the coating. Some of the more commonly
used methods are described below.

Testing for Strength and Tightness

Required overload tests are clearly outlined
in the various governing codes. Whenever the
structure itself, its supports, and foundation
conditions will permit, the overload test is
usually hydrostatic, i.e., the structure is full of
water when the overload, if any, is applied.
For the water and oil tanks of Volume 1, no
overload can be applied other than that
inherent in any difference between the specific
gravity of water and that of the product to be
stored in service. The normal cone roof will
withstand internal pressures only slightly
greater than the weight of the roof plates. It
will not withstand hydrostatic pressure due to
overfilling. Hence, the water test level is
limited to the top capacity line.
The testing of the flat bottom, however, may
warrant brief comment. The liquid tightness of
a flat bottom is usually demonstrated by
means of a soap bubble test. A soapy liquid is
brushed on the weld and a small differential
positive pressure created on the opposite side
of the plate. Leaks in the weld will be
indicated by bubbles as the air passes through

RADIOGRAPHY is an inspection method

that shows the presence and nature of
macroscopic defects or other discontinuities in
the interior of welds. Just as in the case of
medical X-rays with which we are all familiar,
radiography utilizes the ability of X-rays or
gamma rays to penetrate objects opaque to
ordinary light. Radiograph films can reveal
slag (non-metallic) inclusions, porosity or gas
57

PENETRANT utilizes visible instead of

fluorescent dyes. As the dye penetrant rises
from the flaw by capillary action, it stains the
developer (usually a chalky substance) and
clearly marks the flaw.
ULTRASONIC INSPECTION requires a
great deal of explanation for even a
rudimentary understanding of how it works.
Briefly, ultrasonic testing makes use of an
electrically timed wave of the same nature as a
sound wave, but of a higher frequency, hence
the name ultrasonic. The sound wave or
vibrations are propagated in the metal being
inspected and are reflected back by any
discontinuity or density change. The search
unit contains a quartz or similar crystal, which
can be moved over the surface much like a
doctors stethoscope. The search unit applies
energy to the metal surface in short bursts of
sound waves for a very short, controlled
period of time. The crystal then ceases to
vibrate for a sufficient period of time to receive
the returning echoes. The reflected signals are
indicated on a cathode ray tube or
oscilloscope.
From
the
reflection
or
oscilloscope pattern, a trained operator can
determine the distance to the discontinuity
and some measure of its magnitude.
Ultrasonic testing is a valuable tool for
certain applications. But it must be used only
by an operator skilled in the interpretation of
the reflection patterns.
In addition to the above methods the
following can be used: Eddy Currents,
Acoustic Emission, Video Enhancement,
Ultrasonic
Holography,
and
Neutron
Radiography. Only technically qualified
personnel should use these methods.

pockets, cracks, lack of fusion, inadequate

penetration, and even surface defects, such as
undercut. However, welds are rarely perfectly
free of all minor defects nor do they need to
be. As a result, the inspector must have a good
background of experience in reading films,
and knowledge of standards. The various
construction codes, such as AWS and ASME,
define limits of acceptability.
MAGNETIC PARTICLE INSPECTION is an
aid to visual inspection for surface defects too
fine to be detected by the naked eye, plus
those that lie slightly below the surface. With
special equipment, more deeply seated
discontinuities can be detected. The method is
applicable only to magnetic materials. It will
not function on non-magnetic materials such
as the austenitic stainless steels.
The basic principle involved is as follows:
When a magnetic field is established in a ferromagnetic material containing one or more
discontinuities in the path of the magnetic
flux, minute poles are set up at the
discontinuities. These poles have a stronger
attraction for magnetic particles than the
surrounding surface of material.
Normally the area to be inspected is
magnetized between two prods by
introducing high amperage current or some
other convenient means. Then the area is
covered with a powder of finely divided
magnetic particles. These form a visible
pattern of any discontinuity due to the
stronger attraction at those points.
LIQUID PENETRANT INSPECTION is
another method for detecting surface
discontinuities too small to be readily seen by
the naked eye. It is particularly useful on nonmagnetic materials where the magnetic
particle method is ineffective.
The method utilizes liquids with unusual
penetrating qualities, which, when applied to a
previously cleaned surface, will penetrate all
surface discontinuities. The surface is then
cleaned of all excess penetrant and a developer
applied. Penetrant that has entered a crack or
other discontinuity will seep out, make contact
with the developer and indicate the outline of
the defect.
There are two principal types of penetrant
used.
When a FLUORESCENT PENETRANT is
used, the indications will fluoresce when
exposed to near ultra violet or black light. DYE
58

Part VIII
Appendices

A.
B.
C.
D.
E.
F.
G.

Trigonometry.. A-1
Elements of Sections..A-2
Properties of Circles and EllipsesA-7
Surface Areas and Volumes.A-8
Miscellaneous Formulas...A-10
Properties of Roof and Bottom Shapes...A-12
Conversion FactorsA-13

59

Appendix A. Trigonometry

A-1

Appendix B. Elements of Sections

PROPERTIES OF GEOMETRIC SECTIONS

PROPERTIES OF GEOMETRIC SECTIONS

A = d2
A = bd

c=

d
2

c= d

I=

d4
12

I =

S=
r=

d3
6
d

Bd 2
3
d
r=
= .577380 d
3
S =

= .288675 d

12
Z=

bd 3
3

d3
4
A = bd

A= d

bd

c=
b

c= d
I =
S =
r=

d4
3

I =

d
3
d

= .577350 d

= .707107 d

= .117851 d 3

6 2
d

= .288675 d

12
Z =

bd

A = bd

d4
12

r=

b d

6 b2 + d 3

d2

2 2

r=

2
I=

6 b2 + d 3

6 b2 + d 3

c=

+ d2

b2d 3

S =

A=d

c=

b sin a + d cos a
2

I =

bd b 2 sin 2 a + d 2 cos 2 a
12

S =

bd b 2 sin 3 a + d 2 cos 2 a
6 (b sin a + d cos a )

r=

b 2 sin 3 a + d 3 cos 2 a
12

2c 3
d3
=
= .235702 d 3
3
3 2

A = bd

A = bd b1d1

d
c=
2

c=

d
2

I =

bd 3 b1d13
12

S =

bd 3 b1d13
6d

I =
S =

bd 3
12
bd 2
6
d
12

r=

bd 3 b1d13
12 A

bd 3
Z =
4

Z =

bd 2
bd 2
1 1
4
4

r=

= .288675 d

A-2

Appendix B (Contd)

PROPERTIES OF GEOMETRIC SECTIONS

PROPERTIES OF GEOMETRIC SECTIONS

d (b + b1 )
2
d (2b + b1 )
c=
3 (b + b1 )

A = b (d d1 )
c=

d
2

I=

b d 3 d13
12

S=

b d3
6d

A=

d13

d 3 b 2 + 4 bb1 + b1 2
36 (b + b1 )

S=

d 2 b 2 + 4 bb1 + b1 2
12 (2b + b1 )

r=

d
6 (b + b1 )

d 3 d13

r=

12 (d d1 )

b 2
Z =
d d1 2
4

d4

I=

d3
6

Z=

A t + t1

d
2 2

A=

r=

I=

d 4 d14

S=

d 4 d14

64
32d

) =.785398 (d

) = .049087 (d

) =.098175

d12

d14

d 4 d14
d

d + d12

r=

= .235702 d
Z=

bd
2

d
d2
1
6
6
3

A =

2
= 1.570796 R

c =d

S=

= .049087 d 4 = .785398 R 4

18

I =

d 2 d12
d
2

c=

bd
I=
36

A=

R4

d 3 R3

r=

=
= .098175 d 3 = .785398 R 3
32
4
d R
r= +
4 2

I
A

bd 2
24
d

64

S=

I
S1 +
c1

bd
A=
2
2d
c=
3

S=

2 b 2 + 4bb1 + b12

d
= R
2

Z=

= R 2 = .785398 d 2

c=

bt
bt 3
+ bty 2 + 1 1 + b1t1 y12
12
12

= 3.141593 R 2

1 / 2 bt 2 + b1t1 (d 1 / 2 t1 )
c=
A

I
S=
c

d2

A=

A = bt + b1t2

I =

I =

c = R 1
= .575587 R
3

bd 3
12
bd 2
12
d

8
I = R4
= .109757 R 4
8 9
S=

= .408248 d

R3 (9 2 64)
= .190687 R3
24 (3 4)

r=R

A-3

9 2 64
6

= .264336 R

Appendix B (Contd)

PROPERTIES OF GEOMETRIC SECTIONS

PROPERTIES OF GEOMETRIC SECTIONS

4
ab
3
2
m= a
5
16 3
I1 =
a b
175
4
I 2 = ab3
15
32 3
I3 =
ab
105
2
A = ab
3
2
m= a
5
3
n= b
8
8 3
I1 =
ab
175
19
I2 =
ab3
480
16 3
I3 =
ab
105
2
I 4 = ab3
15

1
ab
2
4a
m=
3
A=

A=

8
11 = a3b

8 9

1
12 = ab 2
8
1 3
13 = a b
8
1
ab
4
4a
m=
3
A=

n=

4
12 = ab 2
16 9

A = ab 1
4

m=

a

6 1
4

n=

b

6 1
4

1
1
11 = a 3b
3 16 36 1

t
2 2
t

2
1 3
A= t
6
4
m =n = t
5
11 4
I1 = I 2 =
t
2100

1
a3b
16
1
14 = ab3
16

1
ab
3
7
m=
a
10
3
n= b
4
37 3
I1 =
a b
2100
1
I 2 = ab 3
80

b=

4
11 = a3b
16 9

13 =

A=

a=

4b
3

1
1

12 = ab3
3 16

36 1

* To obtain properties of half circle, quarter circle and

circular complement substitute a = b = R.

A-4

Appendix B (Contd)

PROPERTIES OF GEOMETRIC SECTIONS

AND STRUCTURAL SHAPES

n = Number of sides
REGULAR POLYGON
Axis of moments
through center

s=

180 o
n

a = 2 R 2 R12

IRREGULAR PLANE SURFACE

a
R=
2 sin
R1 =
A=

a
2 tan
1 2
1
na cot = nR 2
4
2
sin 2 = nR12 tan

I1 = I 2 =

ANGLE
Axis of moments
through
center of gravity

) (

A 6 R 2 a 2 A 12 R12 + a 2
=
24
48

r1 = r2 =

6R2 a2
12 R12 + a 2
=
24
48

tan 2 =

2K
IY I X

A = t (b + c ) x =

b 2 + ct
d 2 + at
y=
2(b + c )
2(b + c )

Divide the plane surface into an even number of

parallel strips of equal width. The given figure has
been divided into ten strips of width, d; the ordinates
are h0 to h10. When the ends are curved, h0 and h10 are
zero and cancel out of formulas.

K = Pr oduct of inertia about

X X &Y Y =

abcdt
+ 4(b + c )

I x = t (d y )3 + by 3 a ( y t )3
3

Simpsons Rule:
Area=

1
I y = t (b x )3 + dx 3 c(x t )3

)]

Area=

I w = I x cos 2 + I y sin 2 K sin 2

BEAMS AND
CHANNELS
Transverse force
oblique
Through center of
gravity

Durands Rule:

I z = I x sin 2 + I y cos 2 + K sin 2

Z-Z is axis of minimum I

d
h0 + h10 + 4(h1 + h3 + h5 + h7 + h9 ) + 2 h2 + h1 + h6 + h8
3

K is negative when heel of angle, with respect

to c.g., is in 1st or 3rd quadrant, positive when
in 2nd or 4th quadrant.

d [0.4(h0 + h10 ) + 1.1(h1 + h9 ) + h2 + h3 + h4 + h5 + h6 + h7 + h8 ]

Trapezoidal Rule:
Area=
1

d (h0 + h10 ) + h1 + h2 + h3 + h4 + h5 + h6 + h7 + h8 + h9
2

I 3 = I x sin 2 + I y cos 2

When the ends are not curved, but are the straight

I 4 = I x cos 2 + I y sin 2

lines h1 and h9 then,

x
= M sin +
cos
Ix

Iy

1
(h1 + h9 ) + h2 + h3 + h4 + h5 + h6 + h7 + h8
2

Area= d

Where M is bending moment due to force F.

A-5

Appendix B (Contd)
Thin Wall Sections (Dimensions are to Center of Wall)

A = dt
A = dt
I=
S=

d 3t
8

d 2t
4

r = 0.355d

b=d
A = 4dt
A=

2d 3t
3

A = 0.408d

d >b

A = 2(b + d )t
d 2t
(3b + d )
6
dt
= (3b + d )
3

I11 =
S11

r11 = 0.289d

3b + d
b+d

Sector of thin annulus

A = 2 Rt
sin

y1 = R1

sin

Y2 = R
cos

A-6

Appendix C Properties of Circles and Ellipses

AREA OF CIRCULAR SECTIONS
Circular Sector, m o n p
Area

= (length of arc, m p n x radius, r)

Circumference

= 6.28318 r = 3.14159 d

Diameter

= 0.31831 circumference

Area

= 3.14159 r2

= area of circle x
arc, m p n, in deg rees
360
= 0.0087266 x square of radius, r2, x
angle of arc, m p n, in degrees

triangle, m o n =

Rise

(length of arc, m p n, x radius, r) (radius, r, - rise, b) x chord, c

2

Circular Segment, m q n, greater than half circle

Area = area of circle area of segment m n p

Ac =

Circular Segment

b
d

b
d

CIRCULAR SECTOR

not

r = radius of circle

y = angle ncp in degrees

Areas of Sector ncpo=1/2 (length of arc nop r)

= Area of Circle

b
= 2 7 5 3 = 0.478528
16
32
d

CIRCULAR SEGMENT

Area =d2 x coeff.=25.91629 x 0.371233 = 9.6321

Area of triangle ncp

( Length of arc nop r ) x(r b )

2
Area of Segment nsp = Area of Circle Area
of Segment nop

Given: rise, b, and chord c.

Area = product of rise and chord, b x c, multiplied by the

b
.
c
b
Intermediate coefficients for values of
c
coefficient given opposite the quotient of

VALUES FOR FUNCTIONS OF

=3.14159265359,

not given in tables are obtained by interpolation.

3.9556

Log=0.4971499

2 = 9.8696044, log = 0.9942997

Example Given: rise = 1.49 and chord = 3.52,

Area = b x c x coeff. = 1.49 x 3.52 x 0.7542 x

r=radius of circle x=chord b=rise

Area of Segment nop = Area of Sector neopo

Circular Segment

Coefficient=0.7542

y
360

= 0.0087266 r2 y

Coefficient by interpolation = 0.371233.

1.5028501

log=1.4914496
1
= 0.1013212, log = 1.0057003

Circular Zone, t u w v

= 1.7724539, log = 0.2485749

Area = area of circle (area of segment, t p u +

180

=0.5641896, log = 1.7514251

Area = segment m p n segment, m s n.

A
2

c
A
b = r 1 / 2 4r 2 c 2 = tan
2
4
2 A
= 2r sin
= r y r 2 x2
4

Example Given: rise=27/16 and diameter=53/x

Circular Lune, m p n s

4b2 + c2
8b

Diameter of circle of equal periphery as square = 1.27324 side of square

Side of square of equal periphery as circle = 0.78540 diameter of circle
Diameter of circle circumscribed about square = 1.41421 side of square
Side of square inscribed in circle = 0.70711 diameter of circle

given in tables are obtained by interpolation.

area of segment, v q w).

a
180 b a
= 57.29578
r
r

x = r 2 (r + y b )2

Area = square of diameter, d2, multiplied by the

b 1.49
=
= 0.4233.
c 3.52

= 0.017483 r Ab

y = b r + r 2 x2

Given: rise, b, and diameter, d = 2r.

Intermediate coefficients for values of

Angle

180o

Chord C = 2 2br b 2 2 r sin

= area of sector, m o n p area of

coefficient given opposite the quotient of

a=

Radius r =

Circular Segment, m p n, less than half circle

Area

r Ab

Arc

180
1

=0.3183099, log =

3 = 31.0062767,

=0.0174533, log=2.2418774

=0.0322515, log = 2.5085500

=57.2957795, log = 1.7581226

Note: Logs of fractions such as 1.5028501 and 2.5085500 may also

be written 9.5028501 - 10 and 8.508550 10 respectively.

PROPERTIES OF THE CIRCLE

A-7

Appendix D Surface Areas and Volumes

SURFACES AND VOLUMES OF SOLIDS
CIRCULAR RING (TORUS)
D and R = Mean Diameter and Mean Radius, respectively, of Ring
d and r = Mean Diameter and Mean Radius, respectively, of Section
Surface = 2 Dd = 42 Rr
Volume = 22 Rr2 =

2
4

Dd 2
PRISMOID

End faces are in parallel planes.

Volume = i(A + A + 4M), where
1 = perpendicular distance between ends
A, A = areas of ends
M = area of midsection, parallel to ends
UNGULAS FROM RIGHT CIRCULAR CYLINDER
(As formed by cutting plane oblique to base)
I.

Base, abc, less than semicircle;

Convex Surface
= h [2re(d x length arc abc)] (rd)
2

Volume = h e 3 (d area base abc ) (r d )

3

II.

Base, abc, = semicircle;

Convex Surface = 2rh

III.

Volume = r2 h

Base, abc, greater than semicircle (figure);

Convex Surface
=h [2re + (d x length arc abc)] (r+d)

2
Volume = h e3 + (d area base abc ) (r + d )

IV.

Base, abc, = circle, oblique plane touching circumference.

Convex Surface = rh

V.

Volume = r2 h

Base, abc, = circle, oblique plane entirely above (figure)

Convex Surface = 2 r
(h, minimum + H, maximum)
Volume = r2 x (h, minimum + H, maximum)

ANY SOLID OF REVOLUTION

Let abcd represent the generating section about axis A-A of solid abef.
Let g at distance h from A-A be the center of gravity of abcd.
Let a be the angular amount of generating revolution.
Then
Total Surface of solid abef
= (2 ha 360) x perimeter abcd
Volume of solid abef = (2 ha 360) x area abcd
For complete revolution (2 ha 360) = 2 h

A-8

Appendix D (Contd)
SURFACES AND VOLUMES OF SOLIDS

SURFACES AND VOLUMES OF SOLIDS

CYLINDER
Convex Surface = dh

SPHERE
Surface = d2 = 4 r2

2
Total Surface = dh + d

Volume =

d2 h
4
Volume Cylinder, right or oblique = area of section
at right angles to sides X length of side.
h
Center of Gravity above Base =
2

PRISM
Lateral Surface = h X Base Perimeter
Total Surface= Lateral Surface + (2 X Base Area)
Volume = h X Base Area
h
Center of Gravity above Base =
2

SPHERICAL SECTOR
Total Surface =
Volume =

PYRAMID

SPHERICAL ZONE
Convex Surface = 2 rh
Total Surface = 2 rh +

Volume =

Volume =

12

dh=
2

24

4s d
2

Volume =

+ d ) =

(d

+ d )

4h

Total Surface = s (d
2

Volume =

h
12

Center

h d

4 d

(d

of

Gravity

+ dd + d

(d

+ d ) +

+ d d + d 2

+ 2 dd + 3d 2
2

d )2

(d

+ d

Volume =

above

base

(l +

(3c2 + 3c2 + 4h2)

4
Rr 2
3

4
R2r
3

Where e=

PARABOLOID

Convex Surface =

R2 r 2
R

r 2
r + 4h 2
6h 2

32

r 3

Total Surface = Convex Surface + r2

Surface = Sum of surfaces of bounding planes

wh
6

Use common or base 10, log.

WEDGE
Volume =

+ c 2

2.303r 2
1 + e
Surface = 2 R 2 +
log .

e
1 e

Convex Surface =
(d

ELLIPSOID (II. Revolution about conjugate axis)

FRUSTUM OF CONE

(c

Sin-1e = Angle, in radians, whose sine = e

h
Center of Gravity above base =
4

24

sin 1 e

Surface = 2 r r + R
e

2
Total Surface = Convex Surface + d

ELLIPSOID (I. Revolution about transverse axis)

ds = d d 2 + 4h 2

c2
2 2
2

r h = r2 r r2
3
3
4

SPHERICAL SEGMENT
Spherical Surface= 2 rh = (c2 + 4h2) 4
Total Surface = Spherical Surface + (c2 4)
Volume = h2 (3r h) 3 = h (3c2 + 4h2) 24
Center of gravity above base of segment =
h (4r h) 4(3r-h)

CONE
Convex Surface =

(4h + c )

3
h
r
4
2

FRUSTUM OF PYRAMID
Lateral Surface = s (Top + Base Perimeters) 2
If a = top area and A = base area.
Total Surface = Lateral Surface + (a + A)
Volume = h (a + A + aA ) 3
Center of Gravity above base =
3 a + A + 2 aA

a + A + aA

Center of Gravity above center of sphere =

s
X Base Perimeter
2
Total Surface = Lateral Surface + Base Area
h
Volume =
X Base Area
3
h
, above base
Center of Gravity =
4
Lateral Surface =

h
4

d3

4
= r3
6
3
Side of an equal cube = diameter of sphere X
0.806
Length of an equal cylinder = diameter of
sphere X 0.6667
Center of Gravity of Half Sphere = 3/8r above
spherical center

Volume =

m + n)

Volume =
base

A-9

r 2h
2

Center of Gravity =

h
above
3

Appendix E
MISCELLANEOUS FORMULAS
1. Area of Roofs.

7. Heads for Horizontal Cylindrical Tanks:

D = diameter of tank in feet.

Umbrella Roofs:
Surface area
in square feet

= 0.842 D2 (when radius=diameter)

= 0.882 D2 (when radius=-0.8 diameter)

Conical Roofs:
Surface areas
in square feet

= 0.787 D2 (when pitch is in 12)

= 0.792 D2 (when pitch is 1 in 12)

Hemi-ellipsoidal Heads have an ellipsoidal cross section,

usually with minor axis equal to one half the major axis-that
is, depth = D, or more.
Dished or Basket Heads consist of a spherical segment
normally dished to a radius equal to the inside diameter of
the tank cylinder (or within a ran e of 6 inches plus or minus)
and connected to the straight cylindrical flange by a
knuckle whose inside radius is usually not less than 6 per
cent of the inside diameter of the cylinder nor less than 3
times the thickness of the head plate. Basket heads closely
approximate hemi-ellipsoidal heads.
Bumped Heads consist of a spherical segment joining the
tank cylinder directly without the transition knuckle. The
radius = D, or less. This type of head is used only for
pressures of 10 pounds per square inch or less, excepting
where a compression ring is placed at the junction of head
and shell.
Surface Area of Heads:

2. Average weights.
Steel
490 pounds per cubic foot, specific gravity 7.85
Wrought iron 485 pounds per cubic foot, specific gravity 7.77
Cast iron
450 pounds per cubic foot, specific gravity 7.21
1 cubic foot air or gas at 32 F., 760 mm. barometer = molecular weight
x 0.0027855 pounds.

3. Expansion in steel pipe = 0.78 inch per 100 lineal feet per 100
degrees Fahrenheit change in temperature = 0.412 inch per mile
per degree Fahrenheit temperature change.

(7a) Hemi-ellipsoidal Heads:

S= R2 [1 + K2 (2-K)]
S=surface area in square feet
R=radius of cylinder in feet
K=ratio of the depth of the head (not including the
straight flange) to the radius of the cylinder
The above formula is not exact but is within limits of practical
accuracy.

4. Linear coefficients of expansion per degree increase in

temperature:

STRUCTURAL STEEL-A-7
70 to 200C ........................
21.1 to 93C .......................
STAINLESS STEEL-TYPE 304
32 to 932C ........................
0 to 500C ..........................
ALUMINUM
-76 to 68F ..........................
-60 to 20C .........................

Per Degree
Fahrenheit

Per Degree
Centigrade

0.0000065
-

0.0000117

(7b) Dished or Basket Heads:

0.0000099
-

0.0000178

(7c) Bumped Heads:

0.0000128
-

0.0000231

Formula (7a) gives surface area within practical limits.

S= R2 [1 + K2 (2-K)]
S, R, and K as in formula (7a)

Volume of Heads:

(7d) Hemi-ellipsoidal Heads:

V=2/3 K R3
R=radius of cylinder in feet
K=ratio of the depth of the head (not including the
straight flange) to the radius of the cylinder

5. To determine the net thickness of shells for horizontal

cylindrical pressure tanks:
6 PD
S
P = working pressure in pounds per square inch
D = diameter of cylinder in feet
S = allowable unit working stress in pounds per square inch
T = Net thickness in inches
Resulting net thickness must be corrected to gross or actual
thickness by dividing by joint efficiency.

T=

(7e) Dished or Basket Heads:

Formula (7d) gives volume within practical limits.

(7f) Bumped Heads:

V=1/2 K R3 (1 + 1/3 K2)

V, K and R as in formula (7d)

Note:

Dished Heads - K = M -

6. To determine the net thickness of heads for cylindrical

pressure tanks:
(6a) Ellipsoidal or Bumped Heads:
6 PD
T=
S
T, P and D as in formula 5
(6b) Dished or Basket Heads:
10.6 P(MR )
T=
S
T, S and P as in formula 5
MR = principal radius of head in feet
Resulting net thickness of heads is both net and gross thickness if
one piece seamless heads are used, otherwise net thickness must
be corrected to cross thickness as above.
Formulas 5 and 6 must often be modified to comply with
various engineering codes, and state and municipal regulations.
Calculated gross late thicknesses are sometimes arbitrarily
increased to provide an additional allowance for corrosion.

K in above formulas may be determined as follows:

Hemi-ellipsoidal heads K is known

(M 1)(M + 1 2m)

Bumped Heads K = M M 2 1

MR = principal radius of head in feet

mR = radius of knuckle in feet
R
= radius of cylinder in feet
MR
mR
m=
M=
R
R
For bumped heads, m=0

8. Total Volume of a Sphere:

V = total volume
D = diameter of sphere in feet
V = -0.523599 D3 Cubic Feet
V = -0.093257 D3 Barrels of 42 U.S. Gallons
Number of barrels of 42 U.S. Gallons at any inch in a true
sphere = (3d-2h) h2 x .0000539681 where d is diameter of
sphere and h is depth of liquid both in inches. The desired
volume must include appropriate ullage for the stored
liquid.

A-10

Appendix E (Contd)
MISCELLANEOUS FORMULAS
(CONTINUED)
9. Total volume or length of shell in cylindrical tank with
ellipsoidal or hemispherical heads:
V
L

= Total volume
= Length of cylindrical
shell
KD = Depth of head
V

D 2

(L + 1 KD)
4
= (V +y)-1/iKD

10. Volume or contents of partially filled horizontal

cylindrical tanks:
(10a) Tank cylinder or shell (straight portion only)

sin cos
Q

Q = partially filled volume or contents in cubic feet

R
= radius of cylinder in feet
L
= length of straight portion of cylinder in feet
The straight portion or flange of the heads must be considered a
part of the cylinder. The length of flange depends upon the
diameter of tank and thickness of head but ranges usually
between 2 and 4 inches.
A
= R = depth of liquid in feet

= i
= a ratio
Cos = 1 - , or

= degrees
(10b) Hemi-ellipsoidal Heads:
Q
= 3/4V 2 (1-1/3 )
Q
= partially filled volume or
contents in cubic feet
V
= total volume of one head
per formula (7d)

= k
= a ratio
a
= R = depth of liquid in feet
R
= radius of cylinder in feet
o
= R 2 L
180

(10c) Dished or Basket Heads:

Formula (lob) gives partially filled volume within
practical limits, and formula (7d) gives V within
practical limits.
(10d) Bumped Heads:
Formula (lob) gives partially tilled volume within
practical limits, and formula (7f) gives V.
Note: To obtain the volume or quantity of liquid in partially filled
tanks, add the volume per formula (10a) for the cylinder or
straight portion to twice (for 2 heads) the volume per formula
(10b), (10c) or (10d) for the type of head concerned.

11. Volume or contents of partially filled hemi-ellipsoidal

heads with major axis vertical:
Q = Partially filled volume
or contents in cubic feet
V = Total volume of one head
per formula (7d)
R = Radius of cylinder in feet
(11a) Upper Head
Q = 1 V (I - 2)
a
=
= a ratio
KR
A = KR = depth of liquid in feet
(11b) Lower Head:
Q = 1 V 2 (1 - )
a
=
= a ratio
KR
A = KR = depth of liquid
in feet

A-11

D2 X 2

4
3

3pD

1.3 H + D
2

0.3125D

0.2878D
0.2083D

0.1919D

D2 4X 2

4
3

A-12

D
Calculate

0.3183W r 2

subtract

D2
4

on basis (h-x) and

0.0796WD
h

Calculate new vol.

6pr

2.6r (H + h)

8rh 3h 2
12r 4h

5
h (roughly)
8

Calculate

30

0.276W

0.0451D

0.056D

1.0472D
0.0906D2

Calculate angle

0.8418D2

0.539D3
0.4031D3

38.67

0.198W

0.0596D

0.0755D

0.1195D2

1.080D

0.8822D2

0.07175D3
0.5367D3

0.1755D

90

0.4543D

0.4464D2

1.6661D2

0.2777D3
2.0772D3

0.707D

90

0.6602D

0.5639D2

1.9635D2

0.3272D3
2.4481D3

CONISPH.

CONISPH

SHAPES

0.134D

60

90

STD. UMBRELLA

Calculate sector -

2 rh

1.0472h2 (3r-h)
7.833h2 (3r-h)

SEGMENTAL

vol. - vol.

6 pDa
h

2.6 H Da
h

3h
4

2h
3

0.2618D2h
1.9584D2h
Da
2
Dh
2

CONE

90
90
Calculate
H = water elev. Above belt line (Shell Height).
W = total load carried, including dead load.

D2 3 X 2

4
4

-6pD

3 pD
or
4

+6pD or

+ 2.6 D H +
4

2.6 D H +
3

0.1563D

0.1439D

+4.5pD

+ 1.95 D H + or
3

H D
2.6 D +
4
8

Angle at edge
90
NOTE: All dimensions expressed in feet;

(cu. ft.)
Belt Line
Stress
(pounds)

Within depth X

Partial Volume

Stress due to
Gas pressure
p lbs per
inch

Lbs. per inch

Stress (water)

of Mass

y to Centroid

Proj. Ar.

y to Centroid

1.211D
0.1964D2

1.322D
0.2618D2

1.5708D

1.084D2

0.3927D2

1.24D2

Length of Arc
Projected
Area

D
4
0.1309D3
0.9792D3

ELLIP.

1.571D2

D
4

Surface sq. ft.

D
3
0.1745D3
1.3056D3

ELLIP.

Volume, cu. ft.

Volume, gals.

D
3

D
2
0.2618D3
1.9584D3

Depth or Rise

HEM.

Appendix F. Properties of Roof and Bottom Shapes

90

0.1000D

0.1255D2

1.1043D

0.9286D2

0.0810D3
0.6059D3

0.169D

D.R. = D
K.R. = .06D

F and D
HEAD

A-13

Appendix G. Conversion Factors

Appendix G (Contd)

SI CONVERSION FACTORSa
Qty

Length

Multiply

inch
foot
yard
mile(U.S.
Statute)

millimetre
metre
metre
kilometre
Area

39.370 079 x10-3

3.280 840
1.093 613
0.621 370

inch
foot
yard
mile

square inch

square foot
square yard
square
mile
(U.S.Statute)
acre
acre

cubic inch

0.645 160 x 103

in
ft
yd
mi

4.046 873 x 103

0.404 687
1.550 003 x 10-3

square inch in2

10.763 910
1.195 990
0.386 101

square foot ft2

square yard yd2
square mile mi2

0.247 104 x 10-3

2.471 044

acre
acre

16.387 06 x 103

cubic
millimetre mm3
cubic meter m3
cubic meter m3
litre
l

0.092 903
0.836 127
2.589 998

28.316 85 x 10
0.764 555
3.785 412

cubic
millimetre
cubic metre
cubic metre
litre

61.023759 x 106

-3

0.946 353

35.314 662
1.307 951
0.264 172
1.056 688

ounce
(avoirdupois)
pound
(avoirdupois)
short ton

mm
m
m
km

square
millimetre mm2
square meter m2
square meter m2
square
kilometre km2
square meter m2
hectare

cubic foot
cubic yard
gallon(U.S
liquid)
quart(U.S.
liquid)

litre
Mass

to obtain

millimetre
metre
metre
kilometre

square
millimetre
square metre
square metre
square
kilometre
square metre
hectare
Vol

by

25.400
b 0.304 800
b 0.914 400
1.609 347
b

SI CONVERSION FACTORSa

28.349 52
0.453 592
0.907 185 x 10-3

litre
cubic inch

in3

kilogram

kg

kilogram

kg

ounce
oz av
(avoirdupois)
kilogram
2.204 622
pound
lb av
(avoirdupois)
kilogram
1.102 311 x 10-3 short ton
a Refer to ASTM E380-79 for more complete information on SI.
b Indicates exact value.
gram

35.273 966 x 10-

Multiply
ounce-force
pound-force

0.278 014
4.448 222

newton
newton

newton
newton
pound-force-inch

3.596 942
0.224 809
0.112 985

pound-force-foot

1.355 818

newton-metre

8.850 748

newton-metre

0.737 562

pound-force per
sq inch
foot of water
(39.2 F)
inch of mercury
(32 F)
kilopascal

6.894 757

ounce-force
pound-force lbf
newtonmetre
Nm
newtonmetre
Nm
pound-force
inch
lbfin
pound-force
foot
lbfft
kilopascal kPa

2.988 98

kilopascal kPa

3.386 38

kilopascal kPa

0.145 038

kilopascal

0.334 562

kilopascal

0.295 301

pound-force
per sq. inch
lbf/in2
foot of water
(39.2 F)
inch of
mercury (32 F)
joule
J
joule
J

Bending
Moment

Pressure
Stress

Energy,
Work,
Heat

foot-pound-force
cBritish
thermal unit
ccalorie
kilowatt hour
joule

by

1.355 818
1.055 056 X 103
b
b

4.186 800
3.600.000 X106
0.737 562

joule

0.947 817 X 10-3

joule
joule
foot-pound-force/
second
cBritish thermal
unit per hour
horsepower (550
ft. lbf/s)
watt

0.238 846
0.277 778 X 10-6
1.355 818

cubic foot
ft3
cubic yard yd3
gallon (U.S. liquid)
gal
quart (U.S. liquid)
qt
gram

Qty
Force

Power

to obtain

joule
J
joule
J
foot-pound
force
ftlbf
cBritish
thermal
unit
Btu
ccalorie
kilowatt hourkW
watt
W

0.293 071

watt

0.745 700

kilowatt

W
kW

foot-poundforce/
sec.
ftlbf/s
cBritish
watt
3.412 141
thermal
unit/hr Btu/h
kilowatt
1.341 022
horsepower hp
(550 ftlbf/s)
Angle
degree
17.453 29 X 10-3 radian
rad
radian
57.295 788
degree
Temper Fahrenheit
tC=(tFx32)/1.8
degree
ature
Celsius
Celsius
tF=1.8 x tC+32 degree
Fahrenheit
a Refer to ASTM E380-79 for more complete information on SI.
b Indicates exact value.
c International Table

A-14

0.737 562

N
N

Appendix G (Contd)
SPECIFIC GRAVITY AND WEIGHTS OF
VARIOUS LIQUIDS
Liquid

Acetaldehyde
Acetic Acid
Acetic Anhydride
Acetone
Aniline
Asphaltum
Bromine
Carbon Disulfide
Carbon Tetrachloride
Castor Oil
Caustic Soda,66% Solution
Chloroform
Citric Acid
Cocoanut Oil
Colza Oil (Rape Seed Oil)
Corn Oil
Cottonseed Oil
Creosote
Dimethyl Aniline
Ether
Ethyl Acetate
Ethyl Chloride
Ethyl Ether
Formaldehyde
#1 Fuel Oil
#2 Fuel Oil
#4 Fuel Oil
#5 Fuel Oil
#6 Fuel Oil
Furfural
Gasoline (Motor Fuel)
Glucose
Glycerin
Hydrochloric Acid, 43.4% Sol
Kerosene
Lactic Acid
Lard Oil
Linseed Oil Raw
Linseed Oil Boiled
Mercury
Molasses
Naphthalene
Neatsfoot Oil
Nitric Acid, 91% Sol.
Olive Oil
Peanut Oil
Phenol
Pitch
Rosin Oil
Soy Bean Oil
Sperm Oil
Sulfur Dioxide
Sulfuric Acid,87% Sol.
Tar
Tetrachloroethane
Trichloroethytene
Tung Oil
Turpentine
Water (Sea)
Water (0C)
Water (20C)
Whale Oil

At
Temp.
of F
64.4
68.0
68.0
68.0
68.0
68.0
68.0
68.0
68.0
59.0
68.0
68.0
68.0
59.0
68.0
59.0
60.8
59.0
68.0
77.0
68.0
42.8
77.0
68.0
60.0
60.0
60.0
60.0
60.0
68.0
60.0
77.0
32.0
60.0
68.0
59.0
59.0
68.0
59.0
68.0
68.0
68.0
59.0
68.0
59.0
59.0
77.0
68.0
68.0
59.0
59.0
80.0
64.4
68.0
68.0
68.0
59.0
68.0
59.0
39.2
68.0
59.0

Specific
Gravity

0.783
1.049
1.083
0.792
1.022
1.1-1.5
3.119
1.263
1.595
0.969
1.70
1.489
1.542
0.926
0.915
0.921-0.928
0.926
1.040-1.100
0.956
0.708
0.901
0.917
0.712-0.714
1.139
0.80-0.85
0.81-0.91
0.84-1.00
0.91-1.06
0.92-1.08
1.159
0.70-0.76
1.544
1.260
1.213
0.82
1.249
0.913-0.915
0.93
0.942
13.595
1.47
1.145
0.913-0.918
1.502
0.915-0.920
0.917-0.926
1.071
1.07-1.15
0.98
0.924-0.927
0.878-0.884
1.363
1.834
1.2
1.596
1.464
0.939-0.949
.0.87
1.025
1.00
0.998
0.917-0.924

Weight in Lbs.
Per
U.S. Gal.
6.52
8.74
9.02
6.60
8.51
9.2-12.5
25.98
10.52
13.28
8.07
14.16
12.40
12.84
7.71
7.62
7.67-7.73
7.71
8.66-9.2
7.96
5.90
7.50
7.64
5.93-5.95
9.49
6.7-7.1
6.7-7.6
7.0-8.3
7.6-8.8
7.7-9.0
9.65
5.8-6.3
12.86
10.49
10.10
6.83
10.40
7.60-7.62
7.8
7.84
113.23
12.2
9.54
7.60-7.65
12.51
7.62-7.66
7.64-7.71
8.92
8.91-9.58
8.61
7.70-7.72
7.31-7.36
11.35
15.27
10.0
13.29
12.19
7.82-7.90
7.25
8.54
8.34
8.32
7.64-7.70

Weight in
Lbs. Per
Cu. Ft.
49
65
68
49
64
69-94
195
79
100
60
106
93
96
58
57
57-58
58
65-69
60
44
56
57
44-45
71
50-53
51-57
52-62
57-66
57-67
72
44-47
96
79
76
51
78
57
58
59
849
92
71
57
94
57
57
73
67-72
61
58
55
85
114
75
100
91
59
54
64
62.4
62.3
57

The parameters given are approximate for estimating purposes only. The properties of the stored liquid should be determined
analytically and used in the final design.

A-15

Appendix G (Contd)
A.P.I. AND BAUME GRAVITY AND WEIGHT FACTORS
The relation of Degrees Baume or A.P.I. to Specific
Gravity is expressed by the following formulas:
For liquids lighter than water:
Degrees Baume =

140
140
-130, G =
G
130 + Degrees Baume

Formulas are based on the weight of 1 gallon (U.S.) of oil

with a volume of 231 cubic inches at 60 degrees Fahrenheit in air
at 760 m.m. pressure and 50% humidity. Assumed weight of 1
gallon of water at 60 Fahrenheit in air is 8.32828 pounds.
To determine the resulting gravity by mixing oils of different
gravities:

md1 + nd 2
m+n
D = Density or Specific Gravity of mixture
m = Proportion of oil of d, density
n = Proportion of oil of d, density
d1 = Specific Gravity of m oil
d2 = Specific Gravity of n oil
D=

141.5
141.5
-130, G=
Degrees A.P.I. =
G
131.5 + Degrees A.P.I .
For liquids heavier than water:
145
145
, G=
Degrees Baume = 145 G
145 Degrees Baume
G = Specific Gravity = ratio of the weight of a given volume
of oil at 60 Fahrenheit to the weight of the same volume of
water at 60 Fahrenheit.

PRESSURE EQUIVALENTS
PRESSURE
= 2.31 ft. water at 60F
= 2.04 in. hg at 60F

1 lb. per sq. in.

1 ft. water at 60F

= 0.433 lb. per sq. in.

= 0.884 in. hg at 60F

1 in. Hg at 60F

= 0.49 lb. per sq. in.

= 1.13 ft. water at 60 F

lb. per sq. in. Absolute (psia)

= lb. per sq. in. gauge (psig) + 14.7

A-16

Appendix G (Contd)
WIRE AND SHEET METAL GAGES
Equivalent thickness in decimals of an inch

US Standard
Gauge for
Uncoated
Hot
and Cold
Rolled
Sheets b

Galvanized
Sheet Gauge
For HotDipped
Zinc-Coated
Sheets b

USA Steel
Wire
Gauge

13

.0897

.0934

.092 a

.462 a

14

.0747

.0785

.080

.430

15

.0673

.0710

.072

.394 a

16

.0598

.0635

.062 a

.362 a

17

.0538

.0575

.054

2/0

.331

18

.0478

.0516

.048 a

1/0

.306

19

.0418

.0456

.041

.283

20

.0359

.0396

.035 a

.262 a

21

.0329

.0366

.2391

.244

22

.0299

.0336

.2242

.225 a

23

.0269

.0306

.2092

.207

24

.0239

.0276

.1943

.192

25

.0209

.0247

.1793

.177

26

.0179

.0217

.1681

.1681

.162

27

.0164

.0202

Gauge
No.

U.S.
Standard
Gauge for
Uncoated
Hot
and Cold
Rolled
Sheetsb

Galvanized
Sheet Gauge
for
Hot-Dipped
Zinc-Coated
Sheetsb

USA
Steel
Wire
Gauge

Gauge
No.

7/0

.490

6/0

5/0

4/0

3/0

.1495

.1532

.148

28

.0149

.0187

10

.1345

.1382

.135

29

.0172

11

.1196

.1233

.120 a

30

.0157

12
.1046
.1084
.106 a
a Rounded value. The steel wire gauge has been taken from ASTM A510 General Requirements for Wire Rods and Coarse Round
Wire, Carbon Steel. Sizes originally quoted IO 4 decimal equivalent places have been rounded to 3 decimal places in accordance with
rounding procedures of ASTM Recommended Practice E29.
b The equivalent thicknesses are for information only. The produce is commonly specified to decimal thickness, not to gauge number.

A-17

CoverInQuark:Layout 1

12/19/11

10:40 AM

Page 1