\

VESSEL DES I G N

u z
vi'

0
VI

~ ..a

""
z

i :::;
::c

1i

::c

...

"" a
c E
Co

.c

~ GIl:

>~ LLI
Z

0 "IS

c
C

..a

LLOYD E. BROWNELL
Professor of Chemicol and Nuclear University Engineering of Michigan

EDWIN H. YOUNG
Associate of Chemical and Metallurgical University Professor Engineering of Michigan

IP,.

..

= v;:jg",,__ ".

-.-. .

"

r7.1

-_~.,

.1'.

••

_~

• ..,...,.

·i~

l.' , '"

~
",

(gpy,r~M

,® lr9:5~' by Jt~m
nGl!i¥ed~

WII~'

t, SI;1lJlll.ne •

..tU Jii~lti,
iIi1~
w'.i~i

'!r!l!:iIiI:

:JI.(It ~ ropmdeoed d)" iOO!'! IIi'Ile WiIillml ~ii(o:rmWIN!, !fIf II1.ol! p'U\;J'lC!Ih~']'.

:lt1t;

'boo!;

,00

.RtlJ' p.1i:d

Ul:Kml')l' of C~{I1~i!!in 'C(ilhi:~!ggCard Nwirih;j!if~ ,:S~:(I.Ii!I2

...._

Primwd I'll "frill [1J"!~e-~SIGI~, of Arm·.~'~

'l00 ,o"~r WIVEloi .

I

Ja neili" B'i'O'WlIlH~:I:i

and
,S:iglne' Y CHl!llliIig

1 110 I

(ill

PREFACE

This book was prepared primarily for senior and graduate students in engineering. The needs of design engineers and consultants as well as those of students were considered in selecting the topics and methods of presentation. The book is based upon our experiences gained in industrial design offices and in 16 years of teaching courses in equipment design at the University of Michigan. We both have supervised research and development of process equipment, and have acted as consultants in this field. The book was originally prepared as class notes, which have been used for about ten years in teaching courses in process equipment design at the senior and graduate levels in the Chemical and Metallurgical Engineering Department of the University of Michigan. Typical problems have involved the design of fractionating towers, trays, vacuum crystallizers, condensers, heat exchangers, high-pressure reactors, and other types of process equipment. The design of process equipment requires a thorough knowledge of the functional process, the materials involved, and the methods of fabrication. The design factors to be considered are many and varied and, in most cases, so interwove~ that exact methods of attack are often impossible to formulate. Compromises are necessary and the design engineer often has only experience in similar or related fields to guide him in his choice. Thus, the engineer must realize that considerable engineering judgment is required in applying all recommended specific methods of design. One purpose of this book is to consolidate the basic concepts, industrial practices, and theoretical relationships useful in the design of processing equipment. Many of these considerations and much of this vital information are widely scattered throughout the technical literature, industrial bulletins, appropriate codes, and handbooks. It is not intended that this book should cover all the ramifications of design problems, but it will serve as a guide to the student and the practicing engineer for efficient and economical design of equipment for the processing industries.
vii

viii

Preface

The organization is based on the premise that the vessel is the basic part of most types of processing equipment. For example, a heat exchanger or evaporator is a vessel with tube bundles and a fractionating tower is a vessel with trays. The first 12 chapters are concerned in part with the development of fundamental relationships on which many of the code specifications are based. Chapter 13 is concerned entirely with code practice and covers selected code specifications not covered in the earlier chapters. Chapters 14 and 15 are concerned with the design of vessels beyond the scope of the ASME code. The sequence of chapters was selected to permit the introduction of a brief review of elementary theories of mechanics and strength of materials early in the book. More advanced theory is developed as needed in subsequent chapters. The integration of theory with practice in design eliminates the necessity of a separate section on engineering mechanics. The sequence of presentation allows for an orderly development of theoretical relationships when the book is being used as a textbook in teaching design. The material presented covers the range from simple vessels for low-pressure service to thick-walled vessels for highpressure applications. The experienced designer will find the book useful as a reference in a design office. In all but a few cases derivations of equations and the method of analysis have been given so that the engineer will understand the assumptions and limitations involved. Also, example calculations and designs have been included to illustrate the use of the relationships and recommended procedures. We wish to acknowledge the assistance given by a large number of individuals and companies in providing subject material and illustrations on process equipment design and in making reviews and suggestions. We are particularly indebted to the following: C. E. Freese, Mechanical Consultant, and B. B. Kuist, The Fluor Corporation; W. R. Burrows, Chief Engineer, Manufacturing Department, Standard Oil Company of Indiana; A. E. Pickford, Department Head, Apparatus Design, C. F. Braun and Company; H. B. Boardman, Director of Research, L. P. Zick, Research Engineer, and E. N. Zimmerman, Chicago Bridge and Iron Company; W. T. Gunn and Walter Samans, American Petroleum Institute; J. M. Evans, Chief Engineer, and F. L. Maker, Standard Oil Company of California; R. S. Justice, Chief Engineer, Gulf Oil Corporation; F. L. Plummer, Director of Engineering, Hammond Iron Works; W. D. Kinsell, Manager, Construction Engineering Department, The Pure Oil Company; G. E. Fratcher, Director of Engineering, A. O. Smith Company; F. E. Wolosewick, Sargent and Lundy Engineers; P. E. Franks, Chief Engineer, Sinclair Refining Company; D. W. Carswell and H. B. Peters, Chief Engineer, The Texas Company; W. T. Brown, Manager, Mechanical Division, and Harry Wearne, Construction Manager, Shell Oil Company; F. J. Feeley, Jr., Assistant Director, Engineering Design Division, Esso Research and Engineering Company; J. H. Faupel, E. 1. du Pont de Nemours and Company; W. H. Funk, Lukens Steel Company; and the following additional companies and organizations: Horton Steel Works, Ltd.; BlawKnox Company; Graver Tank and Manufacturing Company; American Cyanamid Company; Inland Steel Company; Ryerson Steel Company; Taylor Forge and Pipe Works; Aluminum Company of America; M. W. Kellogg Company; American Standard Association, Inc.; The Girdler Company, Inc.; Baldwin-LimaHamilton Corporation; Bethlehem Steel Company, Inc.; United States Department of Interior, Bureau of Mines; Great Lakes Steel Corporation; McGraw-Hill
/'

. (

Preface

ix

Book Company, Inc.; Universal-Cyclops Steel Corporation; and the United States Steel Corporation. We also wish to express our appreciation to the American Society of Mechanical Engineers and the American Petroleum Institute for permission to use selected material from the 1956 edition of the Unfired Pressure Vessel Code and the API Specification for Welded Oil Storage Tanks and Production Tanks, respectively. We are also indebted to Dr. J. McKetta, Mr. F. L. Standiford, Dr. H. H. Yang, and Dr. M. D. S. Lay, who assisted in the preparation of the course notes while enrolled in the Graduate School of the University of Michigan, and to Professor Donald L. Katz, Chairman, Department of Chemical and Metallurgical Engineering, University of Michigan, for encouragement and advice in the preparation of this book. Many individuals have given valuable suggestions, comments, and assistance in the preparation of this book and any omissions in acknowledgment are not intended.
LLOYD EDWIN

E.
H.

BROWNELL YOUNG

Ann Arbor, Michigan April,1959

/'

CONTENTS

Chapter

1 ~2 3 4 5 6 7 8

Factors Criteria

Influencing

the Design of Vessels

1 19

in Vessel Design Cylindrical Vessels Cylindrical

Design of Shells for Flat-Bottomed Design of Bottoms Vessels and Roofs for

36 58

Flat-Bottomed for Cylindrical of Flat-Plate

Proportioning and Head Selection Formed Closures

Vessels with 76 and Conical

Stress Considerations in the Selection Closures for Cylindrical Vessels

98
120 141 155 183 Supports 203 219

;J

Stress Considerations in the Selection of Elliptical, Torispherical, and Hemispherical Dished Closures for Cylindrical Vessels Design of Cylindrical Vessels with Formed Closures under External Pressure Design of Tall Vertical Design of Supports Design of Horizontal Design of Flanges Design of Pressure High-Pressure Multilayer Vessels to Code Specifications Vessels Vessels Operating

9
10

for Vertical Vessels Vessels with Saddle

o 11
c)

12 13 14 15

.~

249 268 296 317

Monobloc

Vessels

References Appendix A. Design Conventions Welding Conventions

xi

323 327

B.

xii

Contents

c.
o

Pricing of Steel Plate Allowable Stresses

330 335 346 349

Rolled Structural Members

D. E. F.

Typical Tank Sizes and Capacities Shell Accessories Properties of Selected

"

G. H. I.

353 362

Values of Constant

C of Eq. 13.27
of Cylindrical and

Charts for Determining Shell Thickness Spherical Vessels under External Pressure Properties Properties Strength of Various Sections of Pipe of Materials

364
381 386

(!

J.

and Beam Formulas

-::- K. c: L. Author Index Subject Index

392 395 399

CHAPTER

FACTORS INFLUENCING THE DESIGN O-F VESSELS

/'

rF>

1.1

SELECTION OF THE TYPE OF VESSEL

~hemical engineering invotj:s~ ~he application of the:;, Usually the first step in the design of any vessel is the sciences to the process industries which are primarily conselection of the type best suited for the particular service ~,','cerned with the conversion of one material into another by in question. The primary factors influencing this choice '11' chemical or physical means. These processes require the are: the function and location of the vessel, the nature of the • handling and storing of large quantities of materials in confluid, the operating temperature and pressure, and the neces:.•) tainers of varied construction, depending upon the existing sary volume for storage or capacity for processing. Vessels .~' state of the material, its physical and chemical properties, may be classified according to functional service, temperaand the required operations which are to be performed. ture and pressure service, materials of construction, or For handling such liquids and gases a container, or "ye§Sell' geometry of the vessel. is used. The vessel is the basic part of most types of procThe most common types of vessels may be classified essing equipment. Most process equipment units may be according to their geometry as: considered to be vessels with various modifications neceslOt nk sary to enable the units to perform certain required func2' FIPetnbatt s. d ti I I' dri I t k . .' a - 0 orne ,ver ICa cy In rIca an s. tions. For example, an autoclave may " considered --;---"'7"/-'-'0 3 • VertiIca I cy I' dri I an d hori be to be e ith e • •• In rIca orlzon t a I vesse Is WI torrne d a high-pressure vessel equipped with agitation and heating .d . ","--<:41 en 4 s. S h . I sources; a di '11' isti ation or a b sorption coIumfi may b e consiIddif d heri I I . .. . f' linuid . P enca or mo I te sp enca vesse s. d to b e a vesse I containing a aeries 0 vapor- IqUI conere Vessels in each of these classifications are widely used as tactors; a heat exchanger may be considered to be a vessel storage vessels and as processing vessels for fluids. The containing a suitable provi~~n for the transfer of heat ....... range of service for the various types of vessels overlaps, through tube walls; and an evaporator may be considered and it is difficult to make distinct classifications for all to be a vessel containing a heat exchanger in combination applications. /v' with a Y!lE.~r-disengagin~~~ W~/.~/i, It is possible to indicate some -generalities in the existing lj/ Regardless of the nature of t~lication of the vessel, a num~ortact~~s usua'lly-~~t be considered in designing uses of the common types of vessels. Large volumes of . the unit. The most important consideration often is the nonhazardous liquids, such as briIK!afld other aqueous solutions, may be stored in ponds if of'very low value, or in open ~ selection of the type of vessel that performs the required steel, wooden, or concrete tanks if of greater value. If the "S .service in the most satisfactory manner. In developing the .~.~ <fesigna number of' other criteria must be considered, such fluid is toxic, combustible, or gaseous in the storage condition, or if the pressure is greater than atmospheric, a closed as the properties of,t?-~ material used, the ind~c;,e?" stresses, system is required. For storage of fluids at atmospheric the elastic stability, and the aesthetic appeatlln& of the pressure, cylindrical tanks with flat bottoms and coniunit. Th~kQst Q,f tb.e fa bri.ca.~e§.§~lis.llJ~Qj!!lPQrtllnt.i.D. cal roofs are commonly used. Spheres or spheroids are rel~ ...1Qjgt~e.J:.Y.iC.fUlll.dllSefuLliie.

,'"

Factors Influencing the Design of Vessels

o
I

N ------_j--------

2' _["

Inside diameter

10'-0'

Double fillet butt weld eff. 95%

Fig. 1.1.

10'-0" Unfired Pressure Vessel' 150 Ib.sq in. at 850°F Example of a cylindrical vessel with formed ends designed to the original API-ASME code.

(Courtesy of Amer. Pet. Inst.'

.0

. employed for pressure storage where the volume required is large. For smaller volumes under pressure, cylindrical tanks with formed heads are more economical. ·~I<" a Open Vessels. Open vessels are commonly used 1.1 as surge tanks between operations, as v~ts for batch operations where materials may be mixed and blended, as settling r ~ tanks, decanters, chemical reactors, reservoirs, and so on. \~) Obviously, this type of vessel is cheaper than covered or "' Y closed vessels of the same capacity and construction. The • 'J decision as to whether or not open vessels may be used ~ depends upon the fluid to be handled and the operation. Veiy large quantities of aqueous liquids of low value may be stored inponds, It is doubtful if ponds may be correctly referred to as vessels. (They are, however, the simplest

ie .

containers made from the cheapest of materials, rolled earth. Not all types of earth can he used for storage ponds; a clay which will form an almost watertight bottom is essential. An example of the use of ponds of rolled earth is found in the process whereby salt is crystallized from sea water by solar evaporation (1).. 'When more valuable fluids are handled, more reliable but more ~nsive containers are required. Large circular tanks of stee~) or reinforced (or prestressed) \concrete (3), (4) are often used for settling ponds in which a slowly rotating rake removes sediment from a slightly {J _inclinedconical bottom. Vessels of this t~. as exemplified jf.'by the Dorr classifier, may have diameters ranging from 100 to 200 ft and a depth of several feet. Smaller open vessels are usually o.f a circular shape and

(tV

II

Selection of the Type of Vessel are constructed of mild carbon steel, concrete, and sometimes of wood (5). Other materials find limited use where serious corrosion or contamination problems are encount:U)~(J ered. However, in the process industries in general, the major portion of existing vessels are constructed of steel because of its low initial cost and ~. of fabrication. In many cases such vessels are lined with l~-grass; t.1> or plastic to improve resistance to corr~n. In the food I / industry fir is commonly used for pickle and kraut tanks, ,t ~J' //whereas quarter-sawed white oak is employed for wine and, -! { o spirits. Redwood or cyprus tanks are often employed for water storage reservoirs. Wood is also used in place of , steel for handling dilute solutions of hydrochloric, lactic, and _~~'~aceii~ aC1asand-s~lt solutions and is indispensable as a low- " cost tank in the tanning, brewing, and pickling ind~ries (6t.f I;;lnt~nd pharmaceutical industries it is often neces~. 'sary to add materials to open vessels in the preparation of mixtures. Small open tanks or kettles are usually employed for such purposes. Glass-lined steel, copper, Monel, and stainless steel tanks are widely used in these applications to resist corrosion and prevent contamination of the process " materials. ~I 1.1b Closed Vessels. Combustible fluids, fluids eIDiiting toxic or obnoxious fumes, and gases must be stored in closed vessels (7). Dangerous chemicals, such as acid or caustic, are less hazardous if stored in closed vessels. The combustible nature of petroleum and its products necessitates the use of closed vessels and tanks throughout the petroleum and petrochemical industries. The extensive use of tanks in this field has resulted in considerable effort on the part of the American Petroleum Institute to standI ~,

j~

.) I

ardize design for purposes of safety and economy. Tanks used for the storage of crude oils and petroleum products are generally designed and constructed in accordance with API Standard 12 C, API Specification for Welded OilStorage Tanks. This is the standard reference used in designing tanks for the petroleum industry, but it is also a useful guide for other applications. CYLINDRICAL VESSELS WITHFLATBOTTOMS ANDCONICAL OR DOMED ROOFS. The most economical design for a closed vessel operating at atmospheric pressure is the vertical cylindrical tank with a conical roof and a flat bottom resting directly on the bearing soil of a foundation composed of sand, g,tayel, or crushed rQg. In cases where it is desirable to use a gravity ked, the tank is raised above the ground, and the flat bottom may be supported by columns and wooden joists or steel beams. Cylindrical, flat-bottomed, cone-roofed tanks are provided with "breathers" or vents which permit expansion and cOlUl:aeQon the fluids of as a result of temperature and volume flU£ltuQ*jons. Tanks up to 24 ft in diameter may be covered with a self-supporting roof; tanks with larger diameters, up to 48 ft, usually require at least one central column for support. Tanks larger than 48 ft in diameter are frequently designed with multiplecolumn supports or with a floating or pontoon roof which rises and falls with the level of liquid in the vessel. In general, tanks with conical roofs are limited to essentially atmospheric pressure. If domed roofs are used, pressures from 272 to 15 lb per sq in. gage may be permitted. These vessels are normally smaller in diameter and of greater height for a given capacity than tanks with conical roofs (8, 9).

Fig. 1.2.

Oil reflnery installation.

(Courtesy of C. F. Braun & Company,)

Factors Influencing the Design of Vessels

CYLINDRICAL VESSELS WITH FORMED ENDS. Closed cylindrical vessels with formed heads on both ends are used where the vapor pressure of the stored liquid may dictate a stronger design. Codes have been developed through the efforts of the American Petroleum Institute (10) and the American Society of Mechanical Engineers (11) to govern the design of such vessels. These vessels are usually less than 12 ft in diameter if they are to be shipped by rail. However, field-erected vessels may exceed 35 ft in diameter and 200 ft in length. If a large quantity of liquid is to be stored, a battery of vessels may be used. A variety of formed heads are used for closing the ends of cylindrical vessels. The formed heads include the hemispherical, elliptical-dished, torispherical, standard-dished,

one type or another. Figure 1.2 shows a wide variety of such items in a petroleum refinery. Note that nearly all of the processing equipment shown consists of cylindrical vessels with formed ends. SPHERICALANDMODIFIED SPHERICALVESSELS. Storage containers for large volumes under moderate pressure are usually fabricated in the shape of a sphere or spheroid. Capacities and pressures used in this type of vessel vary greatly. Capacity ranges from 1000 to 25,000 bbl, and pressures range from 10 lb per sq in. gage for the larger vessels to 200 lb per sq in. gage for the smaller ones. Figure l.3 shows a battery of horizontal cylindrical vessels and spherical vessels for storing petroleum products at pressures up to 100 lb per sq in. gage.

,
Fig. 1.3. Spherical and horizantal storage tonics at Crown Central Petroleum Plant near Houstan, TexaL
r

(Courtesy of Hammond Iron Works.)

",

conical, and toriconical shapes. For special purposes flat plates are used to close a vessel opening. However, flat '1N' heads are fa.rely used for large vessels. For pressures not covered by the ASME code, the vessels are often equipped with standard dished heads, whereas vessels that require code construction are usually equipped with either the ASME-dished or elliptical-dished heads. The most common shape for the closure of "pressure vessels" is the elliptical dish. Figure 1.1 shows a drawing of a vertical cylindrical vessel with formed ends designed to the original API-ASME code. Most chemical and petrochemical processing equipment such as distilling columns, desorbers, absorbers, scrubbers, heat exchangers, pressure-surge tanks, and separators are essentially cylindrical closed vessels with formed ends of

under pressure, it is obvious that the required storage volume will be inversely proportional to the storage pressure. In general, for a given mass the spherical type of tank is more economical for large-volume, low-pressure storage operation. At higher storage pressures, the volume of gas is reduced, and therefore the cylindrical type of storage vessel becomes more economical, If allowance is made for the cost of compression and cooling ~the gas, some of this apparent saving is lost. When handling small masses of gas, there is an advantage in the use of cylindrical storage vessels ~ because the cost of fabrication becomes the controlling factor and small cylindrical vessels are more economical than small spherical vessels. Further economy can. sometimes~ realized by using

I:~ Where a given mass of gas is to be stored

..,~
.

·1

Methods of Fabrication

Fig. 1.4. Two multi'pheres for storage of nitrogen under 400 Ib per sq In. gage. (Courtesy of Chicago Bridge & Iron Company.)

modified spherical vessels such as the two multispheres shown in Fig. 1.4. These storage vessels were designed to handle nitrogen at 400 lb per sq in. gage working pressure. Modified spherical vessels are also used for storage of large volumes under moderate pressures. Large ellipsoidal vessels have been built to hold 55,000 bbl at a pressure of 75 lb per sq in. gage. The largest vessels for storage under pressure are the semi-ellipsoidal tanks, which have been made to hold as much as 120,000 bbl at a pressure of 2~-2lb per sq in. gage. As the capacity of an individual vessel is increased, the pressure that the vessel can safely maintain (without very heavy construction) decreases. A hemispheroid with a capacity of 20,000 bbl of natural gasoline at a working pressure of 2Y2' lb per sq in. gage is shown in Fig. 1.5. 1.2 METHODS OF FABRICATION

and greater reliability as compared with cast iron, it is more suitable for high-pressure service where metal porosity is not a problem. The vessel diameter is still limiting because of problems in casting. Alloy cast-steel vessels can be used for high-temperature and high-pressure installations. Forging is a method of shaping metal that is commonly used for certain vessel parts such as closures, flanges, and fittings. Vessels with wall thicknesses greater than 4 in. are often forged. Other special methods of shaping metal, such as pressing, spinning, and rolling of plates,are used for forming closures for vessel shells and are discussed later in the text. Sheet-metal forming is similar to pressing in that metal is shaped by means of presses and dies, but this method is limited to relatively thin stock. The process of sheet-metal forming as a method of vessel fabrication finds its greatest application in the field of nonferrous metals such as copper, Monel, and stainless steel, where cost considerations often preclude the use of heavier stock. Riveting was widely used, prior to the improvement of modern welding techniques, for many different kinds of vessels, such as storage tanks, boilers, and a variety of pressure vessels (12). It is still used for fabrication of nonferrous vessels such as copper and aluminum. However, welding techniques have become so advanced that even these materials are often welded today. Because of the trends away from riveted construction, the designs based upon riveting as a method of fabrication will not be discussed in this text. Machining is the only method other than cold forming that can be used to secure exact tolerances. Close tolerances are required for the mating parts of equipment. Flange faces, bushings, and bearing surfaces are usually machined in order to provide satisfactory alignment. Laboratory and pilot plant equipment for very-high-pressure service is sometimes machined from solid stock, pierced ingots, and forgings. Multilayer vessels for high-pressure

Process equipment is fabricated by a number of wellestablished methods such as fusion welding, casting, forging, machining, brazing and soldering, and sheet-metal forming. Each method has certain advantages for 'particular types of equipment. However, fusion welding is the most important method. The size, shape, service, and material properties of the equipment all may influence the selection of the fabrication method. Gray-iron castings have been widely used for the mass production of small pipe fittings and are used to a considerable extent for larger items such as cast-iron pipe, heatexchanger shells, and evaporator bodies because of the superior corrosion resistance of cast iron as compared with steel. Large-diameter vessels cannot be easily cast, and the strength of gray iron is not reliable for pressure-vessel service. Cast steel may be used for small-diameter thickwalled vessels. Furthermore, because of its higher strength

Fig. 1.5. diameter pressure.

by 35 ft high. Designed for 2Y2 Ib per sq in. gage (Courtesy of Chicago Bridge & Iron Company.)

A 20,000-bbl

hemispheroid

gasoline-storage

tank

64 ft in working

Factors Influencing the Design of Vessels

Fig. 1.6. ferential

Welding

external

circum-

seam of shell of large vessel

with automatic welder. (Courtesy of C. F. Braun & Company.)

services may be fabricated by machining a series of concentric shells and shrink fitting for producing desirable pre-" stress conditions. This method of vessel fabrication i~,discussed in a later section of the text. In general, machining is an expensive operation and is limited to small vessels and parts in which the cost can be justified. .' 1.2a Fuslon Welding. Fusion welding is the most widely used method of fabrication for the construction of steel vessels (12). This method of construction is virtually unlimited with regard to size and is extensively used for the fabrication and erection of large-size process equipment in the field. Often such equipment is fabricated by the method of subassembly. In this process, sections of the unit are shop welded and then assembled in the field. Equipment having a size sufficiently small to permit transportation by trucks, rail, or barge is usually completely shop welded because of the lower cost and greater control of the welding procedure in the shop. _.. There are two types of fusion welding that are extensively used for the fabrication of vessels. These are: (I) the gas welding process, in which a combustible mixture of acetylene and oxygen supply the necessary heat for fusion, and (2) the electric-arc welding process, in which the heat of fusion is supplied by an electric current (13, 14, 15, 16). Arc welding is the preferred process because of the reduction of heat in the material being welded, the reduction of oxidation, and better control of the deposited weld metal. A wide range of arc-welding equipment is available. from the small portable welding units to the large automatic welding machines. Small arc-welding machines are widely used in welding. shops that fabricate small equipment whereas the

automatic machines are better suited for the welding of heavy sections involving the deposition of a large quantity of weld metal. Figure 1.6 illustrates the use of an automatic welding machine in fabricating a large-diameter vessel. Gas welding is the preferred type of welding for light gages of metal (20 gage or less), which are difficult to weld by the arc-welding process. Gas welding equipment is extremely useful in flame cutting either in the field or in the shop! One of the most recent and successful developments in 'the field of arc welding of vessels is the submerged-arc welding process (17). This process was virtually unknown at the beginning of World War II. The necessity of expediting production of, welded equipment during the war years resulted in the realization of the advantages of this technique. The process involves submerging of the arc beneath a blanket of granulated mineral flux. The arc beneath the blanket generates heat to melt the electrode and deposits weld metal. A portion of the granulated flux melts, forming a protective layer on the weld metal, and solidifies with the weld metal. In addition to completely protecting the weld metal from the atmosphere, this process makes the weld metal virtually free of hydrogen, A13the arc is covered, there is no arc flash, and ~ a lesser quantity of smoke and obnoxious fumes is produced as compared with the' earlier welding processes. As the weld can not be observed by the operator, mechanical attachments are used to control the _/dimensions of the weld. Several inches of weld metal can be deposited in one pass, a fact which greatly decreases the welding time involved. However, the greatest advantage

Methods of Fabrication of the submerged-arc process is the elimination of the operator variable. 1.2b Welding Standards. The success of fabrication by welding is dependent upon the control of the welding variables such as experience and training of the welder, the use of proper materials, and welding procedures. An inexperienced welder or a welder using inferior materials or incorrect procedures can fabricate a vessel that has good appearancebut has unsound joints which may fail in service. Thus it is absolutely essential that the welding variables be controlled in order to produce sound joints in the equipment. A number of codes and standards have been established for this purpose. Some of these standards are: "ASME Code Welding Qualifications" (Section IX of the ASME Boiler Code) ASA Code for Pressure Piping (B 131.1, Section 6 and Appendices I and II) Standard Qualification Procedures of the American Welding Society API Standard 12 C, API Specification for Welded Oil Storage Tanks (Sections 7 and 8) The American Welding Society (AWS) established the basic standards for qualifying operators and procedures. These standards of qualification form the basis for most of the standards in the various codes. For practical purposes, therefore the rules for qualifying welders and welding procedures are essentially the same in the various codes and standards. Regardless of whether or not the welded vessel is intended to meet one of the codes or standards, it is advisable that the welding conform to one of the minimum standards. Each fabrication shop should establish welding procedures best suited to its need and its equipment. To meet the welding standards previously mentioned, it is not necessary that welding procedures be the same in all shops. But it is necessary that, regardless of the procedures used, the welded joints pass the qualification tests for welding procedures and that the welding operators be qualified in using these same procedures. To meet welding standards, welds made by the shop procedures must be tested to determine tensile strength, ductility, and soundness of the welded joints. The required tests for the welding procedures specified by API Standard 12 C involve the following:
A. For groove welds: 1. Reduced-section-tension test (for tensile strength). 2. Free-bend test (for ductility). 3. Root-bend test (for soundness). , 4. Face-bend test (for soundness). 5. Side-bend test (for soundness). B. For fillet welds: 1. Transverse-shear test (for shear strength). 2. Free-bend test (for ductility). 3. Fillet-weld-soundness test.

Double-welded butt joint (V- type groove)

Single-welded butt joint with backing strip (may be V-or U-type groove)

Single-welded butt joint without backing strip (may be V-or U-type groove)

=~7.~""'~'7~""m""'~"""~/M
Single-welded butt joint With backing strip

,,'

' ___c~ in. or less, W////ff~/hl

Single-welded butt joint without backing strip

~
Double full-fillet lap joint Single full-fillet lap joint with plug welds

-tfig. 1.7. Examples of welded joints. (Note: The two types of lap welds shown may be used only for circumferential joints and for shell plates not over ,~~ in. thick, and for attachment of nozzles and reinforcements thickness limitation.) (From the API-ASME code [1OJ.) without

shall not be less than 95 % of the minimum tensile strength of the material being welded. 2. The minimum permissible elongation in the free-bend .. test is 20 %. 3. The shearingstrength of the welds in the transverseshear test shall not be less than 87 % of the minimum tensile strength of the material being welded. 4. In the various soundness tests, the convex surface of the specimen is examined for the appearance of cracks or other defects. If any crack exceeds 7~ in. in any direction, the joint is considered to have failed. The individual welders, as well as the shop procedures, must meet certain standard qualifications. The individual welders must qualify under the established procedure according to the test previously described. This is important because a welder may qualify when using one procedure but may be unable to qualify when using another procedure. For example, an operator of an automatic welding machine may produce satisfactory welds with that machine but may not qualify when using manual equipment. 1.2c Types of Welded Joints. A variety of types of welded joints are used in the fabrication of vessels. The selection of the type of joint depends upon the service, the thickness of the metal, fabrication procedures, and code requirements. Figure 1.7 is a diagram from the API-ASME code for unfired pressure vessels which illustrates some of the types of welded joints used in the welding of steel plates for the fabrication of pressure vessels. Other types of weld joints and details for the preparation of such joints are given in Appendix B. Instead of drawing weld details to specify the type of weld desired, most engineering offices now use standard symbols for welding conventions (16). Typical welding symbols are shown in Fig. 1.8.

The minimum results required by tests such as those listed above are described in detail in t he various codes. A few representative requirements are: 1. The tensile strength in the reduced-section-tension test

Factors Influencing the Design of Vessels Type of weld

Groove Bead Fillet Square Weld all

Bevel

Plug and

Field weld

Flush

sial

around

II

VV

l)
Other (or far) side of joint

•0
Both sides of joint t'nerement lenglh ~e

--,

Locationof welds
Arrow (or near) side of joint

~~n < L
See note 5\ Siie --"'-

[Field

weld

$'ncluded 90·

angle~ Size 40· Size1

~~II
[Size

<

Slze"\

;{/

\~~

B2
~

~~
~ Off~t if staggered \_ Pitch of increments ~ Weld all around

Root opening

Flush

See note 5

1. The side of the joint to which the orrow points is the orrow side, ond the opposite side of the joint is the other side. 2. Arrow-side ond other-side welds are same size unless otherwise shown, 3. Symbols apply between abrupt changes in the direction of welding, or to the extent of hatching or dimension lines, except where the allaround symbol is used. 4. All welds are continuous and of user's standard proportions unless otherwise shown. 5. Tail of arrow used for specification process or other reference. may be omitted when reference not used. I ~ Fig. 1.8. Welding symbols recommended {Tail

6. When a bevel- or J-groove weld symbol is used, the arrow shall point with a deflnite break toward the member which is to be chamfered. (In cases where the member to be chamfered is obvious, the break in the arrow may be omitted.) 7. Dimensions of weld sizes, increment lengths, and spacing, in inches. 8. For more detailed instruction in the use of these symbols refer to Standard Welding Symbols, published by American Welding Society.

by API Standard

12 C.

(Courtesy of American Petroleum Institute.)

1.3

TYPES OF CRITERIA IN VESSEL DESIGN

The selection of the type of vessel is based primarily upon the functional service required of the vessel. The functional requirements impose certain operating conditions in respect to such things as temperature, pressure, dimensional limitations, and various loads. If the vessel is not designed properly, so as to accommodate these requirements, the vessel may fail in service. Failure may occur in one or more manners, such as by plastic deformation resulting from excessive stress, by rupture without plastic deformation, or by elastic instability. Failure may also result from corrosion, wear, or fatigue. Design of the vessel to protect against such failures involves the consideration of these factors and the physical properties of the materials. Various types of possible vessel failure and criteria in vessel design are discussed in the following chapter. . 1.4 ECONOMIC CONSIDERATIONS

vessels from more expensive metals or alloys. As the size of the tank is increased to handle larger volumes, the relative costs of using alloys and nonferrous metals increases. Prestressed or reinforced concrete may sometimes be used .to advantage for the construction of large vessels. 1.40 Steel Pricing. The bulk of chemical and petrochemical process equipment is fabricated from plain carbon steel. A know ledge of the method of pricing steel is essen.

Although the chemical-process requirements generally limit the choice of materials of fabrication, the final selection is frequently dictated by economic considerations. For purposes of comparison the relative costs of 10,000-gal tanks fabricated from various materials are tabulated in Table 1.1 (with steel as the unit reference) (IS). An examination of this table indicates that the cheapest construction materials, provided they can be used, are wood, concrete, and steel. These materials can frequently be lined with a thin protective layer; this eliminates the necessity of fabricating the

Relative Costs of Materials of Construction for Tanks Cost Relative to Steel 10...0000 gal 100,000 gal Wood, 0.4 Concrete (reinforced) 0.6 0.5 1.0 1.0 Steel 1.2 Lithcote-lined steel 1.2 2.0 Rubber-lined steel l.S 2.0 l.S Lead-lined steel 2.6 2.0 Copper 2.4 3.0 Aluminuin 2.7 3.0 Glass-lined steel 3.0· 2.7 Ni-clad steel 2.7 Stain-clad steel [ 3 >0 3.4 3.5 . Stainless steel, type 30"'3.4 Monel-clad steel 3.5 3.4 Inconel-clad steel 3.5 / Stainless steel, type 316 4.S 4.4 4.S Monel metal 4.4 12.S Silver-lined steel

Table 1.1.

Economic Considerations Table 1.2. Grade Classification by Size of Flat, Cold-rolled Carbon Steel

(Courtesy of Great Lakes St~el Corpor?tion~ D~vision of National Steel Corporation, Detroit, Michigan) Thickness, 0.2&0 or thicker Bar Strip (2) Sheet (3) Sheet Sheet inches 0.2499 to 0.0142 Strip (1) Strip (2) Sheet (3) Sheet Sheet

Width, inches Up to 12 Over 12 to 24 Over 12 to 24 Over 24 to 32 Over 32

Notes: (1) Up to 72' in. wide and less than 0.225 in. in thickness, and not to exceed 0.05 sq in. in cross section, having rolled or prepared edges is "flat-wire." (2) If special edge, finish, or definite temper, as defined by ASTM Specification A-I09. (3) If no special edge, finish, or temper is specified or required. tial in order to arrive at economical designs for equipment fabricated of steel. Steel may be purchased from two sources-a steel mill or a steel warehouse. The prices paid for the steel from the two sources are very different, the warehouse prices being appreciably higher. The reason for these price differences is found in the methods used by steel mills to obtain maximum-volume production in order to minimize unit costs. It is the present practice of the steel mills to accumulate orders until they have sufficient tonnage to permit economical rolling. Therefore, the steel mills usually serve cu:stomers who require material in reasonably large quantities and who can anticipate their requirements well in advance. It is apparent that this mode of operation is not conducive to quick delivery; three or four months, or more, depending upon the rolling schedule, may elapse before delivery. This situation makes necessary another means of furnishing steel to customers who require material quickly and in quantities too small for mill production schedules. The steel warehouse fills this distribution need, supplying steel immediately from large warehouse stocks. The steel warehouse secures steels from many rolling mills, in a full range of qualities, finishes, shapes, and sizes, and stores these steels. Thus fabricators using steel may purchase any particular product immediately from stock or combine orders for various products and buy all at one time from one convenient source. Obviously the warehouse must be paid an increase over

mill price to compensate for the handling, storage, and delivery of the steel stock. Therefore, the difference between warehouse prices and mill prices is essentially a service charge. The relative amount of "millproduction" that was shipped to warehouses for warehouse distribution for the ten-year period 1944-1954 is indicated in Fig. 1.9. This figure indicates that for the seven-year period 1945-1952 about 18 % of the total steel-mill production on the average was shipped to the warehouses. For the year 1951 the steel mills produced 78,928,950 tons of steel products and shipped 14,399,432 tons (18.50 %) to the warehouses. In the design of equipment for large process plants, it is not unusual to place vessel orders with the vessel fabricator from 6 to 12 months before the required shipping dates to enable the vessel manufacturer to order the steel plate from the mill rather than from a warehouse. MILL PRICING. In general, steel is purchased from the mill or warehouse in the "hot-rolled" or "cold-rolled" condition.· The steel is further classified as sheet, strip, plates, or bars. Alloy steels and structural steels are classified separately. "Hot-rolled, plate steel," or "cold-rolled strip steel," or "alloy steel bars," and so on are combined classifications of types of available steel. Table 1.2 shows the grade classification, by size, of flat, cold-rolled carbon steel by a typical steel mill. Table 1.3 shows the corresponding grade classification by size of flat, hot-rolled carbon steel. The steel mills and the warehouses quote "base prices" for each class of steel product. Table 1.4 shows a section of a typical mill base-price list as of January, 1956. The prices are all F.O.B. cars or trucks at the mill works (Indiana Harbor, Indiana.) The prices quoted in Table 1.4 apply to an order of 10,000 Ib or more of the size ordered at one time (one thickness and one width is considered one size), of one grade or analysis, released for shipment to one destination at one time. For weights of less than 10,000 lb, "itemquantity extras" apply. Item la, of Appendix C lists the quantity extras charged by a typical steel mill (Inland Steel Company, as of May 13, 1953) for carbon-steel plates.

21 20

c:
Q)

19 18

t
Q.

/\

11\

Table 1.3.

Grade

(Courtesy

Classification of Flat, Hot-rolled Carbon Steel of Great Lakes Steel Corporation) Width, inches Over Over Over 3}2 to 6 6 to 12 12 to 48 Bar Plate Plate Bar Strip Sheet Strip Strip Sheet Strip Strip Sheet Over 48 Plate Plate Plate Sheet

17 16

Thickness, T03H inches incI. 0.2300 and thicker Bar 0.2299 to 0.2031 Bar 0.2030 to 0.1800 Strip 0.1799 to 0.0568 Strip

./"t:::::::--

~I

/ \V
.... .... '"
Ln

/f\

_.
Year

Fig. 1.9.

Percentcge

of totcl millproduction

of steel products

shipped

to

wc rehouses.

10

Factors' Influencing the Design of Vessels Table 1.4. Mill Price List (Courtesy of Inland Steel Company, Chicago, Illinois, January, 1956) Base Price per 100 Ib ness, and width are calculated on a dollar-per-l00-lb basis, whereas the extras for circular and sketch plates" are calculated on a percentage basis, as listed in item 3 of Appendix C. The percentage is calculated on the net-per-l.Gu-lb price of the smallest rectangular plates from which each circular or sketch plate is obtained exclusive of freight and extras for gas cutting a quantity . The outside dimension of each circular or sketch plate determines the size of the smallest. rectangular plate from which the circular or sketch plate is obtained. A wide variety of other "mill extras" are quoted by the various steel mills. The reader is referred to company price lists for complete quot.ations on these other extras, among which are: 1. Heat-treatment extras 2. Surface-finish extras 3. Testing extras . 4. Chemical-requirements extras 5. Specification extras 6. "Special-requirements extras 7. Dimensional and workmanship extras 8. Extras for special-shipment requirements 9. Special-marking-of-plates extras "10. Loading extras 11. Bundling-of-plates extras. Each extra is usually separate and distinct. The individual items are combined to form a "full extra" applicable to the order. The steel-mill base prices given in Table 1.4 and the steel-mill extras given in Appendix C are quoted as of January 31, 1956. It must be emphasized that these prices are representative of the prices quoted by steel mills at that time. As economic conditions vary, prices charged for manufactured products fluctuate, and the base and extra prices are subject to change. Figure 1.10 illustrates. the changes in the base price of steel plate in Pittsburgh from"j July, 1938 to January, 1956 (19). The horizontal line to April, 1945 is for the period during which government controls were maintained on steel prices because of the national emergency of World War II. The curve indicates that the price of steel plate at the mill doubled between 1945 and 1956. Reference should always be made to the most-recent available price lists for estimation purposes. WAREHOUSE PRICING. Steel warehouses are strategically" located' throughout the country to provide a convenient source of supply for steel products. Whereas the steel mills produce steel products of standard' length and width, the warehouse will supply steel cut to the customer's requirements. Typical operations in the warehouse include shearing, sawing, slitting, and flame cutting. Some warehouses will supply steel plates rolled to cylindrical shapes and bar shapes, bar stock rolled to lings or bent to other shapes, ami plates with drilled or punched holes. Figure 1.11 shows typical stocks of steel illiit warehouse. Prices vary somewhat from warehouse to warehouse, depending upon the location of the warehouse, the distance from the mill, and the service performed. Item 4 of Appendix .C gives typical warehouse prices from one warehouse (20).

Hot-rolled sheets (IS-gage and heavier) Hot-rolled strip Cold-rolled sheets Hot-rolled carbon-steel bars (merchant quality) Hot-rolled alloy-steel bars Reinforcing bars Carbon-steel plates Carbon-steel structural shapes

$4.325 4.325 5.325 4.65 5.575 4.65 . 4.50 4.60

Carbon-steel plates fall into three classifications: (1) those furnished to chemical requirements, (2) those furnished to physical requirements, and (3) those furnished to both chemical and physical requirements. Item Ib of Appendix C lists the "classification extras." Other mill-price extras of primary interest are given in item I, Appendix C and are classified as: quality extras, length extras, width-and-thickness extras and killed-steel extras. Circular- and sketch-plate extras are involved when items such as blanks for formed heads are purchased. As these plates are usually flame cut, gas-cutting extras also apply. Gas-cutting extras are also charged for rectangular plates when the thickness limits for shearing are exceeded. Item 2 of Appendix C lists the gas-cutting extras per linear foot of cutting. The previously mentioned extras such as quality, thick-

00
C"t)

0'1

In \0 ~ '<T '<T '<T ~ ..... C7\ 0'1 0'1 ~ ~ ..... ..... .....

,.._

s. .....
C7\

..... N 0 In In In LO 0'1 ..... ..... ..... ..... C7\ 0'1 0'1

C"t)

In ~ In ~ 0'1 .....

Year '.
Fig. 1.10. Base price of steel plates in Pittsburgh.

Economic Considerations

11

Fig. 1.11.

Interior view of warehouse showing typical stocks of steel.

I

ICourtesy of Joseph T. Ryenon & Son. Inc.}

'I\}

\J
Fig. 1.14a gives the cutting time in hours per linear foot for flame cutting the shell plate as a function of plate thickness. This curve may be used when the shell is cut from standard plate kept in stock, such as mill plate. If the plate is purchased from a warehouse, it may be obtained, cut to size, and the cutting cost included in the purchase price. In addition to the man-hours involved in flame cutting, a machine rate burden which includes the cost of machine time and gas consumed is also involved in flame cutting. The curve for cutting-machine rate burden is shown in the lower part of Fig. 1. 14a. The number of man-hours involved in edge preparation prior to crimping and rolling are given in Fig. 1.14d. The combined number of man-hours involved in crimping the longitudinal seam ends and rolling the plate into a cylindrical form are given as parameters in Fig. 1.14b; the man-hours are treated as a function of plate lengths and thicknesses. • The parameters given in Fig. 1.14b are based upon the rolling and crimping of a few plates; the figure therefore gives a liberal allowance for these operations when more than a few plates are rolled and crimped at one time. The man-hours required for the fitting and assembling of the shell and closures for the vessel are given by the solid lines in Fig. 1.14c as a function of the plate thickness and with three parameters for different degrees of complexity of the vessel. Also included in this figure are three curves having nearly the same shape as the parameters and indicated by the dotted lines that may be used as a rough check on the total man-hours involved in fabrication. These latter three curves are intended to be used only as a check to disclose any gross errors in the total estimation. In acldi-

lAb Fabrication Costs. The direct costs of producing a piece of process equipment include the cost of materials and the cost of labor. Material costs consist of the shop material used in the fabrication plus the parts purchased from an outside source. The cost of steel plate, which has been discussed in the previous section, usually comprises a major portion of the material costs for vessels. The labor costs involved in the actual fabrication of the equipment are often difficult to estimate accurately in advance. How has reported methods of short-cut estimations of welded process vessels (21). FABRICATION PROCEDURE. One of the first steps in the fabrication of the vessel is usually the preparation of the shell for rolling. The edges of the individual plates for the shell require machining to true the edges and, in the case of code welding, to prepare the edge for welding. Figure 1.12 shows a 40-ft planer machining a double "U" edge on a 1%-in. plate 29 ft long for a vessel shell.· The next step is usually crimping the edges of the plate which will be joined by a longitudinal weld. The crimping step is required because the rolls cannot be used to form the two ends to the desired curvature. Figure 1.13 shows a 350-ton hydraulic press in the foreground, crimping the edge of a plate before rolling. In the background plates are shown being rolled into a cylindrical shape on pyramid rolls. MAN-HOURSANDl\IATERIALS. After the shell has been given edge preparation and rolled into shape, the vessel components must be fitted and assembled by welding. Figure 1.14 gives curves according to How (21) for estimating the man-hours involved in the various stages through assembly of the shell and closures. The upper curve of

t:

12

Factors Influencing the Design of Vessels

Fig. 1.12. Company.)

Machining a double U edge on a plate 1% in. thick and 29 ft long for a vessel shell by means of a 40-ft planer.

(Courtesy of

C. F. Braun

&

<:>
Fig. 1.13. A 350·ton hydrc;lulic press crimping the end of a plate before rolling.

(Courtesy of C. F. Braun & Company.)

Economic Considerations 30 0.15 0.10 ::: 0.08

/

13

V
/'

/
25

i 0.06
.!:

s:

..
oj

_,.._,..

./'

0.04

;, 0.03 c: B :> (.) 0.02

---3.5

CI)I.\\".!.

t\(I\e

:::;..--

b--'"

~
_,..

..... ;..-

l/Iac

";'?

"oI)IOe"_ . e late~

l- :.---

f--

1\\ ,"\ -, 1,\'\ ,,\ ,,\ "\ ,§ 20 1'\'\1\'\ S .'\

I'\.\ "\'\. I'\. -, '\.\ ,'\. \"\

-, -, -, r-, r-, -, .'\. .'\. "\'\. 1-, -: -, '\. -, -, '\. .'\. '\.\ -, .'\. .'\.1\. _"'\.\ "\\ -, -, "\"\

-. -. ,'\. -: .'\. ."'\.

.'\.
,""\_

~ a.
'0
s:

\'\ _"-, 1'\'\ S .'\ ,'\. '\.I'\. 1\'\ .'\ ,,\ ,,\ .'\ \1"'."\ ."\ \ ;"\

-, \ "\ ."\ -, "\'\. -, -, -, -, -, -, -, "\

-,

.'\.

-, "\"\ -,
.

.'"'- ",-",- ~~ .: "'-"- .'"-, r-, '\'\.

<,

0.01

V
~

.s

I~I
~
(a)

1~

1~

I2 I 3 1% 2~

Plate thickness, in.

-, -, '\ -, -, -, -, \.'\ ,\. .'\ "\\ 1\\ ."\ "\ .'\ ."'-, -, -, I\, ,,\'\ ,,\ ,,\ 1,\\ 1'\'\ "\'\ -, 1,\ -, \ ,,\\ ,,\ l'\\ 1,\ -, 1'\'\ .'\ .'\. \'\ -, 1"\ "\ ,\\ .\ 1\","\ \ "\ .\ "\ \. 1\ ,,\ ,,\ .'\ .'\ '\ -, S .".. 10 \\ \ \\. l'\~".'\ -, -, ,,\ ,,\'\ ,,\ ,,\ -, ,,\ -, -, ~\ .\\ .\\ ,,\ -, -, .'\ "\ "\ "\ -, "\ ,\\ ,\ \ \ ."\ 1,\ ,\\ ". \ -, ~\\ \\' ,\\ 1\\ ,,\ :'\ 1'\\ 1,\ \ 1,\ \ ,'\. -, I'\. \1\\ ,\ \ 1\'\ ,,\\ ,\\ 1,\\ 1,\\ 1'\'\ I'\.\ '\. \ 5 ~ % ~
% % ~
Plate thickness. in. (b)

~ ~~ .'\ ,,\ \ I'\.\u' S .'\ ,,\ .'\ \ 1,\\ ~ 15

"\ 1\ \ "\, "\ -"'.'\. \. I

-, I'\. -,

-, "-, '\. '\. '\. -, "\I'\. -, -, " -, -, -, -, ~'\. '\ I'\. '\ 1\.\ I'\. .'\. -,

-,I'\. .'\. -, "\\ -,

-, -,
.". \.

'"

14

-:;;~
u
:>

!\ 3.0

......
12 ~
:> u
'C

~ 0.06
0
Q)

:!2

'C

tID
Q) Q)

I Y,..'---I .....,:~·I r-%" 1.6

I-~~~+-~~'~

%·.,I.~·R·.
~2--11<-

t::v,.' /

! 2.5 :e
8 ......
~ 2.0

"-

L II
/

__ Complicated and small

1"<-

10 'C '5 8

Q)

\,

:e ~
a.

1:
.!l!

1.5

-,

<, ".,

8' ~ ...... ~ Q; ., a.

--

c c

~ 0.05
Q;
a.

:c u
'C

a. c:

1.4 1.2

:'!I..

Welding rod

..c

.a

1:
0 ~ e

'" u 1.0
Q)

'" c

'"
/

-, /
/1\.

~o:/

,,,,Q"''/ Q

,,,,~o~ /

~2

c:-

:0 E Q)
'C C

>.

"

<, /Medium

~ 1.0 ~ 0.5

<~
......,.
-0:::

'" tID

-==== :::-:::-_
%
(e)

--

---.....
~

""

--;
~ ~

~ .a
.!l! c
0

0.04

c: 'i5

tID

1~

~ s: u
..'l!

a;

'" e

<t:! 0.8 :§

-r:i a; ;:

a. a.

=
o

~ 0.03

il'o

~ .,
a.

0.6 0.4 0.2

1:

,<'","0 ~ ~ ",,"0 / ~~'

/
// c,.s

/ V/ ~ /_V ~'V / ~ '\

'\

/ /V
/

V,

i.::

'0 2 f-

(ij

/ ~ ~

-, I'"
'\

1~

Ya

1 1Ja 1"> 1% 1~

Average steel thickness, in.

Plate thickness, in.

(d)
(a) Cutting time and machine rate burden for flame cutting with automatic time (for (Courtime and welding-rod weight for 'code butt welds in carbon steel.

Fig. 1.14. machines.

Curves of How (21) for estimating shop time for vessel fabrication. (d) Welding and edge-preparation

(b) Time for rolling plates 60 to 72 in. in width and of various lengths and thicknesses •. (c) Fitting and assembly, and total fabrication

rough check), for steel tanks and weldmenls. tesy of McGraw-Hili Publishing Co.)

tion to giving the number of man-hours for edge preparation, Fig. l.14d also gives the welding time and quantity of welding rod per linear foot involved in assembly of the vessel ends and shell. Most vessels contain two or more nozzles for charging and discharging operations. The man-hours and weldingrod requirements for attaching different types of nozzles are given in Fig. l.15.

Formed closures such as dished heads can be purchased from fabricators with the edges beveled for welding. Costs for this preparation are given in a later section describing formed heads. However, if it is practical to machine the heads in the shop that fabricates the vessel, the man-hours required for this operation may be estimated from Fig. l.16a and b. Bolting flanges for nozzles may be shop fabricated from flat plates. The machining time for this opera-

14

Factors Influencing the Design of Vessels

1.6 1.4 1.2

N
0 C1>

1.6
-

1.4 ft-:-

~
V

1.2 ~ 1.0
"i5 t;

//
/

LL
/

5.0 4,5 4.0

//

c:

.S:

~0.8
-.:i

1/ /
Labor, hy

13 0
e

Q)

"t:J

8. 2.0

.S: 3.0
N
N

.!! "i5 t;
.j

3.5

" kK
~ <,
Labor, IS0-lb

~Weld

rod, IS0-lb

nozzle

./ Weld rod, 300 -Ib nozzle /

V
V

v
V

r-,
'\

<,

c:

/
~'d rod,lb

:e
-.:i

Labor, 3OO-lb nozzle / nozzle

-, / t>< -, /

/
/
\.

.;.,

8.0.6 0.4 0.2

./

;/
,./
./
".,

g 2.5
8.

/
/

z 2.0
1.5 1.0 0.5

__._
%

/ VV
/'

1/

)V

1\V
\

<,

Kr-..

3 1~ 2 2~ 1~ Nominal nozzle size, in.

(a)

22~33~4

6 Nominal nozzle size, in.

(b)

1Jl
5
<»
Weld rod Ib

i\

1\

r\

20 15

'{f

0 30

.!! c ;I-

Series-IS forged steel

~'~'.~

15 '--"t:J

I
.

_/_;
Schedule-40 steel pipe /

!J 'I

-r-..I'-...

.......

I
I

~ 10
c:

lI-

.!!.

r--

.S:

V 10
Weld rod, Ib_

II

'/

25
C1>

~ c:
GI

...

13 o
c:

ai

8.

/"

r-.... V
...... V

//

/ II

:e
~

c..

r---r---~ 20 r---r---~ r-r---.S: r---.!l 15 r--

/-eS-30 forged steel slip-on flange "'"""'i

"'"' "

v
'\

I
I

II
\

Schedule-SO I'*"'" steel pipe

,.
\

~ c:
lii

~~bzzzz...
/

rV
,

./

c..

Z 10
./ ./

L
hr

./ ..__

Labor, hr

-Labor,

_.... """
2 2~33~4

./

2~33~4

8 10 12 14 16 18 56 Nominal nozzle size, in.

(c)

8 10 12 14 16 18 56 Nominal nozzle size, in.

(d)

Fig. 1.15. Curves of How (21) for estimating welding-tiine in hours and welding-rod requirements for nozzle attachments to vessels. (a) Welding time and welding rod for installing XH steel couplings In unflred p~ressure vessels. (bl Welding time and welding rod for installing long-welding-neck forged-steel nozzles. (e) Welding time and welding rod for fabricating lSD-psi nozzles in unflred pressure vessels. (d) Welding time and welding rod for fabricating 300-psi nozzles In unflred pressure vessels. (Courtesy of McGraw-Hili Publishing Co.l

EconomicConsiderations

15

8 7 6
CD

4 1

5 4

c 1:

T-

~ CI.d\
J !f."
~"R ~

1/ 1/

3 I---

I--I---

_:f 'so" ~"R

/

'"
E
C

l(," ,

l!,"--l~

1.11/ / 1/ I/ 1.11/
/

./

1/

vi/

1/ /

"0

'"
:J I

/1/

Q.

3
~ /

1/

~ G> C ~ :2 u ~ '" E 2~ "0 ~ 1.8 ~

T~.

s
Co

.l~ 36"land')

t CI'~
on steel

1/
/ 1.1 '1

V 1/

;f:'V li," land

1// /

~.__L

1/

II

1/

1/

'" -0
.<::

Gi
CD

rt!.Lv.
1/

./ ./

2 ~ 1.8 l: 1.6
iii

'"

VV
/ ./ ./ /

~~ ~ V ,."Ii- ~""'./ V .... -l

/
/

=:
1/

1.6 1.4

"/i""

.,;,/

.! 1.2
ij
0.7 0.6

-0

5i

I V .1717 ~4'""~~ ..;.

1.4

./ ./ /"

/" /"

./ ./

./

./

./ V 17 ~ V /~ /

[7

VI. /

--; 17

1.2 V 1.0 0.9 0.8 15

/" V

/'

/
.....

./

l7

..... V

./

V
20 25 30 35 40 50 60 70 80 Outside diameter of head, in.
(a)

/'

100 120

0.5 15

20

25 30 35 40 50 60 70 90 Outside diameter of head, in.

(6)

110 130

6.5 6.0 5.0

I-I--

111

:c "0
"0
C

T raised fa:J £Tflatface

4.0 rI--

T"'FI'I-r
ASAstandard drilling

I 1"''2.''''.-.:,...-..'2.~ ..:;... ~
1-

.-~
~

1.4

.......,...---

'"

.l2

G> U

r3.0 2.5 2.0 1.8 1.4

<=,

"t

._,

.....: F-".: ... ....... V

:\. ........

.s::..

~ 0

:\.'R."~\."
'R."

E
c
;:

E 8b

~.£":".:

-l,....-

1.2
c

This chart can also be used to F- . estimate fillet welds on other structures such as bases and supports

:E
~3
:9

~ 1.0 u
"0
C

-0

:2 ~

1>02 c

~0.8

'"

=
c
.<::

~
G> Cl.

.:f.'"

~ ~6'~4.

'"

~0.6
/ / / '/1/

~ 1.6
l:
I-

1.2 1/ [;11/ 1.1 VV, V 1.0 0.9 [.I 0.8 V 12 34 5 5

r;../

/. / V

./

./ '/

./

...

_/

/'

V ,\~

;V

'/ V

V/ -:V
/

./

/"

c;;
I-

00.4

Add 0.1 hr for raised-face _ flanges up to 6 ips inclusive

0.2

10 12 14 16 18 20 22 24 Nominal flange size, in.

26

7 ~
J.(

1/ V~

V~

i(I\.

-: / /y

-:
f
1

-:
_
1~

...............1""

~Si!e

of weld -

Size of weld, in.

Cc)
Fig. 1.16. beveled Curves of How (21) for estimating time, flanged face, carbon-and-nickelman-hours of machining time and welding-rod grooved steel. face, carbon-and-nickelrequirements steel.

Cd)

for miscellaneous 1 to 26 ips,

operations

in vessel fabrication, and dished heads. (d) Welding

(a) Machining

and dished heads, or stainless-clad

or stainless-clad

(b) Machining time, flanged

(el Machining time for carbon-steel

(Courtesy of McGraw-Hili

plate flanges,

~ito

2H in. thick.

time and weight of welding rod for fillet welds in carbon steel.

Publishing Co.)

16

Factors Influencing the Design of Vessels Table 1.5. Engineering News-Record Cost Index Index Year Construction Co.) Year Table 1.7. Twenty-City Average of Hourly Rctes for Skilled Labor .

(Courtesy of McGraw-Hill Year

Publishing

Index

(From Engineering News-Record) . (Courtesy of McGraw-Hill Publishing Co.) Rate, dollars /hr Year Rate, dollars /h-

1913 1915 1920 1925 1926 1930 1932 1935 1940 1945

100 94 235 206 208 202 157 195 242 308

1946 1947 1948 1949 1950 1951 1952 1953 1954 1955

346 413 461 477 510 543 569 600 628 660 (July)

1926 1932 1939 1945 1946 1949

1.27 1.03 1.44 1.66 1.80 2.41

1950 1952 1953 1954 1955 (July)

--:'l

2.52. 2.84 3.01 3.14 3.25

tion is given in Fig. l.16c. Various attachments such as skirts, saddles, and lugs may be added to the vessel, usually by fillet welding. The man-hours per foot and the welding required for fillet welding is given in Fig. l.16d. Additional curves for some alloy and nonferrous metals are given by How (21). 1.5 ESTIMATING CURRENT COSTS

COST INDICES. Because of the constant change of costs for material, labor, taxes, and plant overhead, available cost data rapidly become obsolete. Thus some method of bringing cost data up to date is required. The procedure normally followed is the application of available "cost indices." The cost indices are relative numbers giving the variation in a group of costs with reference to a base year. To use a cost index the estimator simply multiplies the known cost at a given date by the ratio of the current index value to the index applicable at the date of the known cost. CostA=CostB Index A ··d B In ex

(l.1)

A number of indices are in wide use; they differ somewhat because of the basis used in their preparation and the reference year. Three widely used indices are the Engineering News-Record (ENR) construction-cost index (22), the

Marshall and Stevens equipment-cost index' (23),and the Nelson refinery index (24). The ENR construction-cost index (22) reflects labor-wagerate and material-price trends. The index -consists of the cost of a hypothetical block of construction requiring 6 bbl of cement, l.088 M fbm of lumber, 2500 Ib of steel, and 200 hours of common labor. This cost was $100 in the year 1913, which is taken as the reference year. Although this index is intended to reflect average construction costs and has no particular relation to the cost of equipment, it has proved extremely useful in estimating changes in costs for complete plants. Because of its wide use it has often been the basis of estimating changes in equipment costs. Table 1.5 lists some values of this index as a function of time. The Marshall and Stevens equipment-cost index reflects the comparative costs of equipment (23). It is based upon the costs of machinery and major equipment, installation labor, plant furniture and fixtures, tools and minor equipment, and office furniture. These costs are estimated quarterly for 47 different industries, with a separate formula for each industry and with the year 1926 as a reference of 100. The petroleum-industry index contains the following component percentages: process machinery, 25; installation labor, 19; power, 12;maintenance equipment, 2; and administration, 6. Other process industries for which indices are prepared are: the cement industry, the-chemical industry, the clay-products industry, the glass industry, the paint

Table

1.8.

Average Boilermaker Wages 1954, as a Function of Locale

in July,

Table 1.6.

Marshall and Stevens Equipment-Cost Index (Average for All Industries) (Courtesy of McGraw-Hill Publishing Co. [235]) Year Index Year Index

(U. S. Bureau of Labor Statistics) dollars/hr U. S. Average 3.n New England (Me., Vt., Mass., Conn., R. I., N. H.) 3.00 Mid-Atlantic (N. Y., Pa., N. J.) 3.44 Border States (Del., Md., Ky., W. Va., Va.) 3.01 Southeast (Tenn., S. C., N. C., Ala., Ga., Miss., Fla.) 2.90 Great Lakes (Minn .. Wis., Mich., III., Ind., Ohio) 3.13 Midwest (N. Dak., S. Dak., Kans., Nebr., Mo., Iowa) 2.96 Southwest (Tex., Okla., La.) 2.90 Mountain (Mont., Idaho, Wyo., Utah, Ariz., N. Mex., Colo.) ( 3.01 Pacific (Wash., Oreg., Calif., Nev.) \'._ 3.05

1913 1915 1920 1925 1926 1930 1932 1935 1940 1945 1946

57.9 55.9 153.3 105.3 100.0 87.0 66.1 78.0 86.1 103.4 123.2

1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 (June)

150.6 162.8 161.2 167.9 180.3 180.5 182.5 184.6 190.6 208.8 224.1

Typical Procurement

Procedure

for Vessels

17

industry, the paper industry, and the rubber industry .. A weighted-average for the process industries is also reported, which contains, in percentages: cement, 2; chemicals, 48; clay products, 2; glass, 3; paint, 5; paper, 10; petroleum, 22; and rubber, 8. Also, an average for all 47 industries is published, which has differed only by about 1% to 2% from the average for the process industries alone. As this index is based primarily upon industrial-equipment costs, it is considered more reliable for estimating changesin equipment costs than the ENR index. Values of the Marshall and Stevens index for the average of all industries are given in Table 1.6. The Nelson refinery index (24) is a construction-cost index somewhat similar to the ENR construction-cost index but based upon the cost of materials and labor for the construction of petroleum refineries. Although price indices are extremely valuable in estimating costs, it should be mentioned that they are based on national averages and may be inconsistent with price changes for a particular locale. Also, price indices are based on wage rates and material costs but make no allowance for such factors as: availability of materials, productivity of labor, competitive conditions, influence of new techniques, business optimism, relation of demand to production capacity, and other intangibles. LABOR-COST VARIATIONS. he cost of labor varies from T year to year and from area to area. These costs have risen rapidly since World War II, and at the end of 1955 the hourly rates in dollars were at an all-time high. Table 1.7 gives the average hourly rate for skilled labor for 20 cities since 1926 as reported by the Engineering News-Record (22). Skilled laborers such as machinists, welders, and boilermakers are required in the fabrication of vessels. The hourly rates, as a function of locale, that were paid boilermakers in July, 1953, as reported by the Bureau of Labor Statistics (22), are given in Table 1.8. SHOP OVERHEAD. n addition to the direct costs involved I for materials and labor, all fabricators must add an indirect cost often termed the "shop overhead" or "burden." This overhead includes a variety of items, such as the cost of supervision, administration, engineering, sales, utilities, maintenance, depreciation, taxes, and other fixed and indirect costs. These costs vary from shop to shop, area to area, and year to year, and are established by the conditions for a particular shop and by the accounting practice followed. This overhead usually ranges from 100% to 200% of the total cost for labor and materials: PROFIT. The profit of the fabricator is estimated on the total cost to the fabricator, including materials, labor, and overhead. The profit usually ranges from 5 % to 20 % of this total cost but may be higher if the state of competition permits. SOURCES OFPRICEINFORMATION. recent years a conIn siderable attempt has been made to collect, group, and correlate price information. A series of post World War II articles was published in Chemical and Metallurgical Engineering in 1946,and additions to them were made in Chemical Engineering from 1947 through 1955. These articles have been collected and reprinted in three booklets (25). Furthermore, rather recently two texts on the subject of

chemical-engineering costs have been published (26, 27). In October, 1954, Weaver listed a bibliography of 351 articles dealing with equipment costs, operating costs, and estimating methods (28). SCALING EQUIPMENT COST WITHSIZE. Frequently Ji/ piece of process equipment having a size different from tli"at for which the cost is known is desired. A comprehensive study of the cost of a variety of processequipment as a function of size and capacity was made by Chilton (UI). In a subsequent article (229) Chilton analyzed these data and similar data by Williams (230) and concluded that the "six-tenths factor" rule is useful as a short-cut method for approximating the cost of a similar piece of equipment of a different size. This rule states that the cost of a second size is equal to the cost of the first size times the ratio of the sizes (or capacities) raised to the six-tenth power, or Cost A
~'~'.. =

Size Cost B ( -Size B

A)O.6

(1.2)

The general validity of this rule has been well established, but some discretion should be exercised by limiting it to less than a tenfold range unless cost data are available for two or more units over a range of sizes. '
1.6 TYPICAL PROCUREMENT PROCEDURE FOR VESSELS

The typical procedure followed in the procurement of vessels for a process application will be discussed briefly to give a perspective of the sequence of steps involved. Normally a process-design group develops flow sheets for the process involved. The flow sheets include information relative to the operating temperature and pressure, capacities, heat duties, and any particular information concerning corrosion. The equipment-design group prepares detailed sketches of the various items of equipment, specifying the materials of construction, shell and closure thickness, type of closure, and code stamping. The nozzle and manhole types and their ratings; corrosion allowances; stress relieving, radiographing, and hydrostatic or air-testing requirements are also specified by the design group. In addition, shipping limitations; the weight of the vessel empty, with internal attachments, and filled with water; and the operating weight are usually estimated by this group. From the above information specification sheets for each item of equipment are prepared. The procurement group sends copies of the detailed sketches and the specification sheets to various vessel fabricators for quotations on prices and delivery dates. In the meantime the plant-design group are also sent copies of the sketches and specifications. This group prepares a layout of the plant design. This includes specifications for roads, utilities, sewers, fire protection, structural foundations, pumps, piping, and detail design of the various components involved. On the basis of the original estimates received from the fabricators, one or more fabricators are selected and final drawings are furnished for a rigid price quotation on the various vessels and other items of equipment to_ye p,urchased. The fabricator is then granted perrnissionj to purchase material and prepare shop cira,'vipgs. Tilese shop drawings are usually submitted l6c We puichasj for
" -, "..:. '..-)

.,:;;.'i'
-~~~'";-:::;::
.....

or-

----------

18

Factors Influencing the Design of Vessels

approval. On approval of these drawings the fabricator proceeds with construction of the vessels and other items of equipment being supplied. Where major items of equipment are involved, it is customary for the purchaser to set up an inspecling-and-expediling group at the fabricator's plant. The inspectors for the purchaser normally follow

every step in fabrication, from initial inspection of-plate and heads through testing and shipping. -This text covers the design problems of the engineering offices of both the purchaser and the fabricator. No attempt has been made to separate these problems, which often overlap. -

PROBLEMS

1. A vertical vessel designed as shown in Fig. 1.1 and 32 ft from head junction to head junction is in use. If an identicalvessel were to be fabricated of SA-285, Grade C steel today, what would be the estimated cost of: a. The two circular blanks for the two elliptical dished heads at the warehouse (no edge preparation) ~ b. Four shell plates cut to length only at the warehouse P (Note: see Chapter 5, section 5, Common Types of Formed Heads and Their Selection, for diameters of head blanks.) 2. Estimate the cost of labor today for the fabrication of the vessel shown in Fig. 1.1. much weld metal will be required? How

CHAPTER

CRITERIA IN VESSEL DESIGN

For axial compression: unit of process equipment may fail in service for a variety of reasons. Consideration of the types of failV which may occur is one of the criteria which should be used in equipment design. Failure may result from excessive elastic or plastic deformation or from creep. As a result such deformation, the equipment may fail to perform ts specified function without rupture or may fail catastrophically with rupture. Failure can usually be classified in one of the following catagories: excessive elastic deformation, elastic instability, plastic instability, brittle rupture, creep, or corrosion. 2.1 EXCESSIVE ELASTIC DEFORMATION

=P
j
=-

(2.2)

a when j = induced axial stress, pounds per square inch P = load, pounds a = cross-sectional area, square inches Stresses resulting from bending and torsion are more complex, and a large number of texts have been written on the subject of the evaluation of such stresses (29, 30,31,32). Induced stresses result in corresponding induced elastic deformations. The deformations may interfere with the functional operation of the equipment. A common example of this is found in the use of excessively thin flanges for a bolted closure with a gasket at the interface of the flanges. Tightening of the flange bolts in an attempt to seat the gasket in such a way that it will contain the internal pressure may result in excessive elastic bending of the flange between the bolts without transfer of the bolt load to the gasket. Another example is the excessive deflection of a tray in a distillation column under the tray load, a condition which produces a nonuniform liquid seal on the bubble caps and possible instability in tray operation. 2.1 b Modulus of Elasticity. In order to avoid such situations as described in the previous examples, sufficient rigidity must be incorporated into the design of the part to restrict the amount of deformation to a permissible value. The deformation which can be tolerated is determined by the function of the part. Parts in simple tension or compression, such as exist in axial loading, deform in the elastic region in direct proportion to the induced stress and in indirect proportion to the modulus of elasticity of the material of construction. Thus the proportionality con19

2.1 a Induced Stresses. Elastic deformation is induced by a load such that when the load is removed, the part resumes its original shape. A typical example is the steel spring in a watch. Under service conditions the various parts of the equipment will be subjected to a variety of induced stresses. A stress is defined as the force per unit area in the member under consideration. Various types of stresses are induced, depending upon the loading condition, and are classified as: tensile, compressive, shear, bending, and torsion. These stresses may be the result simply of the weight of the-material of construction or may be caused by loads resulting from fluid pressure, forces, wind moments, and so on. Parts under axial-compressive or tensile forces have induced stresses which may be computed by the simple relations: For axial tension:

p j=a

(2.1)

..

-.

_.--_".20 Criteria in Vessel Design loaded beam having uniform cross section and freely supported at the ends as indicated in Fig. 2.2. Consider an element, ds, of a beam having a uniform cross section, supporting a distributed load of w pounds per inch of length of the beam as indicated in Fig. 2.2. The total load acting on the element is w (dx). If w is considered positive when the load acts downward and if dx is positive, the differential shear force, dV, must be negative. By summation of vertical forces: V2 - VI w (dx) = 0, or

70,000---r--H---.---.-----,;------,----. r

50,000

'iii
Co

40,000

ill 30,000

~ e

and dV or dV
20,000

= -w (dx)
-w
(2.4)

dx

Taking a summation of bending moments about point A gives:

50
Percentage of elongation Fig. 2.1: Typical stress-strain curves for various metals.

· . --IgI Smce W (dx)2 IS neg I' ihle, 2

dM or (2.5)

M2 - MI ~

VI dx

stant between stress and strain (under axial loads) is the modulus of elasticity. Typical stress-strain curves for a few selected materials are shown in Fig. 2.1. (Note that two scales are used on the abscissa in order to enlarge the elastic region of the curves.) The elastic portion of the total strain is represented by the straight-line segments of the curves. The slope of these straight-line segments, when the strain is expressed in inches per inch, is the modulus of elasticity of the material, E, or: . (2.3) where J

Any beam under a load deflects. A particular radius of curvature exists for the portion of the beam under consideration. Thus the loaded beam has a radius of curvature r at a distance x from the perpendicular to the neutral axis. The bending 'of the beam will result in a deformation ·.6x of the fiber at any distance y from the neutral axis, as indicated in Fig. 2.3. The corresponding strain or unit deforination Ez is equal to .6x/x, and by similar triangles
E.,

.6x =-

=-

(2.6)

= axial stress, pounds per square inch .

As given by Eq. 2.3, the ratio of stress to strain for elastic deformation is equal to the modulus of elasticity, E, or

E = unit strain, inches per inch E = modulus of elasticity, pounds per square inch

J., = EE.,

2.1 c Elastic Bending. The deflection of a part subjected to forces which produce bending is a more complex phenomenon. In such cases the amount of deflection is inversely proportional to the modulus of elasticity and the moment of inertia of the member. The use of relationships developed in the field of theoretical mechanics· are required for evaluation of the deflections. General procedures for such calculations are presented in a number of texts on the subject of strength of materials (29, 30, 31, 32), and Ion the subject of the theory of elasticity (33, 34, 35, 36, 37). Selected procedures for particular calculations involved in vessel design are presented in later chapters of this text. The basic relationships for such calculations may be devel- . oped by considering the shear and bending in a uniformly

(a)

(b) beam. (01 Uni-

2.2. Forces on an element of a uniformly loaded formly loaded beam. (bl Detail of element dx,

Fig.

/'

Excessive

Elastic Deformation

21

Section -A-A

Stress diagram

Ax

Fig. 2.3.

Stress and strain in an elemental

strip of a curved

plate

or beam.

By substitution

of Eq. 2.6 fx

By definition the radius of curvature r is defined (38) as:

= E-

(2.6a)

d2y/dx2 [1 + (dy/dx)2P~

(2.12)

By summation of forces

f
M
=

+C
-c

fx dA

f+c
-c

Ey

For small deflections the quantity dy/dx is small compared unity; therefore dA

=0
(2.13)

By summation of moments

f+C

yfx dA

= f+c (E 2) dA _)'_
-c

-c

E = - f+C
r-c

Substituting
y2 dA

Eq. 2.13 into Eq. 2.9 gives: M

=

(2.7) By Eq. 2.5: dM

EI d y

dz2

(2.14)

Also, by definition (29) the moment of inertia is: I Therefore M=

+C
-c

= Vdx = Eld3y = V
dx3 (shear force) (2.15)

y2 dA

(2.8) therefore (2.9) dM dx

EI r

And by Eq. 2.4 dV = -w dx therefore

By combination of Eq. 2.9 with Eq, 2.6a, M =fxI

y

(load)

(2.16) of the

or for the outermost fiber where y I

= c,
z (2.10)

f = Me = M
x

Another important relationship, the equation deflection curve, is obtained from Eq. 2.14:

where z I

= = =

L section modulus, c

(2.17) inches! (2.11) These relationships for beams may be applied to plates and shells under certain conditions, as described in subsequent chapters of the text.

moment of inertia of the cross section, inches" distance from neutral fiber to outermost fiber

22 2.2

Criteria in Vessel Design ELASTIC INSTABILITY Euler Column Formulas for Various End Conditions Condition Equation 7r2EI 1. Both ends pivoted (2.18) P critical =--l22. One end fixed, other free 3. Both ends fixed 4. One end fixed, other pivoted
Pcritical

Table 2.1.

Elastic instability is a phenomenon associated with structures having limited rigidity and subjected to compression, bending, torsion, or a combination of such loading conditions. Elastic instability is a condition in which the shape of the structure is altered as a result of insufficient stiffness. It is often the controlling factor when compressive loads are involved. A typical example of elastic instability is the buckling of a cylindrical vessel under an external pressure as a result of vacuum operation. Another example is the buckling of a horizontal vessel as a result of the bending moment induced by the reaction between the vessel and saddle supports. Elastic instability in vessels is usually associated with the use of thin shells. 2.2a Column Instability. The simplest type of elastic instability occurs in the "column" action of an axial, endloaded compression member. The mathematical relationship for critical loading of long, slender columns was first developed by Euler (39) over 200 years ago. The relationship for such a column, pivoted at both ends and free to rotate, may be derived as follows. In reference to Fig; 2.4, the bending moment at distance x is equal to -Py, and by Eq. 2.14 M
=

7r2EI

4J2

(2.21) (2.22) (2.23)

. 47r2EI P critical = ..-l227r2EI P critical = -l-2-

When z = l,.y also equals 0, or y = B sin bl = 0 But B can not be zero if y is to have values other than. zero between z = 0 and z = l. Therefore sin bl must equal zero. If sin bl is equal to zero, the least value of the term bl will be '/r. As the least value of b determines the least value of P, bl is taken as equal to 7r for the critical value of P, or bl

EI dx2

d2y

7r

-Py

therefore

Rewriting with b2 = PIEI, d2y dx2

gives: therefore
=
2

+by

Pcritical

=T

1f2EI

(2.18)

The solution of this differential equation is (40): y = A cos bx + B sin bx where A and B are arbitrary constants. For z = .0, y = 0, or y=AcosO+BsinO=O therefore A

But I may be expressed in terms of the radius of gyration, k, and the cross-sectional area, a., by (41): (2.19) Substituting Eq. 2.19 into Eq. 2.18 gives:
icritical

Pcritical

--a-

z2Jk2

2 7r E

(2.20)

=0
p

Fig. 2.4.

Column pivoted at both ends and free ta ratate.

The stress, icritical, is the load per unit area at which incipient buckling occurs. This is not the maximum stress developed as a very slight increase in Pcritical will result in a considerable amount of deflection and a rapid increase in stress until failure by buckling ensues. For design, an allowable stress appreciably less than the value of icritical is used to provide a margin of safety against buckling. In addition to the condition of a long column pivoted at both ends and free to rotate, column action for other types of end loading may be developed (31). Table 2.1 lists Euler column formulas for various end conditions. For Rankine column formulas, see Chapter 4, section 4.3b. 2.2b Vessel Shells under Axial load. In the design 'of vessels the relationship for the elastic stability of a curved plate subjected to an axial compressive load is of interest because this condition commonly exists in the shell of vertical cylindrical vessels. Timoshenko (42) has given the derivation of the following relationship:

Plastic Instability
!criticsi

23

vi3(1E -

p.)

2 (~)

(2.24)

(for p. where t r
p.

= 0.3)

= shell thickness, inches = shell radius, inches

Poisson's ratio

Experimental tests (42) on the axial compression of thin cylinders have resulted in buckling loads which are about 40 % of that predicted by Eq. 2.24. The safe compressive stress that can be carried without buckling was investigated by Wilson and Newmark (43). AB a result of these tests and others (44), it was found that the safe compressive stress that can be imposed on a steel cylindrical shell without failure by wrinkling can be expressed as follows:
!sllowsbie

= 1.5 X 106

G) ;;:;;; t

yield point

(2.25)

Various applications of the criterion of elastic instability are discussed in subsequent chapters of this book. 2.3 PLASTIC INSTABILITY

2.3a Stress-Strain Relationships. The most widely used criterion in the design of equipment is that of maintaining the induced stresses within the elastic region of the material of construction in order to avoid plastic deformation resulting from exceeding the yield point. These stresses must be limited to a permissible value that is accepted as being safe for the particular application. Usually the results of tensile tests of standard specimens are taken as the basis for establishing the allowable or safe working stress. Ductile materials such as hot-rolled mild steel have two significant stress values, the yield point and the ultimate tensile strength. The yield point is defined as the tensile load at yield expressed in pounds divided by the

original cross-sectional area of the specimen in square inches. The tensile load at yield is the load condition where permanent strain begins to occur. The ultimate tensile strength is defined as the maximum tensile load divided by the original cross-sectional area of the test specimen. The stresses measured by the standard tensile test are induced by a uniaxial load whereas actual loads under operating conditions may induce three-dimensional stresses. Various procedures are employed to handle the problem of combined stresses when using an allowable stress based on uniaxial tests. The procedures differ with the nature of the design problem (see Chapters 6 through 15). Some typical stress-strain curves for various materials were presented in Fig. 2.1. Of the curves shown, only the curve for a hot-rolled mild steel has a well-defined yield point, which occurs at about 30,000 psi. The functional service of a member may be lost if the induced stresses exceed the yield point. For example, a machined flange used as a closure for a vessel may no longer produce a pressure-tight seal if the machined face of the flange is permanently deformed. Thus the allowable stress in such an application should be kept below the yield point. Figure 2.5ashows a simple stress-strain curve for the case of an induced stress within the elastic region. Under load, the part may have an induced stress of A pounds per square. inch and a unit strain of A" inches per inch. On removal of . the load the stress and strain will both return to zero. Such a stress condition is considered satisfactory for design if the induced stress at point A is kept sufficiently below the yield point to provide an adequate margin of safety. Figure 2.5c shows a similar curve for the case in which plastic deformation has occurred. The loading condition has produced a stress which has exceeded the yield point A and has reached a stress of B pounds per square inch. If plastic deformation had not occurred, the theoretical stressstrain condition would have been located at B'. The actual strain resulting from an induced stress B is indicated by B". The permanent residual strain upon' remo~al of

'r------e en

B' I

VI VI

AI

VI VI

en

e
Unloading

o~~----------------------~ Strain __
-c
(a)
Fig. 2.5. Stress-strain diagrams for elastic and plastic loadings. (b) (According to Kerkhof [45].

AU

Strain_

(e) Courtesy of the American Welding Society.)

----,---

24

Criteria

in Vessel Design do not need to conform to pressure-vessel code, specifications but should conform to local building codes. A hotrolled mild steel with an allowable stress of 20,000 psi might be used for such a structure. Inspection of Fig. 2.1 shows that this value is about two thirds of the yield point of SA-285, Grade C steel (which is a typical hot-rolled mild steel). If the vessel supported by this skirt is fabricated from a "code steel," such as SA-285, Grade C steel, having the same physical properties as the skirt steelrthe allowable stress is based on one-fourth of the ultimate tensile strength rather than two thirds of the yield point. Thus the allowable stress for the code vessel using this steel is 12,650 psi rather than the 20,000 psi which might be used for the skirt design. The hazard of an exploding pressure vessel is great, a fact which justifies the use of a greater factor of safety for pressure vessels than for structures. However, the reason for the use of the ultimate strength to define the allowable stress is not obvious and has been the subject of considerable discussion (45, 47, 48, 49). One reason for the use of the ultimate strength as a criterion for allowable stress has been the lack of a plastic zone for brittle materials, such as gray cast iron, and the lack of a well-defined yield point, as in the case of most nonferrous materials. If the yield point is not well defined, the value of a yield point corresponding to some specified permissible strain may be obtained. For example in the 0.2 % offset method a line is drawn parallel to the modulus line from the 0.2 % elongation point, and the intercept of the stress-strain curve with this line is taken as the yield strength of the material. The curve for gray cast iron shown in Fig. 2.1 indicates no yield point, and fracture occurs at the ultimate strength; therefore it is necessary to base the allowable stress for gray cast iron and other brittle materials on the ultimate strength. Because of the great use made of cast iron in design during the period prior to World War I, the policy of basing the allowable stress on the ultimate strength was widely used even for materials which had well-defined yield points. Many engineers still .use a "factor of safety" of three for structural steel and a factor of safety of six for gray cast iron, based upon the ultimate strength, when designing structural parts. The pressure-vessel codes still use a factor of safety of four based upon the ultimate strength for specifying the allowable stresses for pressure vessels. This prior convention of applying a factor of safety to the ultimate strength does not justify the continued use of this / . procedure. Where failure can be expected to occur as a result of plastic deformation, the yield point should be used as a basis for determining the allowable working stresses. However, if the vessel is designed to meet code requirements, the procedure specified in the codes must be used. It should be pointed out that the maximum allowable working stress specified by the codes is not always based on the ultimate strength, The criteria used in establishing the allowable stresses in the ASME code (U) follow.

the load is indicated by point C. If the entire cross section of the part undergoes plastic deformation, as shown in Fig. 2.5c, there will be no residual stress upon removal of the load. Such a design is usually considered unsatisfactory because of the excessive permanent deformation and the possibility of rupture. If part of the cross section is subjected to elastic strain, as indicated in Fig. 2.5a, and the remainder of the section undergoes plastic deformation, as indicated in Fig. 2.5c, residual stresses and strains will remain in the cross section upon removal of the load condition. Figure 2.5b shows the loading and unloading condition that results when only a portion of the cross section undergoes plastic deformation. It is important to note that the plastic strained condition resulting from stress B' is much more limited in this case than in that shown in Fig. 2.5c because of the restraint offered by the adjacent elastic portion of the cross section which is undergoing elastic strain. Thus the total strain in the portion undergoing plastic strain is limited to that predicted by the modulus-of-elasticity line extended to point B'. The 'actual induced stress in the plastic portion corresponds to B pounds per square inch, and the actual strain, to B" of Fig. 2.5b. Upon unloading, the portion having undergone plastic strain has a residual compressive stress as indicated by point C. This residual compressivestress condition is in equilibrium with residual tensile stresses in the adjacent region that has been subjected to only elastic loading. If the portion undergoing plastic deformation is small in comparison with the portion undergoing elastic deformation, the residual strain will be imperceptible. Thus the prevention of significant plastic deformation does not require all calculated elastic stresses to be below the yield point since appreciable plastic deformation can occur only if the material yields across the entire area. Such a loading condition as' shown in Fig. 2.5b often exists where local stress concentrations (which are nonuniform across the section) occur, as at the junction of vessel shell and heads. Because the major part of the cross section is in elastic strain, the small amount of plastic strain relieves the high stress from B' to B without serious deformation. Also the mean stress across the elasticplastic zone may be sufficiently below the yield point to allow an adequate margin of safety. Thus such a condition may have advantages in relieving high local stresses but may become undesirable if excessive repeated loading and unloading occur. Such a cyclic operation may result in strain hardening with corresponding loss in ductility and subsequent failure by rupture (45, 46). ~ 2.3b Allowable Stress. On reference to Fig. 2.5a, the percentage of the yield strength used as the allowable stress is controlled bya number of factors, such as: the accuracy with which the loads can be estimated; the reliability of the stresses computed from these loads, the uniformity of the material, the hazard if failure occurs, and other considerations like local stress concentrations, impact shock, fatigue, and corrosion. For structural steels, one half to two thirds of the yield strength is often used as the allowable stress for static loads in structures. For example, the skirts used to support tall vertical vessels may be considered structures and therefore

1. At temperatures below the creep range, allowable stress values were established at the lowest value of stress obtained from: (a) 25% of the specified minimum tensile strength at room temperature; or (b) 25% of the minimum

Plastic 663s% of yield Material tow-carbon nickel 2S aluminum Copper 54S aluminum 18-8 stainless Carbon steel Carbon steel Low-alloy steel Seamless quenched.and-tempered steel Navy G steel Yield Percentage of ultimate strength 25 30 33.3 40 50 60 70

Instability

25

ASME code designation SB-162 SB-178-996A

sa-n
SB-178-GR40A SA-240 and SA-167 SA-201 SA-212 SA-302 Code case '*1134 None None

T-l steel

~ ultimate 62.5% of yield Percentage of ultimate strength

.\ Fig. 2.6. Comparison of allowable stress and yield stress as a percentage of the ultimate strength for materials within atmospheric temperature range.

(According ta Zick [48].

Courtesy of the American Welding Society.)

expected tensile strength at operating temperature; or (e) 62H % of the minimum expected yield strength for 0.2 % offset at operating temperature. 2. For bolting material used at temperatures between -200 F and 4000 F the stress value were based on 20% of the minimum tensile strength or 25 % of the minimum yield strength, whichever was lower. Criterion Ic introduces a further restriction for materials that have low yield-strength-tensile-strength ratios. Appendix D lists allowable stresses for various materials as specified by selected codes and standards. The appendices of the ASME code for unfired pressure vessels (11) describe the basis for establishing values of allowable stresses for both ferrous and nonferrous materials. Zick (48) has shown graphically (see Fig. 2.6) the allowable stresses and yield strengths for various materials as functions of the percentage of the ultimate strength. The percentages given in Fig. 2.6 are those specified by the code (11) except for the cases of the two high-strength steels shown at the bottom of the figure, which as yet have no code designation. The nonferrous materials shown in

Fig. 2.6 have allowable stresses based on two-thirds of the yield stress, which for these materials is an allowable stress less than one-fourth of the ultimate. Ferrous materials shown have an allowable stress based on one-fourth of the ultimate strength, which for these materials is an allowable stress less than 62H % of the yield stress except for 18-8 stainless. The allowable stress for 18-8 stainless satisfies both criteria (25% of ultimate and 62H % of yield strength). If the criterion is a factor of safety applied to the yield strength, then the ratios between allowable stress and yield strength are inconsistent for the higher-strength materials. An allowable stress based on the yield point assumes that failure occurs by plastic deformation. If failure may be expected to be caused by rupture rather than by excessive plastic deformation, the use of the ultimate strength as a criterion for the allowable stress may have justification on the basis that the fatigue limit, which controls failure by rupture, is usually proportional to the ultimate strength (48). It should be noted that failure by rupture has seldom occurred in vessels fabricated of code-approved low-carbon steels haying high ductility.

26 2.4

Criteria in Vessel Design BRITTLERUPTURE strain hardening of ductile materials in a test to rupture, and strain hardening resulting from local overstressing in repeated cyclic loading. 2.4a Notch Brittleness. Mild steels show high elongation in the simple tensile test and are normally considered to be ductile materials. Such materials can fail with little or no evidence of plastic strain if the material contains a crack or notch and if the material is at a service temperature below the "transition temperature" of thematerial, This type of failure is known as "notch brittleness" and has resulted in the catastrophic failure of a number of welded ships and a number of storage vessels (53, 64). Several texts and articles discuss the phenomenon of notch brittleness (52-66). The transition temperature is defined as the temperature above which the ductile type of failure occurs. Belo\V the transition temperature a transition range may exist in which the material hassemibrittle properties. At still lower' temperature the material becomes completely, brittle.

The current trend toward the use of higher-strength steels having lower ductility increases the possibility of failure by rupture. This has resulted in a number of investigations and technical papers dealing with this problem (45, 46, 50, 51, 52, 106). Stress concentrations are known to exist in a part under load where there are changes in shape or cross-sectional area. Very often these stress concentrations may be evaluated in such parts as the junction of vessel closures and vessel shells. The code for the design of pressure vessels states that such "stress shall be considered" but does not indicate the procedure for this consideration. The customary practice has been the use of generous factors of safety and due-tile materials. The use of high factors of safety results in the overdimensioning of vessel sections. Such overdimensioning, when used in conjunction with highly elastic materials, usually permits the dissipation of local stresses

Fig. 27.

Ductile rupture in a multilayer vessel purposely tested to destruction.

(Courtesy of A. O. Smith Corp.1

by limited plastic deformation without failure by rupture. Currently the Pressure Vessel Research Council is supporting research on materials having higher yield and tensile strengths. If no special corrosion problems are involved, failure, if it does occur, is usually caused by either: (1) excessive plastic strain (ductile rupture) or (2) brittle rupture (45). Figure 2.7 shows failure by excessive plastic deformation in a multilayer pressure vessel purposely tested to destruction. Such failure, discussed in the previous section, can occur only if a high stress is distributed over a large area. Local stresses never produce great plastic deformation because small plastic deformation serves to relieve these stresses. This type of failure seldom occurs in a properly designed vessel. Figure 2.8 shows the fragments of a 5000 psi monobloc vessel purposely tested to destruction. The fragmentation is typical of brittle rupture. Brittle rupture may result from: the use of brittle materials, "notch brittleness,"

Below this temperature of complete embrittlement, brittle fracture may occur even though no notches or cracks exist in the material. In the transition range a notch or crack must exist for brittle fracture to occur. Above the transition temperature brittle fracture will not occur even if such, a notch exists. Thus, failure below the transition temperature is referred to as "notch brittleness." Some materials have a very narrow transition range; therefore 'a single transition temperature is sufficient to define the transition from the ductile to the brittle type offailure. Other materials, such as plain-carhon or low-alloy-ferritic steel, have transition ranges of hundreds of degrees. IMPACT TESTS. The transition-temperature range is usually determined by making Charpy or Izod impact tests at various temperatures. The procedure for the Charpy impact test for plate steels for vessel construction is described in ASTM designation A 370-54T (67), and the minimum impact strength permitted is given in ASTM designation A 300-54aT (67). The general procedure involved in such

,j

Brittle Rupture

27

STRAIN ENERGY. When a load is applied to an elastic material, the material deforms in the direction of the force, and work is done upon the material. This work is equal to the product of the average force times the .distance through which the force moves. If the initial force is zero, the average force is equal to one half of the final force. When the load is removed from an elastic body, it returns to its original shape, and in so doing it has the capacity for doing work. Thus an elastic material under load may be said to have "strain energy." Consider a cubic inch of elastic material initially under no load and apply a force sufficient to produce stress, 1. The average force in terms of stress is equal to U/2), and the unit deformation resulting is equal to:
E

=-

fig. 2.8.

Brittle fracture

in a monoblac vessel designed

for 5000 psi

(2.3)

and purposely tested to destruction.

(Courtesy of A. O. Smith Corp.)

Therefore,

the strain energy,

U, is equal to:

U=--=-

11

12
2E

2E

(2.26)

tests is to prepare rectangular specimens with either a V, V, or keyhole notch machined across one face of the specimen. The specimen is clamped in a vise with the notch exposed and facing a heavy pendulum. The pendulum is released in such a way that it strikes the specimen with an impact blow which causes the specimen to fracture at the notch. The maximum swing of the pendulum after interaction with the specimen is measured. From this measurement the foot-pounds of energy expended in causing the fracture may be determined. A brittle material fractures with little or no plastic deformation, and the amount of energy required for fracture is small. A tough material undergoes considerable plastic deformation prior to plastic fracture. This plastic deformation absorbs a good deal of energy. Thus the foot-pounds of energy absorbed in causing fracture is a measure of the toughness of the material. Figure 2.9 shows curves for typical data from Charpy U-notch tests for a variety of mild steels (53). EFFECT OF COMPOSITION. All of the four steels shown in Fig. 2.9 have carbon contents between 0.20 % and 0.25 % and yield strengths between 33,000 and 40,000 psi. The curves for the three steels shown to the right in Fig. 2.9 exhibit complete brittleness at temperatures of 0 to. -10°F, whereas the ABS class C steel exhibits this phenomenon at a temperature about 30° F lower. The transition temperature for all four steels covers a wide range. In general, steels with lower carbon contents exhibit greater toughness at lower temperatures. The presence of phosphorus has been shown to decrease the transition temperature. A correlation has been developed (60) indicating that the transition temperature is a function of the sum of the percentage of carbon plus 20 times the percentage of phosphorus (C + 20P). The addition of nickel to steel can greatly increase its toughness and lower the transition temperature, Steel designated as AISI 2800 (8H % nickel) and 304 stainless steel will withstand impact loads at temperatures as low as -320° F (liquid-nitrogen saturation temperature) (61).

For a volume of elastic material larger than 1 cu in., the total strain energy becomes equal.
Utotal vol.

= (U)(volume)

(2.27)

GRIFFITH THEORY. A crack extending through a unit thickness of a plate under stress deforms to produce an opening in the plate which may be considered to have the shape of an ellipse with a very large major-to-minor-axis ratio. The volume of such an opening having a length equal to 2c is a function of c2• The strain energy released as a result of formation of the new crack volume in the elliptical opening can be determined by use of Eqs, 2.26 and 2.27. U t.v.

= 2E (ac2l r

(2.28)

where the volume of the elliptical opening is ac2l.

40 36 32 ABS class C....

= .£ tiD s
t;

:9 28 I

24 20 16

V
I

j;7

l..-

.... u
'"

1/

"-M A-285~ Grade C-

§ 12 8 4 -60

Co

/
/
-40

I ASTMA)-.,I

"_
-20

./ '/ /

'_J
/

V/

Vj 7 /'

I.....?'

v::V
/'

k-':::

10-f-

r-

BS-13

I
60 80

o 20 40 Temperature, degrees F

.;. fig. 2.9. Charpy U-notch data for some mild steels (53). American Welding Society.)

(Courtesy of the

28

Criteria

in Vessel Design
B'

em (70). The energy required to form surface' area of metal at the crack interface is given by:
Ucrack

= 4cT

(2.31)

The rate at which energy is consumed as the crack propagates is:

dUcrack

= 4T

dc

(2.32)

The criterion for propagation of the crack is determined by the ratio of the rate of strain energy released to surface energy absorbed in creating the new surface at the interface, or (2.33) or

V 2ET /7rC

(2.34)

c::

.~

Strain _

~
E

Q.

It should be noted that when the ratio given by Eq. 2.33 . exceeds unity, a case of instability exists. For this condition the crack will propagate at an increasing rate, approaching one third the speed of sound in the material (several thousand feet per second for steel). Such failures occur so rapidly and are so extensive as to be catastrophic in most instances. Substituting the approximate value of 2 X 106 ergs per sq cm (70), which is equal to 11.4 in-Ib per sq in., for Tin, Eq. 2.34 and rounding off the product (2X 11.4)/7r to 10 as an approximation gives:

f = VI0E/c
p
.,:t Fig. 2.10. Strain hardening due to cyclic load. (According to Kerkhof [45]. Courtesy of the American Welding Society.,

(2.35)

If the stresS f reaches the yield point, plastic deformation 'will occur. Therefore the smallest crack that can initiate brittle fracture is determined by setting f in Eq. 2.35 equal to the yield POn1t (y.p.) and solving for c.
Ccritical

= f----:i
y.p.

10E

(2.36)

Griffith (68, 69) in a rigorous derivation has shown that for a plate of unit thickness this expression becomes: '7rc2f2 Ut .v.. =--E (2.29)

The rate of release of strain energy with increase of crack dimension c as the crack propagates is obtained by differentiation of Eq, 2.29. dUt.v. dc
=

27rcj2 E

(2.30)

The surface area of metal in the 'crack interface at the time of formation of the crack is twice thep;odli-Ci of the crack length times the metal thickness, or 4c, for a plate of unit thickness. The formation of this surface consumes energy which is normally equal to the product of the area times the surface tension, T, for brittle materials in which fracture occurs with little plastic deformation. If some plastic strain accompanies the fracture, T will be greater than the surface tension and must be experimentally determined. The value of T for low-carbon steel at room temperature is approximately equal to 2 X 106 ergs per sq

Thus a steel (S~-283, Grade C) having a yield point of' 30,000 psi would not propagate a crack shorter than % in. However, a low-alloy steel (ASTM A-242) having a yield strength of 50,000 psi would propagate a crack having a length of about. ~iin. lt2.4b Repeated Cyclic Loading. Brittle rupture can occur without appreciable plastic deformation as a result of local high stresses and repeated cyclic loading. Such failure may occur near limited areas of stress concentration, near defects in the plate, or pear weld joints. This type of failure does not occur during hydrostatic tests in spite of the fact that the stresses are higher than those induced in service because they are not repeated so that they produce fatigue. Failure by rupture usually begins by the formation of a tiny crack after the vessel has been in service for a considerable period of time with cyclic loading operation. These small cracks continue to propagate with time. The material surrounding the cracks becomes strain-hardened and brittle. The extension of the cracks continues through the strain-hardened area and stops when ductile material is encountered. After continued stress cycles, the material at the root of the cracks becomes strain-hardened and the

Brittle Rupture

29

Stress-strain diagram

"L "L "11=

(f (f (f

EO

EO

EO

(a)

(b)

(c)

Type of curve Strain hardening considered

fig. 2.11. Engineers.)

Power function Yes

Straight line Yes

Idealized (ex 0)

No

Stress-strain curves flHing mathematical

equations.

(According

to Burraws et al. [77].

Courtesy af the American Saciety of mechanical

cracks progress further. The continued strain hardening and progression of the crack result in eventual failure by brittle fracture. Failure by rupture as a result of strain hardening can be explained with reference to the stress-strain diagram given in Fig. 2.10-according to Kerkhof (45). During the application of the hydrostatic test load or during the first loading, local stress concentrations develop according to the pattern given in Fig. 2.5b. If a small portion of the cross-sectional area of the shell thickness is under a sufficiently high stress intensification, removal of the load will result in reverse yielding as shown by region CD in Fig. 2.10. Reverse yielding occurs as a result of an induced compressive stress which exceeds the compressive yield point C and produces plastic strain to point D. Such a strain can occur where bending stresses exists such as near the junction of vessel closures and attachments. Bending stresses may be very high on the shell surface and zero at the center of the shell thickness. Upon reloading, the path DEF is followed, and upon unloading, the path FGH is followed. After each successive cycle the yield point of the material increases as a result of strain hardening. After a sufficient number of cycles a stress of Q is reached in the' loaded condition, and of P in the unloaded condition without appreciable plastic deformation. If these stresses reach the fatigue limit of the material, minute cracks will form in the strain-hardened material. Continual cyclic operation will result in eventual brittle rupture It should be noted that if the parallelogram OB'QP, the sides OB' and PQ have equal length, and if Q is the tensile stress equal and opposite to the compressive stress P, then Q = ~-2B' (45). Thus if the maximum value of the theoretical stress B' does not exceed twice the yield point A, the maximum value of Q will not exceed the yield point A, and fracture by strain hardening will not occur. The maximum stress range QP must not exceed:

The above relationship indicates that in the consideration of failure by brittle rupture, discussed previously, the allowable stress should be based upon the yield point rather than upon the ultimate strength of the material. Equation 2.37 does not take into account the phenomenon of stress relaxation at elevated temperatures and was developed for uniaxial stresses (45). Such considerations and the application of Eq. 2.37 .are discussed in a later section of the text. A number .of articles have appeared reporting studies on the problem of brittle rupture as a result of repeated cyclic loading (50, 52,71-76). For purposes of analysis of stress-strain relationships, the stress-strain curves can be approximated as indicated in Fig. 2.11 (77). The stress-strain curves shown in Fig. 2.11 a, b, and c are termed: power-function, straight-line, and idealized curves respectively. They can be represented by the following equations:

1 = a1Eo.51! 1 = I» + (10 1 = 1o
2.4c

(power-function, Fig. 2.lla) EO) tan ex

(2.38) (2.39)

(straight-line, Fig. 2.pb) (2.40) to Brittle Rupture.

(idealized, Fig. 2.11c) Other Factors Contributing

HYDROGENEMBRITTLEMENT ND BLISTERING. HydroA gen will diffuse into steel under certain conditions. The action of hydrogen at high temperatures and pressures differs from that at low temperatures and pressures. When steel is exposed to hydrogen at high temperatures and pressures, the steel loses its tensile strength, becomes brittle, and often cracks or blisters. The mechanism of diffusion of hydrogen athigh temperatures and pressures into steel is believed to result from the dissociation of hydrogen molecules to monoatomic hydrogen. The partial pressure of monOatOmic l\I.ydrogencauses the hydrogen to diffuse into the steel (78). '. As the hydrogen diffuses into the steel at high temperatures, it reacts with the carbon in the steel to form methane. The methane does not diffuse out of the steel and accumulates to form blisters and cracks. Figure 2.12 shows sections through hydrogen blisters in pressurevessel steels (78).

1 range max
where Imax
range =

2Iy.p.

(2.37)

!Y.P.

maximum local stress range not producing fatigue failure, pounds per square inch initial yield point of the material at the operating temperature, pounds per square inch

Fig. 2.12. Sectian cut through hydrogen blisters in pressure-vessel (Courtesy Society.) of Shell Development Company and the American

steel (78). Welding

30

Criteria

in Vessel Design the precipitation and may be measured by impact tests as a function of aging (79, 80). ' TEMPEREMBRITTLEMENT. he phenomenon of temper T embrittlement occurs when hardened medium-carbon structural steel is cooled slowly or held within a critical range of temperatures below the temperature at which austenite is transformed to ferrite. This critical=temperature ...range usually occurs in somewhere between, 850 and .11000·F. Welded joints and service at elevated temperatures are subject to this phenomenon. The effect is accentuated by high Mn, P, and Cr and is retarded hy Mo. Thepheneme-, non is not completely understood but is believed to result from a precipitation mechanism (80, 81, 82, 83). 2.5 CREEP

At low temperatures and pressures the mechanism of hydrogen diffusion is believed to be associated with the formation of hydrogen ions as a result of corrosive attack. The hydrogen ions are converted to monoatomic hydrogen by means of electron exchange. This conversion takes place, for example, when a steel part is cathodic and deionization occurs with the flow of a small amount of current. Energy is required to cause the hydrogen to penetrate the steel. This energy is supplied by the current from the galvanic action of corrosive attack. The driving energy of a fraction of a volt in an electrolytic cell i,s equivalent to many thousand atmospheres of hydrogen pressure (78). The embrittlement caused in a vessel by hydrogen diffusion is temporary. If the equipment is shut down for a period of time, the hydrogen will diffuse from the metal. If the equipment is cooled slowly, the rate of hydrogen diffusion from the metal will be greatly increased. Annealing for two hours at 12000 F or for one day at 2250 F will return the ductility to normal. The loss in ductility caused by hydrogen diffusion is not excessive. For example, a hydrogen content of 14 ppm in 1020 steel will reduce the tensile-test elongation from 40 % to 22 %. Many process plants operate normally without giving any particular consideration to hydrogen embrittlement. However, blistering and cracking is a serious problem with equipment handling hydrogen at high temperature and pressure. STRAINAGING. When metal is permanently deformed beyond the elastic limit by cold working, a precipitation may occur because of local supersaturation along slip planes in the microstructure of the metal. Thus cold working has two effects. First, the hardness is increased at the time of cold working; and this increase is followed, on the aging of the metal, by an additional increase resulting from precipitation. Second, the toughness decreases because of

c: o 'iii c:

Time_
oj.. Fig. 2.13.

Schematic creep curve, extension plotted against elapsed time. (A, elastic extension; B,creep at decreasing rate;. C, creep at approximately constant rate; 0, creep at increasing rate; E, elastic contration; F, permanent change of sha pe.)

The criteria for design discussed previously have be~n based upon the premise that strain under load does not vary with time. This premise is essentially true for ferrous materials under load at temperatures up to about 6590 F. However, beyond this temperature range the material "creeps" under load, causing an increase of strain with time. An increasing rate of creep is encountered as the service temperature is increased; Some materials, such as lead, creep readily at room temperature. The rate of creep depends upon the prior history of the material and' the stress as well as upon the temperature. 2.50 Creep Test. In studying the creep characteristics of a material, a small tensile-test specimen is placed under constant axial load while held at constant temperature in an electric furnace. The rate at which the sample elongates is recorded as a function of time for each temperature and load. Depending upon the test conditions and the material, the duration of the test may last from afew hours to several months. and on occasion has been continuedfor several years. Figure 2.13 shows a typical creep-rate curve obtained from such an investigation. The general specifications for conducting a creep test are covered by ASTM specification E-22-41 (67). Upon application of the initial load an instantaneous elastic strain occurs, resulting in an extension of the specimen as indicated by extension A of Fig. 2.13. The initial creep begins and continues at a decreasing rate for the time interval B; this region is known as the "first stage of creep." This period is followed by a constant-rate period extending over the time interval C; this region is known as the "second stage of creep" and is the region which is used to limit the service life of the equipment. The constant-rate zone is followed by an increasing-rate period over the time interval D. This is known as the "third stage of creep" and ends in fracture of the specimen if the test is continued. Usually the test is interrupted before fracture occurs, and the specimen undergoes an elastic contraction as indicated by extension E. The amount of permanent strain is indicated by extension F. Two typical creep curves for a high-alloy steel tested at 1200° F with stresses of 20,000 and 25,000 psiare shown in Fig. 2.14 (84). It should be noted that increasing the stress from 20,000 to 25,000 psi greatly increases the creep rate and shortens the service life of the material. If a number of curves such as those shown in Fig. 2.14 are obtained, the

0.080 0.070 0.060

J
/_

Creep
Fractured after 3545 hours
I I I I

31

<.::::.

0.050 0.040 25,0~/ 0.030 0.020 0.010

+- Fig. 2.14. at (84).

Time-elongation curves "16-15-6" alloy-from 2150° F

.E

1200°F,

solution quenched Bearing Co.)

(Courtesy of Timken Roller

'"
I:

tIO 0

".,20,000 psi In progress 6683 ~rs

~
1000

500

1500

2000

2500

3000 3500 4000 4500

Time, hours

5000

5500 6000

6500

7000

slopes of the second stage (constant rate of creep) may be determined and a cross-plot prepared such as the one as shown in Fig. 2.15 (85). In Fig. 2.15 the creep rate is plotted against stress with temperature as a parameter. Two standards of creep strength are commonly used for design purposes in this country. One is the stress -that will produce a creep rate of 1 % per 10,000 hours (approximately 1 % per year), and the other is the stress that will produce a creep rate of 1 % per 100,000 hours (approximately 1 % in 11 years). 2.Sb Creep-Rupture Test. Another consideration is the rupture life of the part under the stress condition at the service temperature. A test known as the stress-rupture or creep-rupture test (67) is used to determine this rupture life. It is similar to the creep test except that high loads are used which produce greater creep rates so that rupture occurs in a reasonable length of time. The total strain obtained in a creep test is usually below 7-2 whereas in the creep% rupture test the total strain usually exceeds 50 %. Figure 2.15 shows curves plotted from both creep-test data and

creep-rupture-test data. The time to rupture is plotted as a function of stress with parameters, as indicated in Fig. 2.16 (85). The breaks in the three curves of Fig. 2.16 are the result of a change from ductile to brittle fracture. It is essential that the rupture tests be conducted for sufficient time to insure that the slope of the curve beyond the break is established. It is customary practice to extrapolatethis portion of the curve to 100,000 hours in order to obtain the stress to produce rupture for longer periods of time. The ASME code (11) specifies that the allowable stresses at high temperatures are based.on 100 % of the stress to produce a creep rate of 7100 % per 1000 hours (0.00001 % per hour) or on 60 % of the average stress or 80 % of the minimum stress to produ.ce rupture at the end of 100,000 hours, whichever is the lower value. The stress to produce rupture after a specified length of time varies with the material. Figure 2.17 compares the stress to produce rupture in 1000 hours as a function of temperature for a variety of materials (86). The superalloys shown in Fig. 2.17 may be rather expensive for use in

Creep test
't:J

Creep- rupture test

'"
c/)

fij 100

,. Fig. 2.15.

Correlation

of creepdata Car-

test and creep·rupture-test steel (85). (Courtesy of

o :5

""
10

for 18-8 molybdenum {type 316) neg ie-Illinois Steel Corp.)

'iii

c.
vi
c/)

= 5 =l--T

1l00°F 1300°F

•
~ ~

e en

~
I IIIII

~
0.0001

--- ----r

0.00001

0.001

om

0.1

10.0

100

Minimum rate of extension. per cent per hour

32
100

Criteria

in Vessel Design

-------------

--

- r--

1300°F

I---

___,__

-iii 10
0-

.,;

f:
l-

'"

5 II-

fo-

f:

--- -------- -------..... ------- r--____-_

r--_

1l0O°F

- ---_._-- r-- __

9"

Fig. 2.16.

Stress vs. rupture time

for 18-8 inolybden~m (type 316) steel at 1106, 1300, and 1500° F (85). (Courtesy" of CarnegieIllinois Steel Co~p.l

--

1 0.1·

I 11111

10

100
Rupture time, hours

1000

10,000

100,000

~·I

vessel construction; they find their principal use in such items as turbine blades, parts for jet engines, and similar applications, but may be used for vessels for extreme service. It should be noted that creep- and stress-rupture-test data are obtained under atmospheric exposure conditions in laboratories and under uniaxial loading. The stress condition existing in a vessel part under field service conditions usually comprises stresses in three directions, a fact which complicates the application of the. experimental data. In addition, the vessel material may be exposed to a corrosive atmosphere and be subject to scaling, hydrogen embrittlement, intergranular corrosion, and strain hardening. 2.6 CORROSION

The extent to which corrosion will occur in process equipment depends upon the nature of the films that form on the

surface of the parts. The excellent corrosion resistance of copper and its alloys, for example, is the result of their ability to form thin, protective films on their surfaces. These films may be the result of simple oxidation or may be composed of insoluble salts. To be protective the coating should be thin, adherent, continuous, and relatively insoluble. Equipment operating under conditions that allow the formation of a uniform protective film generally corrodes slowly and may last for a great many years. Under severe corrosive conditions, however, rapid corrosion occurs, resulting in costly delays and replacements. By judicious selection of material and by careful improving of operating conditions, corrosion can be reduced or retarded, and substantial savings in operating and maintenance costs may be realized. Thus an appreciation of the factors which contribute to corrosion is of value in the design of equip-

.~ 40 I---j.-~~--I

§
.,;

~ 30

-i Fig. 2.17. Stress to rupture in 1000 hours vs.! temperature for various materials (86). (Courtesy of Universal-Cyclops Steel Corp. and Battelle Memorial Institute.)

Temperature, degrees F

Corrosion ment. A thorough treatment of the theory of corrosion and its control is presented in the Corrosion Handbook (87). 2.6a Uniform Corrosion. When corrosion occurs at the surface of equipment by the formation of soluble salts, a uniform thinning of the wall occurs. The rate of corrosion depends upon the corroding medium, the velocity of fluid flow, the temperature, and other factors. This type of corrosion is encountered in acid solutions (particularly those containing oxygen), in waters carrying a high oxygen or carbon dioxide content (such as mine water), and in solutions having a solvent action on the corrosion products (such as those containing ammonium hydroxide, which dissolve the corrosion products from copper alloys). Attempts have been made to reduce uniform corrosion by the application of an external electric current to provide cathodic protection. The limited success attained by this method has been attributed more jo the formation of protective films than to cathodic protection. 2.6b Impingement Attack. Under normal operating conditi.ons certain localized areas of the parts may be exposed to the destructive forces of a relatively high-velocity circulating medium. Corrosion under such conditions is described as "impingement attack." Turbulence,of the fluids causes a rapid and repeated destruction of the protective film with subsequent corrosion of the exposed .nietal. This condition is considerably aggravated when the circulating medium carries in suspension an abrasive material. 2.6c Concentration-cell Attack. Corrosion inay be caused by differential aeration with the formation of concentration cells at the metal surfaces under certain operating conditions. Cracks, crevices, porous coatings, and breaks in protective films are sources of trouble since they trap liquid and set up differences in concentration of salts, ions, or gases in the circulation medium. As a result of an electrochemical type of concentration-cell action, severe pitting of the metal surface and localized perforation of the material occur. An example of this type of corrosion is the "rusting" of plain carbon steel. This type of corrosion can be materially reduced by the observation of the following suggestions (88):
1. Specify butt joints and emphasize the necessity for complete penetration of the weld material to guard against minute crevices. 2. Avoid the use of lap joints, or completely seal them with weld metals, solder, or a suitable caulking compound. 3. Avoid sharp corners and stagnant areas or other sites favorable to the accumulation of precipitates and other solids. 4. Endeavor to provide uniform flow of liquid with a minimum of turbulence and air entrainment. 5. Provide suitable strainers in lines to prevent local obstruction within the equipment which may start deposit attacks or result in impingement attack.

33

1'- Table
(Courtesy

2.2. Galvanic Series in Sea Water (236) of American Society of Testing Materials)
(AN,oDIC on LEAST N,oBLE)

C,oRR,oDED END

Magnesium, magnesium alloys Zinc, galvanized steel, or galvanized wrought iron Aluminum (52 SR, 4S, 3S, 2S, 53S-T in this order) Alclad, cadmium Aluminum, (A 17S-T, 17S-T, 24S-T in this order) Mild steel, wrought iron, cast iron Ni-resist ' 13%-chr.omium stainless steel, type 410 (active) 50-.50 lead-tin solder 18-8 stainless steel, type 304 (active), 18-8-3 stainless steel, type 316 (active) Lead, tin Muntz metal, manganese bronze, naval brass Nickel (active), Inconel (active) Yellow brass, admiralty brass, aluminum bronze, red brass, copper, silicon bronze, Ambrac, 70-30 copper-nickel. eomp. G-bronze, comp. Mvbronze Nickel (passive), Inconel (passive) Monel 18-8 stainless steel, type 304 (passive); 18-8-3 stainless steel, type 316 (passive)
PRQTECTED END (CATHQDIC

oa

MQST NQBLE)

2.6d Deposit Attack. When small particles settle out or lodge on the wall of the equipment, part of the metal wall becomes protected by the deposit, and a special type of concentration-cell action may take place. Usually the shielded area becomes anodic and intense pitting results. Filtering of the circulating medium and frequent cleaning will mini, mize the occurrence of deposit attack.

2.6e Galvanic-cell Attack. When dissimilar metals and alloys are Tn contact ~itheach 9t;her in a conducting medium, a galvanic action iue~l!E. which results in the dissolution of the less-noble or cat!ki'Mc metal. From the electromotive or galvanic series it is' possible to predict the tendencies of metals and alloys to form galvanic cells and to predict the probable direction of the galvanic action. Table 2.2 gives the galvanic series as determined for sea water (89, 236). When use is made of such a series determined for the fluid under consideration, one may be relatively safe in choosing metals from the same group; however, if the metals are distant from each other in the list, the metal higher in the list will corrode rapidly. When one is using metals that produce a galvanic action, the relative areas of the two materials have a very important bearing on the extent of corrosion. Usually the extent of galvanic action will be proportional to the ratio of the area of the metal lower in the series to the area of the metal higher in the list. Thus it is wise to avoid galvanic couples where the exposed area of the cathodic metal is much greater than that of the anodic metal. (For example, a steel part in a copper vessel would rapidly corrode, but a copper part in a steel vessel would be relatively safe). 2.6f Stress Corrosion. As a result of the simultaneous action of stresses and certain corrosive conditions, parts may fail by cracking. 'When the stress is applied externally, the break often is called a "stress-corrosion crack." When residual internal stresses are involved, the resulting break often is called a "season crack." Annealing to relieve residual stresses greatly reduces season cracking. When the part is subjected to repeated cyclic stresses during service, failure by "fatigue cracking" may occur, as dis-

34

Criteria

in Vessel Design
to both plastic and elastic deformation in the same 'section

cussed previously. Such failures are characterized by their suddenness and by the absence of plastic deformation in the failing section. Stress-corrosion data on systems free of corrosion in the unstressed state indicate that there is little or no attack on material that is either completely elastically deformed or completely plasticly deformed. But such parts subjected

may undergo severe stress corrosion. Thus stress corrosion is caused by strain hardening in combination with high elastic stresses rather than by plastic deformation (45). Such strain hardening followed by corrosion has been observed in boilers and pressure vessels where local stress concentrations exist without severe plastic deformation.

PROBLEMS'

1. A full-floating-head horizontal condenser has 400 %-in.-diameter No. 16 BWG admiraltymetal tubes. The floating tube sheet and head assembly weigh 1000 lb when the head is filled with water. If the tube-support plate is located 18 in. from the center of load of the floating head assembly and the overhanging tubes and head assembly are assumed to behave as a simple cantilever beam, a. What bending stress is developed in the tubes as a result of this load ~ b. What is the floating-head deflection from the horizontal at the center of load P c. Is the design satisfactory if the allowable stress is 6000 psi~ For a simple cantilever,
y=

3EI

PI3

For a tube, I,•• taneular =

r(do. - di4) 32

For a %-in. No. 16 BWG tube, OD = 0.625 in. ID = 0.495 in. Modulus of elasticity of admirality metal = 15 X 106 psi

2. A horizontal stiffener supporting a bubble-cap tray in a fractionating column may be considered to act as a uniformly loaded, simply supported beam. The deflection equation for such a beam is: y=

2.. [Wlx3
EI

12

wx·. _ x]
W1 3

24

24

where y = deflection at point z, inches x = distance from end of stiffener, inches I = total length of stiffener, inches E = modulus of elasticity, pounds per square inch I = moment of inertia, Inches+ W = uniform load, pounds per linear inch The stiffener has a moment of inertia of 1.785 in.4 and a section modulus of 1.02 in.8 and is of steel (E = 30 X 106 psi). The stiffener has a length, I, of 100 in. and carries a uniform load, w, of 2.4 lb per linear in. Determine: a. The maximum deflection at the center of the span, x = 50 in. b. The maximum stress at the center of the span. c. The stress 20 in. in from the-support, x = 20 in. d. The shear load, pounds, at x = 20 in. 3. A copper fractionating column has vertical tray-support rods between trays. The trays are ,20 in. apart and the rods are spaced so as to support 100 lb each, under column action. What is the diameter of the rods required to limit the column loading to one half of the critical column load if (a) the rods are free to pivot at both ends or (b) the rods are fixed at both ends P k = ~. E = 15 X 106 psi 4'

Problems 4. A vessel is to be fabricated of type 316 stainless steel (see Figs. 2.12 and 2.13). The vessel is intended to be used at 13000 F for an expected service life of five years. The creep deformation permitted during the service life of the unit is not to exceed 5 %. Determine the allowablestress if the allowable stress is not to exceed either (a) two thirds of the stress to produce the creep rate or (b) 50% of the stress to produce rupture. Also estimate the total creep and rupture life if the allowable design stress as determined in this manner is used in the design of the vessel.

35

CHAPTER

DESIGN

OF SHELLS
CYLINDRICAL VESSELS

FOR FLAT-BOTTOMED

hro~g.hout the chemical and petrochemical industries, gases, liquids, and solids are stored, accumulated, or processed in vessels of various shapes and sizes. Such a large number of storage vessels or tanks are used by these industries that the design, fabrication, and erection of these vessels have become a specialty of a number of companies. As a result of economic considerations, only a few companies in the process industries now design storage vessels having large volumetric capacity, and the usual practice is to contract for this equipment with companies specializing in this field. However, the design of this equipment involves basic principles which are fundamental to the design of other types of more specialized equipment. This and the followingchapter cover these fundamentals. Before the extensive use of welding as a means of fabrication, vessels fabricated of metal were assembled either by bolting or riveting (90). At the present time most fluids held at atmospheric or low pressure are contained in cylindrical tanks of welded construction. Because of the large quantities of petroleum and its products stored in this manner, the American Petroleum Institute has established specifications governing the design of welded, vertical, cylindrical tanks for oil storage above ground. This specification code, API Standard 12 C (2), is intended to provide the oil industry with design specifications for tanks of adequate safety and reasonable economy in a wide variety of capacities. It is recommended for use by the oil industry in all sections of the United States except those areas which are subject to local regulations that conflict with it. Although these specifications were devised primarily for the design of storage vessels for petroleum and its products, they are useful guides ~inthe design of a variety of vessels for
36

storage of other fluids. oil-storage tank. 3.1

Figure 3.1 shows a typical welded

MATERIAL SPECIFICATIONS

. The materials used in the construction of storage vessels are usually metals, alloys, clad-metals, or materials with linings which are suitable for containing the fluid. Whe~e no appreciable corrosion problem exists, the cheapest and most easily fabricated construction material is usually hotrolled mild (low-carbon) steel plate. The particular types of steel plate specified by API Standard 12 C are SA-7 (open-hearth or electric-furnace processes only), SA-283, Grade C for all thicknesses greater than U:i: in., or SA-283, Grade D for thicknesses less than l~:i: in. Copper-bearing steel, which is not specified by this code, has some advantages in resisting atmospheric corrosion. SA-7 is specified for structural steel shapes such as angles, channels, and I-beams, and ASTM-A-27 grade 60-30 fully annealed is specified for. steel castings. The physical properties and chemical composition of these plain carbon steels are listed in Table 3.1. Low-carbon steels are rather soft and ductile and are easily sheared, rolled, and formed into the various shapes used in fabricating vessels. These steels are also easily welded to give joints of uniform strength relatively free from localized stresses. The ultimate tensile strength is usually between 55,000 and 65,000 psi and the carbon content between 0.15% and 0.25%. However, both the ultimate strength and the carbon content cannot be specified since one is a function of the other. In Table 3.1 the ultimate strength is specified but not the carbon content. ooLow-alloy, high-strength steels are a specific class of lowcarbon steels which are made stronger by the addition of a

Material

Specifications

37

-t. Fig. 3.1.

Welded

oil-storage

tank, Humble Oil and Reflning Company,

Baytown, Texes,

Diameter,

120 ft. height, 40 ft; nominal cap,acity, 80,000 bbl.

(Courtesy of Bethlehem Steel Company.)

small amount of alloying elements. Such steels have yield points that are considerably higher than plain carbon steel of the same carbon content and also have higher ultimate tensile strengths. The chief disadvantage of some of the high-strength alloy steels of moderate alloy content is that they are more difficult to fabricate because they have low ductility. Some have an increasing tendency to air-harden, which may result in localized stresses in welded joints. However, these disadvantages may be avoided if the alloy content and the carbon content are both kept relatively low, as theyare in low-alloy, high-strength steels. Many steel manufacturers do not produce these steels to standard specifications as they do plain carbon steels but use a variety of formulas marketed under various trade names.

Most manufactures claim their low-alloy, high-strength steels to be readily "weldable." However, this is not a precisely defined characteristic, and it should be understood that such "weldability" refers to conventional metal-arc welding processes usually employed for plain carbon steel. Table 3.2 gives properties of low-alloy, high-strength steels recommended by API Standard 12 C, 1955. In selecting material for large-diameter vertical vessels, consideration should be given to the types of failure that have occurred in the past. In 1952 two unusual tank fail~"'-"'" ures occurred in England while the tanks were being hydrostatically tested with water at 40° F (91). Figure 3.2 shows the catastrophic nature of the failure. One tank was a crude-oil storage tank 140 ft X 54 ft with a floating roof,

»r=>

;d Table 3.1.
(1)

Properties of Plain Carbon Steels, Recommended by API Standard (Courtesy of American Petroleum Institute) (2) (3) Phosphorus, max, per cent Acid 0.06 Basic 0.04 Acid 0.06 Basic 0.04 (4) (5) (6) (7)

12 C

Plates

Steel Specifications SA-7 and SA-283, Grade D SA-283, Grade C

Tensile Sulfur, * max, Strength, Yield Point, min, psi per cent psi 33,000 0.05 60,000-72,000

(8) Elongation, min, per cent In 2 in. In 8 in. 22 ( 1,500,000 Tensile strength+

0.05

SA-7 Requirements ASTM-A-27, 0.05 0.06 Grade 60-30, fully annealed * From ladle analysis made by manufacturer; check analysis from finished material by purchaser may show 25 % more. t See exceptions listed in particular specification. Copper, when copper-steel plates or structural shapes are specified, minimum percentage. 0.20.

Structural Shapes Steel Castings

1,500,000 Tensile strength are same as given above for A-7 plates. 60,000 min 30,000 24

55,000-65,000

30,000

24

38

Design of Shells for Flat-bottomed

Cylindrical Vessels results of these tests are given in Fig. 2.9 (53), in the previous chapter. The conclusion made was that the -British steel had properties within the specification given and was as good as similar American steels of equivalent quality. The steel at the time of hydrostatic test was at a temperature (40° F) within the transition-temperature range (see Fig. 2.9). If the flaws had been smaller in size, or if a more ductile steel had been used, or if the hydrostatic test had been carried out at a higher temperature, the vessel prob, ably would not have failed. These failures demonstrate .the necessity of sound welding procedures and thorough inspection of welded joints. Even if cracks cannot be avoided, a high degree of protection against brittle fracture can be obtained with low-carbon steels if the steel has a minimum specification of l5 ft-l):>in the Charpy V-notch test at the lowest service temperature (47, 96). When specifying fine-grained steels or semikilled steels of low carbon with manganese, a 30 ft-lb Charpy V-notch value would give similar protection (96). The majority of the steels presently used for storage-tank construction do not meet this specification at temperatures below 50° F. It is recommended that for service temperatures below 50° F, steels meeting the above specification be used, such as ABS, class C or SA-201, Grade B (67) and that all weld joints in the shell be radiographed. 3.2 ESTIMATED COST OF TANKS

and the other was a gas-oil storage tank 150 ft X 48 ft with a cone roof. A fairly large number of similar tanks have failed in this country. These failures and those encountered in welded "Liberty" ships and T-2 tankers (92, 93) have stimulated studies on the reason for such failures (53, 94, 95,96). The failures of the two British tanks were traced to flaws in the welds in the shell. In the case of the crude-oil storage tank, the flaw initiating failure was incomplete penetration in a weld-probe replacement located across the first horizontal welded joint. The crack progressed vertically in both directions with brittle fracture up to the fifth course; the fracture of the upper four courses was by shear. It is significant that the rupture occurred entirely across the plates and not along any of the vertical welded joints in the nine courses. The gas-oil storage tank failed in a similar manner. In this case the flaw was in a partially repaired crack in the top 10 in. of a vertical joint in the top of the first course. As in the case of the crude-oil tank, this crack progressed vertically in both directions with brittle fracture in the first four courses and with shear fracture in the remaining upper courses. The fracture occurred chiefly across the plates as in the other failure except for the fifth course, where it traveled through a vertical welded joint. These two ruptures appear to be examples of failure by notch brittleness as a result of a crack existing in the plate at a location of high stress and having a length greater than the critical crack length for crack propagation. See Chapter 2 for discussion of notch brittleness and critical crack lengths. A complete investigation was made of samples of steel from the two tanks which failed. The samples were impact tested at various temperatures and the results were compared with corresponding data for other steels used for vessel construction (53, 91, 94, 95). The steel used in the construction of these tanks was identified as BS-13, a British steel similar to SA-283, Grade C. Some of the

Storage tanks require considerable capital investment. If a limited quantity of fluid is to be handled, a single tank may suffice, in which case the magnitude of the proportions is controlled by the volume of fluid to be stored. Where a large number of tanks are required, it is generally true that larger tanks give a lower cost per unit volume of storage than smaller tanks. This is indicated in Fig. 3.3, which shows that the total installed cost of a 1,000,000-gal coneroofed tank is approximately $32,000 whereas the corresponding cost of a 10,000-gal tank is about $3000, a hundred-

~ Table 3.2.

Principal Provisions of Specifications on Low-alloy, High-strength Steels, Recommended (Courtesy of American Petroleum Institute) (3) (1) (2) (4) (5) Chemical Requirements, max, per cent Manganese 1.25 1.30 Phosphorus (6) Tensile Strength, - psi

by API Standard

12 C

Carbon ASTM-A-242 Ladle Analysis Check Analysis Thicknesses: 7i 6 to % in., incl. Over % to IH in., incl. Over IH to 4 in., incl. SAE 950 Thicknesses: 0.0710 to 0.2299 in., incl. 0.2300 to H in., incl. Over H to 1 in., incl. Over 1 to 2 in., incl. 0.22 0.26

Sulfur 0.050 0.063

Silicon

(7) (8) (9) Tensile Requirements, min Yield Elongation, min, per cent Point, psi In 2 in. In 8 in.

70,000
/

50,000 46,000 42,000 24

18 19 19

67,000 63,000 . 0.050 0.90 70,000 70,000 67,000 65,000

0.20

1.25

0.150

50,000 50,000 47,000 45,000

22 22 { 1,500,000 22 Tensile strength 22

Optimum Tank Proportions

39

Fig. 3.2.

Failure of British storage

tank at Fawley, England

(91).

(Courtesy of Esse Research

and Engineering

Company

and Petroleum

Publishing Co.)

fold increase in volume for approximately 11 times the cost of the smaller tank. However, the large tanks are not always selected because of the greater flexibility permitted in storing a variety of fluids in a battery of smaller tanks. For such reasons no general rule can be made for the selection of an optimum number of tanks. The total cost of various types of large-sized storage tanks is given in Fig. 3.4 (97). The cost does not include the cost of the foundation. The cost of tanks fabricated from various materials (18), field-erected steel tanks (98), and small tanks under 1000 gal (99) has been reported in the literature. (See also 26, 27, and 28.) Figure 3.5 shows the weight of steel and the cost in dollars per ton for large-diameter tanks (97). 3.3 OPTIMUM TANK PROPORTIONS

105

«:~~
..!l!
't:I

(5

104
~'

~
,\).l

>'~

.5 u;
0
U

~\'V'"

~ ~ en
't:I

.5

~ I-

103

l4

""

Before a storage tank can be designed, the proportion of height to diameter must be established. The diameters of standard steel tanks for storage at atmospheric pressure usually range from 10 to 220 ft, and the heights vary from 6 to 64 ft. Typical tank dimensions are listed in Appendix E. No general rule can be given for the selection of the height-to-diameter ratio because this ratio is often a function of the processing requirements, available land area, and height limitations. Figure 3.6 shows a group of oil-storage tanks of various height-to-diameter ratios. The volume of a single tank, which may be one of a

Volumetric capacity in gallons 1'- Fig. 3.3. Estimated cost of small and medium-sized roofed storage tanks. Assumptions: Steel base cost Aat-bottomed cone-

Total cost = total weight X (6c + fabrication cost erection cost) Marshall and Stevens "Process Industries Average" cost index = 180

extras

freight

= 6c/lb

40
140 130 120 110

Design of Shells for Flat-bottomed

Cylindrical Vessels

I
Double-deck floating-roof tanks <,

./

./

<,
/ /"

_//

e =a "0
<0

100 90 80 70 60 50 40 30 20 10
/.

/'

Pcntoon-roof tanks"
/' ,,/'

'0
"0
<II

» y{

",,/

V
/-

V
V

./

"V ./

V./

V
/
/

V
V
./

lA V-tV

c:

:5
0

::> 0

<0 <II

",.

V /-

"".;(
~

P<

/'

.......... yne-roof

t"

I----

~ig. costs (97).

3.4. Estimated installed of large-diameter 'tanks (Courtesy of Petroleum Pub· McGraw-Hili

-= t> ..,
I-

/"

V
V

/ ..1

/'
/'

~/V ,/" ./_./'-VV

/'"
tanks

r- This

S 0

/~/

t'

YC%V ¥
V
10 20 30

~ v/~

V '-Open-top

curve is a· straight line b~tween starred points, and the formula -I-$ = 0.6175 x bbl + 10,000 (to determine total cost of . cone-roof tanks) applies to this portion of curve only.

Processing, lishing Co.)

+ Denotes
of a

I
40 50 60 70 80 90

frr

actual purchases. Remainder m mrufaTrer's (stimrs,_

I----

100 110 120 130 140 150 160 170 180 190 200 .

Capacity.in thousands of barrels

M(f

elf ....£1, I

:::a

0.611)

F.(;; 3
/

battery of tanks, may be determined by the process requirements and other considerations such as production flexibility and seasonal variations in storage requirements: The optimum proportion of the tank diameter, D, to height, H, varies between two limits. The lower limit for the optimum ratio of D/ H occurs when the shell, bottom, and roof costs per unit area are independent of D and H. This condition

exists with tanks of small volume, in which elastic stability and corrosion control the thicknesses. The upper limit for the optimum ratio of D/Hoccurs when the shell thicknesses is a function of both D and H, and the unit area costs of the bottom and roof are independent of D and H. This condition exists .with tanks of large volume. The optimum proportions of a tank are influenced by the

700 650 600

c:

550 500 450 400 350 300 250 200 150 100 50

;,~
Weight of tanks f~ with pontoon roof

/~

v:

s
.!!!

'GI Q.

:g
.s t> 0
'-

~ ~
~

/.~

..,
0

/cost

-= ....
~

<II

c:

-.
20

of cone-roof· tanks per ton

vY
.y
/.~

~ /~ .......

Weight of tanks with cone roof . 3.6 (bbls) W(tons) = +8

---woo

.;_ Fig. 3.5. Weight 'and cost per ton of large-diameter tanks (97). (Courtesy of Petroleum Processing, McGraw· Hill Publishing Co.)

<, ......_

d-

~V"

/'/'
o
~~
10

/~

30 40 50

---60

Weight, pontoon ... roof only .~

-\ ..Weight, cone roof only I I 70 80 90 100 110 120 130 140 150 160 170 180 190 200 Capacity in thousands of barrels

----- --;;:::.

Optimum Tank Proportions

41

fig. 3.6. diameter

Oil-storage

tanks of various height-to-diameter in the background.

ratios.

Three 96,000-bbl

tanks 120 ftin

diameter

by 48 ft high with three fire walls 172 ft in

by 24 ft high appear

(Courtesy of Hammond Iron Works.)

cost of the foundations and the cost of the land in the tank area that is chargeable to the tank as well as by the cost of the bottom, shell, and roof, if required. The value of the land chargeable to the tank may be expressed in terms of an annual cost per unit area. For purposes of tank proportioning, it is also convenient to express the cost of the foundation, bottom, shell, and roof, if used, in terms of cost per unit area as follows. then Let D
= =

annual cost of the installed foundation under the vessel, dollars per square foot of tankbottom area Cli = annual cost of the land in the tank area chargeable to the tank, dollars per square foot of tank-bottom area C = total annual cost of the vessel, dollars per year
C4

H
V or

diameter of the vessel, feet height of the vessel, feet . volume of the vessel, cubic feet

(3.2)

= -----'4

7rD2(H)

Substituting
C

for the areas Al and A2, we obtain: /

= 7rDHcI

+ -4

7rD2
(C2

+ Ca + + cs)
C4

(3.1) When Al = area of the shell, square feet = 7rDH A 2 = area of the vessel bottom or the projected area of the roof, square feet 7rD2
CI

Simplifying the equation by substituting of D, we obtain:

C

for H in terms

D +4

4 VCI

7rD2
(C2

+ Ca + + cs)
C4

(3.3)

= = =

C2

Ca

4 annual square annual square annual square

cost of fabricated shell, dollars per foot cost of fabricated bottom, dollars per foot cost of the fabricated roof, dollars per foot of projected area

To determine the optimum tank proportions by using Eq. 3.3, it is necessary to determine which of the cost terms are variables prior to differentiation. 3.30 Tanks Having Shell Thickness Independent of D and H. The stress in the shell is a function of both the diameter and the height of the tank, as indicated in Eqs. 3.18 and 3.19. For reasons of elastic stability, the minimum shell thickness is limited to M 6 in. for 45-ft tanks and smaller and to ~i in. for tanks of larger diameters. Therefore,

42

Design of Shells for Flat-bottomed

Cylindrical

Vessels proportioning may be assumed to be directly proportional to the shell thickness as follows: Cl =

tanks having shell thicknesses of ~iin. or less may be considered to have a shell thickness independent of Hand D. Substituting ~i-in. shell thickness into Eqs. 3.18 and 3.19, results in the following: 0.25 = 0.0001456 (H - I)D or D(H and D(H 1) = 1515 for lap-welded shells (3.5) Thus all tanks with butt-welded or lap-welded shells having a value of D(H - 1) equal to or less than 1720 or 1515 respectively fall into this category. Tanks of small volume falling into the category of Eq, 3.4 or Eq. 3.5 have shell, bottom, and roof costs per unit area which may be considered independent of D and H. Using Eq. 3.3 and differentiating the total cost, C, with respect to the diameter of the vessel, D, and considering the volume, V, to be known and the cost factors, Clo C2, C3, C4, and cs, as constant known factors for the volume considered, we obtain:
-

c6(H -

l)D

(3.7) (3.8)

or
C6

1)

1720 for butt-welded shells

(3.4) Substituting _ C - 4 V[c6(H D - I)D]

(H - I)D

Eq. 3.7 for ci in Eq. 3.3 gives: {-

+ 7rD2 4
_

C2

+ +' + )
C3 C4 Cs

Expanding and substituting for H by Eq, 3.1 gives: C

= 4 VC6 ( 7rD"2 - 4 VC6

-32c6 V2 D3 -0
7r

4V)

+ 7rD2 4

(C2

+ +. C4,+
C3

cs)

Differentiating and setting equal to zero gives: dC dD or

7rD

+-4

27rD

(C2

+++
C3 C4

cs)

dC dD

= -_-2-

4 VCl D

+ 7rD -2

(C2

+++
C3 C4

cs).

"""2 (C2 + Ca + C4 + C5) =

32c6 V2 7rD3
C6

Actually, the individual unit costs, Clo C2, C3, and C4, of the various tank components will vary somewhat with tank proportions and with tank volume and other factors such as design considerations. Many correlations for estimating tank costs have been presented in the literature. These correlations indicate that for purposes of estimating installed tank costs, the cost is primarily a function of the total volume and that total installed costs do not vary appreciably with the unit costs of component parts. Therefore, the consideration of unit-cost factors as constants is a reasonable approximation. For minimum cost the slope of the curve of cost versus the diameter of the vessel to contain the fixed volume, V, must be equal to zero (dCjdD = 0).· Therefore

Substituting
7r~

Eq. 3.1 for V and Eq. 3.8 for

C3 C4

(C2

+ + + C5) = + _+ +
C3 C4

eH
"_

gives:

~2~;D47r)

(7r~2Hy

or (C2 "2
7rD
cs)

32cI (H - I)D

7r

) (7r2D4fi2) --16

Since H - 1 for large tanks is approximately equal to H, it will be assumed that (H - 1) = H. Therefore D(C2 or D
=

+ + + C5) =
Cli C4

4c1H

- -- + -2

4 VCI D2

4H

C2

7rD

(C2

+++
Ca C4 Cl

Ca ~

C4

c)

(3.9)

cs)

=0

Solving for D, we obtain: D3 _ 8V (

7r
C2

+++
Ca C4

) cs

Substituting for the volume, V, we obtain:

or D

= 2H
C2

+++
C3 C4

Cl

(3.6)
Cs

3.3b Tanks Having Shell Thicknesses Dependent upon D and H. Tanks having heights and diameters such that the quantity D(H - 1) exceeds 1720 for butt-welded shells or 1515 for lap-welded shells have shell thicknesses which are dependent upon D and H. The cost of the shell per unit area, Clo is a function of D and H and for purposes of

3.3c Estimation of Cost Factors. In most designs the cost factors, CI, C2, and so on, cannot be accurately evaluated until the design of the tank is known, a requirement which necessitates successive approximations in determining the _optimum proportions. The costs of the tank components, shell, bottom, and roof, are a function of the plate thickness, grade of steel, cost of forniing, .post of field welding, and so forth, and include the cost of the accessories such as nozzles, manholes, pumps, ladders, platforms, and so forth which are attached to the various components. These factors are all interrelated, and in order to make an estimate it is usually more convenient to express them in terms of the cost of the fabricated component parts of the vessel per pound of fabricated material because information concerning cost per pound is more readily available. 3.3d Simplifled Cases of Optimum Proportions. To demonstrate the use of Eq, 3.6, three simplified cases will be presented. For the first case, consider a small open tank and disregard the costs for land and foundation. For small tanks

~hell ~eslgn the shell thickness is often the same as the bottom thickness. If the cost per square foot of shell, Ch is taken as equal to the cost per square foot of bottom, C2, then Cl = C2; and if Ca, C4, Co = 0, then, by Eq. 3.6, economical designs result diameter ratios.

01 large
in tanks

~torage Tanks having

43

low height-to-

t- 3;4 SHELL DESIGN OF SMALL AND MEDIUM-SIZED

VESSELS (PRODUCTION TANKS) Small and medium-sized vertical vessels with flat bottoms, called production vessels in contradistinction to storage vessels) are usually fabricated from steel plate of a single thickness. Their optimum proportions are similar to those discussed previously in case two (D = H). The design of such vessels is simple and has been standardized for the oil industry (100, 101) as described in Fig. 3.7 and Table 3.3. The shells of such vessels are usually fabricated of either~ 6-in. or }i-in. plate with plate widths preferably not less than 60 in. A mild steel meeting SA-7, SA-283, Grade C or D (open-hearth or electric steel only) is specified. The shell plates are usually either double-welded butt joints with complete penetration of weld metal or single-welded butt joints with backing strips with complete penetration of weld metal. The design of the bottom and roof (deck) is covered in Chapter 4. 3.5 SHELL DESIGN OF LARGE STORAGE TANKS

2H
Cl

+0 +0 +0

Cl

2H

(3.10)

However, the shell is generally more expensive than the bottom partly because of the additional cost of rolling the shell plates. For the second case, consider a small closed vessel having the same cost per unit area for shell, roof, and bottom and disregard the cost of foundations and land.

And if

C4

and

CD

= 0, by Eq. 3.6,
Cl

D = 2H -

+ +0
Cl

Cl

=H

(3.11)

For closed vessels also the cost of the shell per unit area is generally greater than the cost of the bottoms per unit area for the reasons previously stated. Furthermore, the roof costs are generally greater than the bottom costs per unit area because of the structural steel required for the roofs of all but small vessels. For the third case, consider a large closed tank in which the roof and shell costs are twice the cost of the bottom per unit area. And if and

The majority of tanks and vessels are cylindrical because a cylinder has great structural strength and is easy to fabricate. Several types of stresses may occur in a cylindrical shell. These may be recognized as: 1. Longitudinal stress resulting from pressure within the vessel; 2. Circumferential stress resulting from pressure within the vessel; 3. Residual weld stresses resulting from localized heating; 4. Stresses resulting from superimposed loads such as wind, snow, and ice, auxiliary equipment, and impact loads; 5. Stresses resulting from thermal differences; 6. Others, such as may be encountered in practice.

C4

CD

0, C2

D = 4H

+ 2C2 + 0

2C2

= -iH

(3.12)

Introducing the actual values of foundation and land costs, C4 and co, into the equations has the obvious effect of increasing the height-to-diameter ratio. It is apparent that in areas where land is cheap and the tanks can be easily supported on the soil without expensive foundations,

3.50 Stresses in Thin Shells Based on Membrane Theory. Simple equations may be derived to determine the minimum wall thickness of a thin-walled cylindrical vessel

Nominal Capacity, bbl 90 200 210 300 400 H-500 750 L-500 H-1000 1500

Approximate Working Capacity, bbl 72 166 200 266 366 479 746 407 923 1438

Table 3.3. Typical Dimensions for Production Tanks (100, 101) (Courtesy of American Petroleum Institute) Height of Height of ", Outside Walkway £ c: Overflow ", Diameter, Lugs, Height, Connection, ft in. ft in. ft in. ft 7-11 7-7 9-6 10 12- 0 7-7 10 9-6 10- 0 12-7 15 14-6 12- 0 12-7 15 14-6 12- 0 17-7 20 19-6 15- 6 16 15-6 13-7 15- 6 23-6 21-7 24 21- 6 5-7 7-6 8 21- 6 13-7 16 15-6 21- 6 21-7 24 23-6
?-

Location of Fill-Line Connection, in. 14 14 14 14 14 14 14 14 14

U
Ll

Size of Connections, in. 3 4 4 4 4 4 4 4 4 4

A

~-7

44

Design of Shells for Flat-bottomed

Cylindrical

Vessels

Drain-line connection

Tank

f
20"-diam dome Detail, thief-hatch outlet

Overflow-line connection

Vent-line connection

"
Plan

Overflow-line connection

fm~=~=~-::

Fill-line connection 22' B.C. 28 'Yu' holes , (in top flange ~ only)

Vent-line connection

20'-diam dome Fill-line

~-+=l
B
C D

r..
~13
I

91./ c/> bolt holes Walkway bracket lugs

13'.-1

40'

Shell plate

~
Detail, wllkway bracket lugs

Elevation
J

Fig. 3.7. Standard design for small ani:! medium·sized vessels (production tanks) as .ecommended (Courtesy of American Petroleum Institute.)

by API Standard

12 D.

(See Table 3.3 for dimensions.)

Shell Design of large Storage Tanks

45

p~

-+-p

Fig. 3.B. pressure.

Longitudinal

forces

acting

on thin cylinder

under

internal

with an internal pressure. Figure 3.8 shows a diagram representing a thin-walled cylindrical vessel in which a uniform stress, I, may be assumed to occur in the wall as a result of internal pressure. The nomenclature used in Figs. 3.8 and 3.9 is: 1 d l
p

= length, inches
=

inside diameter, inches

= thickness of shell, inches

internal pressure, pounds per square inch gage

A comparison of Eq, 3.13 with Eq. 3.14 indicates that for a specific allowable stress, fixed diameter, and given pressure, the thickness required to contain the pressure for the condition of Eq. 3.14 is twice that required for the condition of Eq. 3.13. Therefore. the thickness as determined by Eq. 3.14 is "controlling" and is the commonly used "thinwalled equation" referred to in the various codes for vessels. This equation makes no allowance for corrosion and does not recognize the fact that welded seams or joints may cause weakness. JOINT EFFICIENCIES AND CORROSIONALLOWANCE. In vessels for atmospheric storage the welded joints are 'seldom stress relieved or radiographed. The welded seam may not be as strong as the adjoining rolled-steel plate of the shell. It has been found-from experience that an allowance may be made for such weakness by introducing a "joint efficiency factor," E into Eqs: 3.13 and 3.14. This factor is always less than unity and is specified for a given type of welded construction in the various codes. The thickness of metal, c,. allowed for any anticipated corrosion is then added to the calculated required thickness, and the final thickness value rounded oft' to the nearest nominal plate size of equal or greater thickness. Equation 3.13 becomes: l and Eq. 3.14 becomes:

Longitudinal Stress. If one limits the analysis to pressure stresses only, the longitudinal force, P, resulting from an internal pressure, p, acting on a thin cylinder of thickness i, length l, and diameter d is: P
= =

=-+c
41E
pd

pd

(3.15)

=-+c
21E
where l d

(3.16)

force tending to rupture vessel longitudinally p7rd 4

2

= thickness of shell, inches

=

p = internal pressure. pounds per square inch

and a

= area of metal resisting longitudinal rupture = t7rd

E=
c
=

1=

inside diameter, inches allowable working stress, pounds per square inch joint efficiency, dimensionless (see Table 13.2) corrosion allowance, inches

Therefore

1=
or

stress

= _ = --_

P a

p7rd2/4 l7rd

=-

pd 4l

. induced stress, pounds per square inch (3.13)

=-

pd

MODIFICATION EQUATIONS. For storage vessels the OF maximum allowable working stress is considered approximately one third of the ultimate tensile strength of the steel; that is, a safety factor of 3 is employed, which is common for static structural loads on steel. The stresses

41

CIRCUMFERENTIALTRESS. If one refers to Fig. 3.9 and S considers the circumferential stresses caused by internal pressure only, the following analysis may be developed: P = force tending to rupture vessel circumferentially

= pdl
a
=

area of metal resisting force

= 2tl

1=
or

stress

=_

P a

=-

pdl 2tl

=-

pd 2t pd (3.U)
Fig. 3.9. pressure. Circumferential forces acting on thin cylinder under internal

21

""

~-,--~'---,
46

,.

Design of Shells for Flat-bottomed

Cylindrical

Vessels point. When one uses double-welded butt-joint. and lowalloy high-strength steel construction Eq. 3.18 becomes:'

are computed on the assumption that the tank is filled level with water at 60° F (density 62.37 lb per cubic ft) and that the tension in each ring is calculated at a point 12 in. above the center line of the lower horizontal joint of the horizontal row of welded plates being considered. The hydrostatic pressure in cylindrical storage tanks varies from a minimum at the top of the upper most course to a maximum at the bottom of the lowest course. In determining the plate thickness for a particular course, a design based upon the pressure at the bottom of the course results in overdesign for the rest of the plate and therefore does not represent maximum economy. A design based upon the pressure at the top of the course would result in underdesign, which would not be good engineering practice. However, some consideration should be given to the additional restraint offered by the plates adjoining a particular course. In the lowest course, the plates of the vessel bottom offer considerable restraint to the bottom shell course. This additional restraint of the bottom edge is effective for an appreciable distance or height from the bottom of the lowest course. In an intermediate course with a course of heavier plates below, the top of the heavier plates will be understressed; this will tend to cancel any overstressing of the bottom of the course in question. Therefore, a design based upon the pressure at a height of 1 ft from the bottom of the course may be considered conservative. The following equations may be derived if one assumes that the density of the fluid will not exceed that of water, which is used for the hydrostatic test of the tank.
p=P

5.096(D)(H

- 1)

+c

!Y.P.
For lap-welded construction,

(3.20)

t = 5.775(D)(H

- 1)

!Y.P.
where pounds per square inch

(3.21)

!Y.P. = minimum specified yield point of steel plate,

(H - 1) 144

(3.17)

where p H

= density of water at 60° F = 62.37 lb per cubic ft

=

height, in feet, from the bottom of the course under consideration to the top of the top angle or to the bottom of any overflow which limits the tank's filling height internal pressure, pounds per/~quare inch
, I'

For double-welded butt-joint construction, the above definition of p may be substituted into Eq. 3.16. When one uses an allowable design stress of 21,000 psi for SA-7 plate and a joint efficiency of 0.85 for doubled-welded buttjoint construction, the following substitution results: t = 62.37(H - 1)(12D) 2(21,000)(0.85)(144) therefore t where t H D

= 0.0001456(H - l)D

+c

(3.18)

= = = c=

shell thickness, inches height as defined in Eq, 3.17, feet inside diameter, feet corrosion allowance, inches

If double-full-fillet lap-joint construction is assumed, the corresponding joint efflciency, E, is 0.75. Then Eq. 3.18 becomes: t = 0.0001650(H - l)D c (3.19)

If low-alloy high-strength steel is used, the maximum allowable stress is taken as 60 % of the minimum yield

Steels of different composition should not be mixed in any course of a vessel, with the exception that it is permissi-" ble to use a different steel for providing reinforcing area for shell openings. In case two different steels are used in"any part of an area of reinforcement the stress corresponding to the weaker steel should be used to determine the thickness of shell metal to be reinforced. Equations 3.18 and 3.19 are given in API Standard 12C and apply only to the steel materials approved by the code. When other materials of construction, are used, the constants of these equations must be recalculated through the use of the proper value for the allowable stress. 3.5b Practical C;::onsiderations in Selecting Shell-plate Dimensions. It should be emphasized that Eqs. 3.15, 3.16, 3.18, 3.19 are useful only for predicting the thickness of metal required to withstand the internal pressure. Other factors such as structural stability, live loads, wind, -ice and snow loads, and fabrication procedures must be considered. Minimum-thickness specifications for tanks by A.PI Standard 12 C include some of these other considerations; they are listed in the tables of Appendix E. To summarize the Important points of these tables, consider first the thickness of the tank shell. It should not be less than ~::t: in. for tanks 50 to 120 ft in diameter, ~ 6 in. for tanks 120 to 200 ft, and % in. for tanks over 200 ft; or less than ;li 6 in. for tanks smaller than 50 ft in diameter. These minima are derived from practical considerations of stiffness, corrosion allowance, )Vind loads, and So on. Tanks having shell thicknesses greater than these minima may use decreasing thicknesses for the upper courses. The thickness for the upper courses necessary to contain the hydrostatic pressure may be determined by substituting the appropriate depth of liquid into Eq. 3.180r 3.19. However, the thickness should not be less than the minimum. It should be noted that these minima are expressed in fractions which correspond to mill plates of standard thickness. In general it is more economical to fabricate the smaller vessels from mill plate of standard thickness than to order plate rolled to a specified thickness. However, for the large vessels the shell must be thicker to withstand the hydrostatic pressure. With this greater shell thickness, rigidity and corrosion are no longer the controlling factors. Some reduction in cost may result from ordering these plates rolled to thickness, especially if the required plate thickness lies about midway between standard plate sizes. In specifying shell-plate widths a compromise must be made between the costs of the material and the costs of field welding including plate preparation such as edge r ,

Shell De~ign of Large Storage Tanks

47

Double-V double-welded butt joint

Double-U double-welded butt joint

(c)

(d) Double-bevel double-welded butt joint

(partial penetration)

(i)
Square-groove double-welded butt joint

UJ

I~

......_----~
Double-welded

Double-welded lap joint Horizontal Joints

full-fillet
lap joint

Vertical Joints by API Standard 12 C. (Courtesy of American Petroleum Institute.)

'" Fig. 3.10.

Typical shell joints recommended

working for butt welding. Plates 80 to 90 in. wide may be purchased at base cost without the inclusion of a price extra for width. Plate widths over 90 in. carry "width extra" charge which increases appreciably with increase in width. Therefore, it is advantageous to use the widest plate possible that does not involve an excessive extra cost. As a result, plates having a width of 96 in. are most extensively used. In specifying plate lengths, there are no price extras for lengths between 8 and 50 ft when mill plates are purchased. Therefore, the longest plates which can be readily handled and shipped are specified. Thus plate lengths of approximately 20 to 30 ft, are selected since longer plates are difficult to handle. The exact length of the plates is determined by dividing the circumference by the number of shell plates, with proper allowance made for the vertical weld joints. 3.Se Butt-welding versus Lap-welding'. The plates of the shell may be butt- or lap-welded depending upon the design and economic considerations. However, % in. is the maximum plate thickness for lap-welded horizontal joints and % in. is the maximum plate thickness for lapwelded vertical joints. Butt-welded joints may be used for shell plates for all thicknesses up to and including lH in. for plain carbon-steel-plates and up to and including 1%6 in. for low-alloy high-strength steel plates. The plates for butt welded joints must be squared. Squaring of the plates for lap-welded joints is not necessary. For this reason, plates for lap-welded joints are less expensive; however, erection by butt 'welding is somewhat faster. Because of the present high labor costs most tanks are now fabricated

by butt welding. Each course of the tank must be inside the course beneath it when the horizontal joints are lap welded. Vertical seams should not be in alignment for any of three consecutive courses. This is a precaution against localized conditions of stress at welds and aids in assuring the distribution of the stresses uniformly throughout the vessel. The requirement of a minimum distance of 2 ft between vertical joints in adjacent courses is an additional safety measure. In the butt welding of the shell, the joints should preferably be doubled welded with complete penetration and fusion. A single-welded butt joint with backing may be used instead, with the same joint efficiency. It is particularly important that the vertical butt joints have complete penetration and fusion because these joints are under the full tensile stress in the shell. The horizontal joints are not under this tensile stress, However, for structural strength against wind loads, and so on, and for prevention of failure by notch brittleness, all horizontal single-beveled joints should have complete penetration, as shown in Fig, 3.l0b. With squared plates (square-groove) and doublebeveled butt plates for horizontal joints, as shown in Fig. 3.l0a and c respectively, incomplete penetration may be used for the sake of economy. However, with partial penetration the thickness of the unwelded portion should not exceed one-third the thickness of the thinner plate, and the unwelded portion should be located at approximately the center of the thinner plate. If a horizontal butt joint is offset because of different plate thicknesses, the inside surfaces should be flush.

48

Design of Shells for Flat-bottomed

Cylindrical Vessels size is not over 3 in. nominal pipe size. For small pipe sizes, screwed fittings are usually preferred because they are cheaper than flange fittings. However, ..pipe with screwed fittings having a nominal size greater than 2 in. is rather difficult to fit because such heavy pipe is so rigid that it cannot be deflected easily to aid in aligning the threads during fitting. Therefore, for practical reasons, it is recommended that any nozzle having a nominal size greater than 2 in. have flange fittings. It is usually desirable to locate nozzles for filling and discharging .near the bottom of the tank to obtain the benefit of gravity in discharging and to avoid pulling a partial vacuum on fluids which are volatile. However, water and sludge may separate and collect on the bottom. To avoid pumping this sludge out of the discharge line, the discharge nozzles are usually placed on the shell a short distance above the bottom. Another nozzle with a sump is placed at the bottom to remove material accumulating below the discharge nozzle and to completely drain the vessel. Typical nozzles of both the screwed-fitting and flange-fitting type are shown in Fig. 3.14, and standard dimensions for these nozzles are given in Items 1 and 2 of Appendix F. . SHELL MANHOLES. Manholes are necessary in closed vessels to permit inspection, cleaning, repairs, and so on. These manholes may be located on the shell or on the roof or at both locations. Manholes located on the shell have the advantage that it is somewhat easier to use a shell opening to clean or repair a vessel. Shell manholes have the disadvantage that they usually cannot be opened unless the vessel is empty and therefore are not used as often for inspection as roof manholes. Items 3,4, and 5 of Appendix

(a)

(bl
joints for tonk shells. (Courtesy of Hommond

Fig. 3.11. Iron Works.)

Circumferential

Lap joints should have an overlap of at least 5 X t inches, and in no case should the overlap be less than 1 in. Vertical lap joints should have continuous full-fillet welds both inside and out. On horizontal lap joints, as shown in Fig. 3.11, the fillet should have a size not less than one third the thickness of the thinner plate, and in no case should it be less than :J.i6 in. Figure 3.11 also shows a horizontal butt-welded shell of a storage vessel. Note that the heavier courses at the base are V butt-welded, whereas the upper courses are plain butt-welded. . 3.5d Cold Forming of Shell Plates. Plain carbon-steel shell plates having a thickness of :J.i6 to % in. for tanks having a diameter of 40 ft or more or low-alloy high-strength steel plates for tanks having a diameter of 50 ft or more can be deflected on erection and therefore need not be cold formed by rolling to the radius of curvature of the shell. If the diameter is 60 ft or more for plain carbon steel or 100 ft or more for low-alloy high-strength steel, shell plates having a thickness of % to H in. may be deflected on erection without cold forming. Plain carbon-steel plates Of thicknesses from H to % in. and diameters over 120, ft need not be cold formed. However, all plain carbon~s"ieel plates having a thickness of % in. and over and all low'·alloy high-strength steel plates having a thickness of H in. or over must be cold formed to the.shell radius regardless of the shell diameter. Figure 3.12 shows the field welding of horizontal butt-welded seams of the shell of a large tank. 3.5e Shell Parts. In addition to the shell plates a variety of other shell parts and accessories must be considered in the shell design. Figure 3.13 shows typical tank accessories including shell nozzles, manholes, ladders, and so on. SHELL NOZZLES. Pipe lines which bring the fluid to and from the tank are attached to short pipe connections welded into the tank shell. These connections are called "nozzles" and may be fabricated of screwed pipe fittings if the pipe

Fig. 3.12. Field welding of shell circumferential Hammond Iron Works.)

butt joint.

(Courtesy of

Shell Design of large F gives typical dimensions for shell manholes and manholecover plates designed as shown in Fig. 3.15. REINFORCEMENT F SHELLOPENINGS. All openings such O as nozzles and manholes, made in the shell in which the opening is over 2 in. in diameter should be reinforced. The reinforcement prevents local overstressing of the shell around the opening. The minimum cross-sectional area of the reinforcemerit should not be less than the product of the vertical diameter of the hole cut in the tank shell times the shell plate thickness. The cross-sectional area of the reinforcement is measured parallel to the axis of the shell across the center of the opening.

Storage

Tanks

49

sidered as available for reinforcement out to a distance of four times the neck-wall thickness, measured from the outside of the shell. The metal in the neck lying within the shell-plate thickness may also be included. If the neck of the fitting extends both inward and outward as shown in the center and right of Fig. 3.15, credit may be taken for the metal of the neck over a distance of eight neck-wall thicknesses plus the shell thickness. In the case where two or more openings are located close together and the toes of the fillet welds fixing the reinforcing plates for these openings come within twice the shell thick-

Conservation vent

I I I I I

Float~:

Spiral stairway Indicator

Winch

CD Shell

manhole

Fig. 3.13.

Usual accessories and RHings on standard

cone-roof tanks.

(Courtesy of Hammond Iron Work s,) 15. Connection for foam chamber 16. Drain 17. Flame arrester 18. Antifreeze valve 1-8 Extra units 5. (a) Inside ladder

Included as standard 1. One 20" shell manhole 2. One 20" roof manhole 3. One 6" gauge hatch 4. Roof nozzle for vent (12 or 13) 5. ladder (small tanks only) 6. Spiral stairway 7. Two shell nozzles S. Flange for water draw-off 9. Sump 10. Swing line unit complete

Included as extra

11. Water draw-off 12. Conservation vent (volatile products) 13. Free vent {nonvolatile products} 14. (a) Target-type Root gauge Root gauge

14.

(b) Ground-reading-type

Reinforcement metal may include anyone of the four metal parts listed below or any combination of them: 1. The metal in the attachment flange of an attached fitting, 2. The metal of a reinforcing plate, 3. The metal of any excess shell-plate thickness beyond that required from a calculation of the minimum plate thickness, 4. The metal in the neck of a fitting. This can be considered as part of the reinforcement area. If the fitting extends only outward, t he metal in the neck may be con-

ness of each other, a single reinforcing plate should be used. This plate should be proportioned for the largest opening in the group. If the reinforcing plates for one or more small openings are of such a size that they lie entire}" vithin the area of the reinforcement plate for the largest opening, the small openings can be included in the normally designed reinforcement plate for the largest opening without increasing the size of this reinforcing plate. If, however, any opening intersects the vertical center line of another opening, the width of the reinforcement (along the vertical center line of either opening) should not be less than the sum of the widths of the two plates that would normally be used.

50

Design of Shells for Flat-bottomed

Cylindrical

Vessels

Permissible alternative square cut

A circular reinforcing plate may' be substituted for the plate shown, for the 3- to 10-in. size nozzles inclusive, provided the { diameter of the circular plate is made equal to l-l:.' .

Bend reinforci ng plate (_to radius of tank shell Bolt holes shall straddle the flange center lines Reinforcing plate

See_~_efl~~~~Fm==~~ note •

Detail B

Victaulic~ groove or threads

Single flange (a)

Double flange (a)

Special flange (a)

TOiher d~tails as previous

LLhI-JL
(b) (c) Detail B (c) Shell noules recommended by API Standard 12 C (see items 1 and 2 of Appendix F for typical dimensions.) (Caurtesy of American Petroleum

t_noZZles~

m
T

Coupling

~ Fig. 3.14. Institute.)

Finally, if the normal reinforcing plates for the smaller openings do not fall within the area limits of the reinforcement for the largest opening, the group reinforcing plate should include within its outer limits the normal reinforcing plates for all openings in the group. 3.5" Reinforcement of Top Course of Shell for Lorge Open Tanks. Open vessels of large diameter may not have

the necessary inherent rigidity to withstand wind loads without deforming and excessively straining the structure. Two methods of stiffening are available: shell plates may be made thicker, or suitable stiffening girders may be added to- the structure. The use of thicker shell plates usually is more costly than the use of stiffening girders. Wind girders or stiffening rings for open tanks are located at or

Shell Design of large

Storage

Tanks

51

Gasket: 20· manhole-25,,· 00 x 20· 10 x 16· thick 24· manhole-29"· 00 x 24· 10 x 16· thick Long-fiber asbestos sheet unless otherwise specified

Shape manhole flange to suit curvature of tank

1~·

rr
t

(min)

30· T
Increase if necessary for clearance

Gasketed face shall be machine finished to provide a minimum gasket-bearing width of J( in. for minimum thickness at bolt circle. see Appendix F.

Alternative designs of manholes

Fig. 3.15. Institute.) Shell manholes recommended by API Standard 12 C (see items 3, 4, and 5 of Appendix F for typical dimensions). (Courtesy of American Petroleum

near the top of the vessel on the top course of shell plate. The stiffening ring is placed preferably on the outside of the shell rather than the inside. The required section modulus as specified by API Standard 12 C for the stiffening ring may be computed by Eq. 3.22: (3.22) where z D

= height of shell including any "freeboard"

vided above the maximum filling height, feet

pro-

= nominal diameter of the tank, feet

section modulus, inches.'

The calculation of the available section modulus of the stiffening ring may include a portion of the tank shell which is considered to be effective for a distance of 16 plate thicknesses from the ring, as indicated in Fig. 3.16. When curb angles are attached to the top edge of the shell ring by butt welding, this reinforcing distance should be reduced by the width of the vertical leg of the angle. Table 3..!

52

Design of Shells for Flat-bottomed

Cylindrical Vessels appropriate dimension for the minimum width of the girder web, b, is 12 in., which corresponds to a section modulus of 28.1 in. 3 The rest of the sectional dimensions, including wind-girder plate thickness of }4 in., are fixed by detail E of Fig. 3.16. A stiffening ring such as this is fabricated by bending plate steel into the appropriate shape. Such a wide-webbed ring cannot be easily rolled to the required

t
Detail D

16t

Detail B

rj
=t
J
t

-f.

~2ZZ'Z1~~~

_j_

61

t 16t
_j_ r2!i'_'

f Table 3.4. Section Moduli of Various Stiffening-ring

Sections on Tank Shells, Recommended by API Standard 12 C (Courtesy of American Petroleum Institute) Member Size, in. Shell Thickness, in. h~ }:4 Section Moduli 0.42 0.52 0.91 l. 72 2.04 2.48 3.35 4.41 5.82 l. 78 2.12 3.73 4.45 5.95 6.60 7.61 10.56 11.78 13.67 16.24 18.89 17.70 20.63 28.92
I

Detail C

1
_l
16t

6'

_l
~ fig. 3.16.

16t

Typical reinforcement by API Standard

for top course of shells for open vessels 12 C. (Courtesy of American Petroleum

recommended Institute.)

lists the section moduli for the rings shown in Fig. 3.16 for two shell thicknesses. Stiffening rings may be made either of structural sections or formed-plate sections or combinations of the two. The minimum size of the angle specified by Standard 12 C either alone or with a built-up section is 272' X 272' X X and the minimum plate thickness is X in. If the stiffening rings are located more than 2 ft below the top of the shell, a minimum of a 272' X 272 X ~6toP curb angle is required for ~6-in. shells. A minimum of a 3 X 3 X X angle is required for }:4 -in. shells. Other members of equivalent section modulus may be substituted. Drain holes should be provided in rings that may trap liquid. Stiffening rings are sometimes used as walkways, in which case they should provide at least 24-in. of clear walking space and should be located preferably 3 ft, 6 in. below the top curb angle. Any such angle- having a horizontal web exceeding 16 times the web thickness requires suitable vertical support. ExAMPLE DESIGN 3.1, WIND GIRDER FOR OPEN VESSEL. A wind girder is required for an open vessel 80 ft, 0 in. in inside diameter and 40 ft high, having a top-course plate thickness of }i in. When one uses Eq. 3.22, the minimum section modulus of the wind girder is:
2

Top angle: detail A, Fig. 3.16 2}2 X2}2 X H 0.41 272 X 2}2 X ~6 0.51 3 X3 X% 0.89 Curb angle: detail B, Fig. 3.16 272 X 2}2 X}:4 l.61 2}2 X2H X ~6 l.89 3 X3 X}:4 2 .32 3 X3 X% 2.78 4 X4 X}:4 3.64 4 X4 X% 4.17 One angle: detail C, Fig. 3.16 ~}2 X 2}2 X H l.68 272 X 2}2 X M6 1; 98 4 X3 X}:4 3 . 50 4 X3 X ~6 4.14 5 X3 X ~6 5.53 5 X3HXP6 6.13 5 X 3}2 X % 7.02 6 X4 X% 9.02 Two angles: detail D, Fig. 3.16 4 X3 X~6 11 .27 4 X3 X% 13.06 5 X3 X~6 . 15.48 5 X3 X% 18.17 5 X 3H X ~6 16.95 5 X 3}2 X % 19.99 6 X4 X;B 27.74
{

= 0.0001D2H
= 0.0001(80)2(40)

= 25.6 in.3

When one uses the type of construction shown in detail

E of Fig. 3.16, referring t~ Table 3.4, he finds that the

Formed plate: detail E, Fig. 3.16 b, in. 10 '22.3 12 28.1 14 34.3 16 40.8 18 47.7 20 54.9 22 62.4 24 70.3 26 78.5 28 87.0 30 95.9 32 105.1 34 114.7 36 124.5 38 134.7 40 145.3

Shell Design of large diameter because of its stiffness. Therefore, it is more convenient to weld straight sections of formed plate, making a polygonal stiffening member. The inside edge will be flame cut and made smooth to form an arc having a radius equal to that of the shell outside diameter. Usually two or more sections are welded end-to-end in the shop to minimize the field welding required. Figure 3.17 shows a preliminary sketch of. a win irder subassembly, using 20 equal sections with every two secti welded end-to-end for subassembly in the shop. For determination of web dimension, z, cos 9 °

Storage

Tanks

53

40 ft, 0-1-in. 40 f 01.

t,

+ 12 in.
+X
-

continuous double-welded butt joint or a continuous doublewelded lap joint. SHELLS WITH SELF-SUPPORTINGROOFS. A self-supporting roof is one which is supported only on its periphery, without added structural support. Such roofs cause a compressive stress in the roof plates, which is transferred to the shell as hoop tension. A stiffening angle should be added to the top shell course at the junction of the roof and shell to absorb the stress as a tensile load. The forces acting -on the ring are shown in Fig. 3.19. he following nomenclature will be used in the equations explaining Fig. 3.19: D

"4 Ill.

= 0.98769

492.25 Ill. . 0.98769

480 •25 Ill. .

= = (J = P=
a

= 18.135 in. = 1 ft, 6i in.

For determination Chord -2 Chord of inside chord length of subassembly,

Tl

1• 40 f t, 0TIn. (si 180) SIll

T2 = Ta =
F

= 2(480.25 in.)(0.30902) = 296.814 in. = 24 ft, 8* in.

of outside chord length of subassembly

cross-sectional area of stiffening ring, square inches nominal tank diameter, feet. angle of the cone element with the horizontal, degrees roof load, pounds per square foot (live load of 25 lb per sq ft plus dead logd) compressive force perIinear inch along an element of the cone, pounds per linear inch tensile force per linear inch in a circumferential direction, pounds per linear inch horizontal component of TJ, pounds per linear inch circumferential tensile force acting in stiffening ring, pounds I
7rD2

For determination Chord 2 Chord

W = total load on roof, pounds = f

= (40 ft, oi in.)

+ (1 ft,

4 tensile stress, pounds per square inch

6i in.) sin 18°

= 2(41 ft, 6i in.) (0.30902) = 2(498.375)(0.30902)

=

308.016 in. 25 ft, 8 in. of outside chord length of section,

=

For determination Chord

2(41 ft, 6-i in.) sin 9°

= 2(498.375)(0.15643) = 155.9216 in.

12 ft, 11 H in.

Figure 3.18 shows typical detail designs for a wind girder for an open tank 80 ft in diameter and 40 ft high as described in example design 3.l. [j.5g Reinforcement of Top Course of' Shell for Large Closed Tanks. SHELLS W.ITH ROOFS HAVINGCOLUMNSUPPORT. If tanks are closed with a roof, the roof provides additional structural rigidity to the upper course of shell plates. As a result, smaller stiffening rings are used for closed vessels. For tankroofs with column supports having diameters of 35 ft or less, 27~ x 27~ x 7'4 in. is the minimum-size stiffening angle. For tanks having diameters of 35 to 60 ft; the size of the angle should be increased to 27~ x 27~ x ~i in. 6 For tanks over 60 ft in diameter the size of the angle should be 3 x 3 x % in. The outside leg of the angle may be extended either inside or outside of the vessel. The angle is attached to the upper edge of the shell plate by either a

Fig. 3.17. design 3.1.

Preliminary

sketch of wind-girder

subassembly

for example

54

Design of Shells for Flat-bottomed

Cylindrical Vessels
Note: , Cut back both ends, as shown, together. See section E-E
in shop after welding sections
r

----

_------B.;J~

Arc on edge of PL. 25'-1%"

Section E-E Wind-girder assembly, 10 required

12/-11'~.'" ~ 12' 11~." Bend

upl

l'~,J
1

I--l'~_

-+--

- ~t - ~l
12'-6,."

t-'~.,</>

. t
Orai!, holes

~l'-l$

@approx, 2'-0" centers

Section A-A Section B-B

Detail of wind- girder section, 20 required Fig. 3.18, Details far wind girder far example design 3.1.

In reference to Fig. 3.19a, the roof load, W, results in compressive force, TIt in the roof plates as follows: TI
= -----

also F

af

PD
48 sin 8 component,

T _ af _ PD cot 8
a-

sin 8(-II'D)12

6D -

48

Compressive force, TI, has a horizontal

Ts. or

To solve for a,

r, _-T

cos 8 = P D cot 8 48
\.

PD2 cot 8 a = --,--8f

In reference to Fig. 3.19b,

f = 18,000 psi, allowable

Therefore a=-----8(18,000) tan 8
a=------

2F = 12DTs
or F 6D

PD2

Ts

PD2

=-

144,000 tan 8 (for

if: live load

~F

= 25 lb per sq ft : = 11 lb per sq ft = 36 lb per sq ft

36D2
144,000 tan 8

t in.

plates)

dead load P

By substituting, a=
-F
(a)

D
4000 tan 8 equal to sin 8;

For small angles, tan 8 is approximately therefore a=---4000 sin 8

(b) Loads on conical roofs.

D2

Fig. 3.19.

Example Design 3.2, Complete Shell Design for a Closed Vessel The API Standard 12 C recommends the use of Eq. 3.23 for determining the required reinforcing area. a
= ----

55

Since corrosion allowance il

= 0,
=

0.57 in.

D2

3000 sin 8

(3.23)

In applying Eq. 3.23 credit may be taken for the crosssectional area of the shell and roof plates within a distance of 16 times their thickness from the stiffening angle. In other words, the sum of these areas must be equal to or greater than (D2/3000 sin 8) for cone-roof construction. For dome or umbrella-roof construction a similar equation may be derived. a where D a

By the use of 10 plates and with an allowance of %2 in. for vertical weld joint, the center length of each plate is calculated from the circumference. L

= 1I"d - weld length = 3.1416(1200.57) - 10(%2)

12n 12(10) (or 31 ft, 5 in.)

= 3770.136 = 31.4178 ft
120

DR
1500

(3.24)

R = radius of curvature of dome, feet

diameter of tank, feet

= reinforcing'area,

square inches

3.6 EXAMPLE DESIGN 3.2, COMPLETE SHELL DESIGN FOR A CLOSED VESSEL The design calculations and drawing for the steel shell only for a 55,000-bbl oil-storage tank having a cone roof are required. The cone roof is to be supported by internal columns, girders, and rafters. 3.6a Proportioning. It is estimated that the ratio of annual cost of the shell per unit area, Ch is two times the annual cost of the bottom per unit area, C2, and that the annual cost of the roof, C3, is 1.8 times the bottom annual cost per unit area. The annual cost of the land area, C4, and preparation, C5, together is estimated at 0.40 of the cost of the tank bottom, C2' Corrosion allowance will be negligible. When we substitute the above information into Eq. 3.9,

Standard mill plates of 96-in. width will be specified and butt welding of the shell plates will be used for fabrication. Shearing of the shell plates is required to square the plates for butt welding. Therefore, the final height of the shell plates will he slightly less than 96 in. The thickness of the shell course above the bottom course can also he determined by Eq. 3.18. The proper height, however, for this calculation will be, (40 ft - 8 ft) or 32 ft t2

= 0.0001456(32 - 1)100 = 0.452 in.

+c

Specify that t2 L

= 0.46 in.
10 X 12

= 11"(1200.46)- 10(0.15625) = 31 ft, 4ft in.

the thickness ta
=

Accordingly, 24 ft) will be:

of the third

course (H

0.0001456(24 - 1)100

+c

= 0.335 in.

+0

When we substitute H

the same information

into Eq. 3.1,

= 4(55,000 X 42/7.48)
(2.5H)211"

.r~---'----1r-----,L-J---,--_.L____j
~ o
I

Ha = 62,910 cu ft or

H = 39.8 ft
Since D

Ll-------r------i±=~~
~
Fig. 3.20. Elevation view of shell for example A. Cable sheave B. Winch C. 4" steam nozzles D. 10/1 shell nozzle E. 1 ~2/1 extrc-hecvy F. 4/1 wcter-drew-eff couplings nozzle swing joint design 3.2.

= 2.5H,
D

= 2.5 X 39.8 = 99.5 ft

The tank dimensions will be 100 ft inside diameter by 40 ft high. Appendix E, item 3 indicates that such a tank will have a volume of 55,950 bbl. 3.6b Design of Shell Courses. The thickness of the bottom shell course can be determined readily by Eq. 3.18 (for butt-welded assembly). tl
=

0.0001456(-10 - 1)(100)

+c

= 0.568 in.

+c

G. Shell nozzle for double H. 24/1 shell manhole

56

Design of Shells for Flct-bcttorne d Cylindrical

r
,

Vessels

50'-03a"-

7
~

,+

_[P
314'-3~" Outside circumference

50' -Ola'
Radius to

Top

3x3x~'

J 0!
I

1
i'
t
I

t PL
~~

11~'
# 5 shell PL
------..._

314'-3~' Outside circumference

50' -Ola' Radius to

1
100 ,
0'>

t Pl.
Pl. #3-0.34'

314'-4' Outside circumference

;:_

~-: ----.II~
# 4 shell Pl.

lila'

, Radius to

5D'~'l<.' r---'---

L31'-5~" --....
Fig. 3.21. Shell details for example design 3.2.

t Pl.
PL. #2-0.46"

~
;:_

314' -5' Outside circumference

50'-O~2 Radius to

¢. Pl.
PL #1-0.57'

t
t

~
;:_

314'-5~' Outside circumference

SO'-O~" Radius to ¢. PL. 50'-0' Inside radius

Section through shell

~?~
~ #1 shell 8i/l of Mo/eria/s
Lenqfh ff. in.

~v-.;o2
pC
Section through vertical Joints

r.~ ,

SJ,1~;1Mark
/0 /0 #/ #2

Descr/pt/o» PI. 95ft4-.¥0.ST" PI. 95~~'Xo.46' PI. 93 G;8'X 0.34PI. 951;8'X PI95~8-X

Order
/0 10 PI 96"%0.57PI. 96·Xo..~·

10 /0 /0 /0

#3 114
#5

J4"

y.,.-

lA5

L 3·.r3"%~8"

3/ 4% 3/ 4~- /0 3/ 4'~; /0 31 4'~t;- /0 3/ 4'r,i /0

3/

PI

94" X 0..34-

PI 96- X

fI.,."

57': 3/ 5yi31 5~s"

31 31 3/ 3/ 50/850/8' 8

Lenqfh If. m.

PI 94" Xy..,."

L 3"%3" X~-

Specify that t3 L

0.34 in. 10 X 12 '

= 71"(1200.34)- 10(0.15625) = 31 ft 415 in.

Tlf

Likewise, the thickness of the fourth course (H

will he:
t4

15 ft)

= =
=

0.0001456(15 - 1)100 0.204 in.

+c
=

71"(1200.25)- 10(0.125) 10 X 120

31 ft/4H

in.

Specify t4 and t6 = 0.25 in. since the thickness as determined from the appropriate relationship results in a thick-

ness of less than ~ in. This means simply that hydrostatic pressure stress considerations are not controlling and the structural stability of the thin shell is the prime consderation. Thus the minimum shell thickness of ~ in. for shells of this size as set by API 12. C for tanks 50 ft in diameter and larger will he specified (see Appendix E, item 4). The required thicknesses for the various shell courses could have been determined from Appendix E, item 4. It will be seen that the calculated and tabulated shell-course thickness agree. Only the shell plates of the bottom course need to he cold formed to the shell diameter. Frequently, however, thianer , shell plates are rolled to facilitate erection. Accordingly, the shell' specifications will call for rolling of the bottom three courses.

Example Design 3.2, Complete Shell Design for a Closed Vessel

3.6c Design of Top Angle. The minimum-size top angle for a tank larger than 60 ft, 0 in. in diameter with a roof supported on columns is 3 in. X 3 in. X % in. and will, therefore, be used. Specifications will call for butt welding of this angle to the top course. By using 10 lengths of top angle, the length of each angle section is calculated as follows:

57

L L

11"(1200.375) - 10(0.15625) 31 ft,

4-»

10 X 12
in. view of tank shell, and

Figure 3.20 shows the elevation Fig. 3.21 shows the shell details.

PROBLEMS

1. A cone-roof tank having a filled capacity of 100,000 bbl is to be designed. Determine the optimum proportion of D / H from the following cost considerations. The fabricated shell, roof (including plates, rafters, girders and columns), and bottom are estimated to cost 18, 20, and 14 cents per lb respectively. Foundation costs are estimated to be $4000. Fixed annual charges including amortization, interest, and so on are estimated to be 40 % per year based on initial installed cost. The annual charge for the land allocated to the tank ~~ea is $500. 2. A wind girder is required for an open vessel 120 ft inside diameter and 48 ft high. The top course of the shell is fabricated from ~ -in. plate. If the wind girder consists of 30 identical sections corresponding to detail E of Fig. 3.16, determine the minimum section modulus and the girder dimensions. 3. Determine the required cross-sectional area of the stiffening ring for a self-supporting conical roof 30 ft 0 in. in diameter having an angle 8 of 15° with the horizontal. .4. The re~uired shell plate ~icknesse~ for the vessel de:>cribed in problem 2 may be determined from Item 2 of Appendix-E. Using these plate thicknesses and 18 courses, prepare a plot of circumferential stress versus height for the conditions of the hydrostatic test in which the vessel is filled with water.

----7---

--:-

C H A -P'T

ER

DESIGN

OF BOTTOMS AND
/

ROOFS

FOR FLAT-BOTTOMED

CYLINDRICAL VESSELS

he bottoms and roofs of vertical cylindrical storage vessels are usually fabricated of steel plates having thicknesses less than those used in the shell. This is possible for the bottom because it is normally supported by a prepared base of sand or gravel resting on the soil. The roof load is usually limited to wind and snow load with a proper allowance made for any anticipated additional loads. 4.1 BOTTOM DESIGN

11

The shape and design of the bottom for a storage vessel will depend upon such considerations as: the foundation used to support the vessel, the method for removal of the stored material, the degree of sedimentation of suspended

JL ~
Intermittent weld Shell-to-bottom joints Single-welded full-fillet lap joint Bottom-plate joints

nSide Bottom plate Added plate

"Groove

~ptlonal

"CTaCkweld Single-welded butt joint with backing strip

1" Fig.

4.1. Typical bottom joints recommended (Courtesy of American Petroleum Institute.'

by API Standard

12 C.

solids, corrosion of the bottom, and the size of the storage vessel. If the considerations mentioned dictate the use of a flat bottom and the safe bearing capacity of the soil is at least 3000 lb per sq ft the bottom is usually placed on a sand or gravel pad directly on the soil. . If the tank bottom is directly supported by the ground, flexure of the bottom is prevented, and the bottom plates are under a simple compression load. Theoretically a lightgage sheet metal, 16-gage or less would be sufficient for such a bottom. However, to provide greater ease in welding and to allow additional metal for corrosion, plates having a thickness of at least X in. should be used. For many years ~6-in. plate was the most common plate thickness used for tank bottoms. Bottom plates of 72 in. or more in width are preferred and plates 96 in. wide are usually specified. Plates of ~~-in. thickness are usually lap welded with a lap margin of at least IX in. for all joints. The bottom plates should extend beyond the shell-plate bottom weld at least 1 in. No more than three plate laps should be located within 12 in. of each other or of the shell. Typical welded joints for shell-to-bottom and bottom plates are shown in Fig. 4.1. Figure 4.2, a and b, gives alternate methods of shaping the sketch plates under the shell ring. The sketch plates should be formed and welded in such a manner that a smooth bearing surface for the shell plates is produced. In regard to selecting the plates for the bottom, the largest-sized plates available that can be conveniently handled and that have no cost extra for size are usually the most economical. Plates 96 in. wide x 20 or 30 ft long are often used. If the bottom plates are laid symmetrically in relation to the center lines of the bottom plan, the number of different shaped plates will be reduced to a minimum.
58

Example Design 4.1, Bottom for a Tank 150 Ft, 0 In. in Diameter plate

59

(a)
;.. Fig. 4.2. Methods of shaping sketch plates under the shell ring recommended by API Standard

(b)

12 C.

(Courtesy of American Petroleum Institute.)

This is an advantage because the plates can then be scribed and cut in groups of four, whereas if the bottom plates are symmetrical in relation to only one center line, only two plates can be scribed and cut at one time. A bottom asymmetrical along both center lines makes a large number of plates of different sizes necessary. The simplest symmetrical layout is to arrange the corners of four plates to intersect at the center of the tank bottom. However, this layout should not be used with lap-welded construction because four plates will lap at the joint. Also, this layout is sometimes wasteful in that with some groups of dimensions considerable scrap may result from the plates at the perimeter. In such a case the bottom plates may be rearranged with one plate centered on the bottom. In this layout the center row is single, but all other rows have mates. The center row will have two perimeter plates of the same size, but there will be four identical perimeter plates for each succeeding row from center. The sizes of plates and the location of cuts on perimeter

plates can be readily calculated by use of Eq. 4.1 and by reference to Fig. 4.3. A2=B(D-B) (4.1) (4.1a)

D2 A2 = - _ C2
4

Means must be provided for the removal of liquid from the bottom of the vessel. A sump, shown in Fig. 4.4, may be used with a sump-pump discharge. Flat-bottomed tanks using gravity or pump discharge may discharge by means of a draw-off elbow, as illustrated in Fig. 4.5. Dimensions for a draw-off elbow are given in Table 4.1. 4.2 EXAMPLE DESIGN 4.1, BOTTOM FOR A TANK 150 FT, 0 IN. IN DIAMETER

Table 4.1. Dimensions of Welded Draw-off Elbow, Recommended by API Standard 12 C, All Dimensions in Inches-See Fig. 4.5 (Courtesy of American Petroleum Institute) Distance Distance from Distance from Diameter Diameter Center of Nomifrom Center of of Hole of Rein- Elbow to nal Center of Outlet to in Tank forcing Face of Pipe Elbow to Bottom, Bottom, Plate, Outlet ,Size* Shell, B C o; DR Flange, E 2 7~'2 6 3~B 6~~ 1~ 3 87'2 7 4~~ 7% 13 4 9}-'2 71YJ:6 5~~ 9% 14 6 11 9% 7% ·12% 16 8 13 12% 9% 16H 18 * Extra-strong pipe, API Standard 5 L

A bottom is required for a tank 150 ft, 0 in. in inside diameter. The minimum allowable (2) plate thickness is ~~in.; however, because of the large tank size a plate thickness of ~ 6 in. will be specified to provide additional protection against loss by corrosion. The bottom course of shell plates for this vessel is 1H6 in. thick, and a ~ 6 in. fillet weld will be used between the shell and the bottom plates.

} Fig. 4.3.

Relationship of bottom plate dimensions. -

- -_----

60

Design of Bottoms and Roofs for Flat-bottomed

.:
!

Cylindrical

Vessels

~
2' -0' to 3' -0'

28%"
r-

25~" opening in bottom plate

~
co

"""
0/,.

/"'V
"

'll6

'--- f---

+I-I

vL~"
I
I

Tankshell~

2' x Y. • ring I.

«:Tank bottom f

:--Sutt

weld

Fig. 4.4. Draw-off sUIT!P recommended by API Standard 12 C. (Courtesy of American. Petroleum Institute.)

1
%

[/'V
v"'"
~24'

Y,6

rU'"

k=1 J
Alternative for bottom corner
I

24%"

The bottom plates must extend a minimum of 1 in. beyond the shell weld, or in this case, the radius must be increased a minimum of (1 1~16 ~6 in.) or 2% in. A radius of 75 ft, 3 in. will be used for the bottom. The central bottom plates will be 96 in. wide by 31 ft, 8H in. in length. The bottom plates will be lap welded, and the joints will be staggered so that no more than three plates are lapped within 12 in. of each other or of the shell. Figure 4.6 shows a layout for such a bottom. It should be noted that the layout is symmetrical with respect to one axis, and thus only half of the bottom is shown. It is also evident that, except for the necessary staggering of the plates, it is symmetrical with respect to the other axis. To demonstrate the use of Eq. 4.1 in this design, dimensions A and C will be calculated for points x and y on the layout. At point x dimension C is equal to one half the plate length of 31 ft, 8~~ in., and Dis 150 ft. 6 in.

By Eq. 4.1a, A2
=-

D2
4

C2

(150 ft, 6 in.)2 4

(15 ft,

lOi

in.) 2

= 5662.562 - 251.024 = 5411.538

therefore A

73 ft, 6t in.

To obtain the width of plate 8-2 at point z, SH-in. plate widths less (10 - 1) laps of Hi in. must be subtracted from dimension A. The plate width is 96 in. 8-2 plate width at x = (73 ft, 6t in.) - Si(S ft, 0 in.) . - 9(1! in.) = (73 ft, 6t in.) - (67 ft, Of in.) = 6 ft, 6 in.

31 ft,

S! in.

15 ft, 10"8'm,

1•

DR----l>i

Tank shell

l __ ~
.
-

r
c

J(

4~r~~~f:~~~:::f'~t~.~~Extra-strong pipe

Fig.- 4.5. Welded draw-off recommended by API Standard

elbow 12 C.

(See Table 4.1 for dimensions.) (Courtesy of American Petroleum' Institute.)

Example
21'-S,,' 19'-9~'

Design 4.1, Bottom for a Tank 150 Ft, 0 In. in Diameter

61

II'-IO~' '-~ 4

,~
S-S 1%'
I

'",

Ie-" ';S·12~ I :=S'II~ 1~ Ie";' ~ :: :: ~ ." I~ 5W24

. ~LI,,
I

!t
S·12
I

: S·II

63'!2'"

~6'-4Y.."

f• 55'-31jj"
t
41'-5%'

34'-1" 5W32

f *.
t

L
5' 8'-31!i.%

1
.1

SW24 10'-10~'

SW24 12'-10%" • 5W32 16'-10%'

~
I

1
I

S·1O I

S·9
/

5W24

36'-1' 5W32

~.

&4'-S~'1 5-1 -2,,· II'-Ili.· 5-6 16'-9,,·

,1

46'-5_%'

~
.

14'-10~'

J "t
5W32

1
48'-5~'

J'_ 10'-2,,·
S-6

I~

2'-8'..{,-

39'!615"
31' 1,,"

f 23'-81jj"
t
t

-I'~:t!:._
s-4

~:S'-91Y,," S-5 3'-1'"

;;;_;;..,;'

5W32

5W32

5W32

5W32

5W32

5W32
-+

5W32

5W32

5W32

5W32

5W32

~
5-5 _\

1+-""
SW24

55'-3~," I SW24 I I

~"
I I I

39'-6,,· 31'-1,," ~ ~ I

__,;:

23'-S%' 15'-10%" 1'-11"", ~ ....._.;: I

L8'-3'.1,' 5-3
10'-5%· 5-2 1'-9,,·

SWl6 63'-2%"

5W24

15,110%.

.,....6'-6·
61'-2" ...=:..;.

~ ~ ~
I

-r- ~ r
1'-11'"

59'-3""

SI'-Os'l ~I

43'-5,,·, ---' I

~35'-1"

,,:;...-.'

I 21'-8,,·'

19'-9~'1 I~

5-2
'~

~
S-I 5W32 5W32 5W32
5W32

5-1 8'-2'"

-5W32

5W32

5W32

5W32

5W32

5W32

5W32

5W32

1 5W32
l

5W32

5W32

5W32

5W32
"'-

I
Half plan of bottom
AllPLS.-12.75" - 1"· laps 2940 Ii n It of If,' weld

Radius .. 75'-3"

Radius

= 15'-0'

10 inside of shell

Radius = 15'-3' 10 .. outside edge of bottom

Section through corner

Section through lap

S:;~~~i Mark
47
10
444 44-

Bill of Material Oercription F).eng:; Wi No.

Plr. - 96"' 12.75- 31 8:4 PII.-96"X 12.7511 23 9Yz Pll.- 96·X 12.75* 15 IO*' Pll. - SK X 12.75* PI1.-SK X 12.75# P~I."SK XIZ.75# PI1.-SK X 12.75# PIs.-SK X 12.75' PIs.- 5K X 12.75# PI$.- 5K X 12.75* P/I.-SK XI2.75# PIs:-5K X 12.75 # PII." SK X 12.75# Pis.- SK X 12.75PII."SK of 1275# PI/. - SK X 12.75# PIs.- SK X 12.75# PII." SK X 12.75#

SWJ2 5W245WI6 5-1 5-2

ling'" Order Ft In. 47 PI$.-96"X 12.75- 31 m

5-3
S-4-

4 4
2

Z
2 2
4 44 4

5-5 5-6 5-7

S-7-A

10 PII.-96"XI2.7S# 4 PII.-96·XI2.75# 2 PIs.-96"XI2.75# 2 PII.-96'X 12.75* 2 PIs:- 96"X 12.75" 2 PIr.-96·XI2.75" 2 PIs.-96"XI2.75" ~ PI's: -96"112.75" 2 PIs.-96"XI2.75#

23 15 16 14 19 13 22 28 15

9~ 10~ 2Y,. 6Jz

Similarly, at point y dimension C is equal to 1H plate lengths plus 2 plate widths less 3 plate laps of 1H in., or C
= = =

1.5(31 ft, 8i in.)

+ 2(8 ft, 0 in.)

- 3(li in.)

Olz
9%

(63 ft, 6i in.) - 3t in. 63 ft, 2* in.

31"
3~

By Eq. 4.1a, A2
=-

5~

D2
4

5-8
S-8-A S-9 S-IO S-I/ S-12 S-I]

2 PI$.-96·XI2.7511 2 PIs.-96·XI2.75#
2 2 2

21 /If,

C2

13 51;¥ PIs.-96 "X12.75* 16 8 Plr96"XI2.75# 20 IIlz PIs.-96"XI2.75# 23 s-!4 PIs.- 96" X12.75/1 12

5662.562 - (63 ft, 2* in.)2

= 5662.562 - 3996.610

=
A
=

1665.952 40 ft,

9H in.

Fig. 4.6.

Typical bottom layout for a 150-ft-diometer

tonk.

62

Design of Bottoms and Roofs for Flat-bottomed

Cylindrical

Vessels

en og e:

~ 8.

.t::.~ .... 0

en e:og cue: enca.

..J-

8:0

Roof Design Therefore, for the layout shown in Fig. 4.6, the dimension for S-8 at point y is: (40 ft, 9H- in.) or 4.3 ROOF DESIGN (34 ft, 7 in.)

63

+ Ii- in.

= diameter, inches t = cone shell thickness, inches () =iangle between cone element and horizontal

6 ft, 4-ft in.

The most common shape for a tank roof is a cone although dome or umbrella roofs are also used. In addition to these shape classifications, tank roofs may be classified into two types, self-supporting and nonself-supporting. Regardless of shape or method of support, tank roofs are designed to carry a minimum live load of 25 lb per sq ft , in addition to the dead load. This live load is an average figure which allows for combined wind and snow loads and for the weight plant personnel who may travel across a roof to inspect the vessel or to reach a manhole and so on. Figure 4.7 shows the maximum snow loads to be anticipated in various parts of the United States (137). 4.3a Self-supporting Conical Roofs. A self-supporting roof is one which is supported only on its periphery without the aid of additional support from columns. Tank diameters for self-supporting roofs generally do not exceed 60 ft and usually are less than 40 ft. Any greater spans require such heavy rafters that it is simpler to use one or more supporting columns and thereby reduce the span. Such roofs usually consist of roof plates supported on rafters. Small and medium-sized flat bottomed cylindrical tanks having capacities of 400 and 3000 bbl or under respectively are extensively used in the petroleum industry (100, 101). Figure 3.7 of the previous chapter shows proportions for such tanks, and Table 3.3 gives typical dimensions. The roofs of these tanks are known as "decks" and are fabricated of mild steel having specifications meeting ASTM-A-7, ASTM-A-283 grade C or D (open-hearth or electric-furnace steel only). The deck plates have the same thickness as the shell plates, and a slope of 1 in. in 12 in. is used for the cone. If M 6-in. plates are specified, the deck must be reinforced with structural support if it is 15~~ ft or larger in diameter but does not require additional reinforcing if it is smaller in diameter. If ).~-in. plate is used, no support is needed for 1572-ft diameter tanks, but support is required for all larger-diameter tanks. The deck may be attached to the shell by one of the following methods. The deck may be flanged and welded by: a double-welded butt joint with complete penetration, a single-welded butt joint with backing strip, or a full-fillet double-welded lap joint. If the deck is not flanged, it should have full-fillet welded joints both inside and outside. ' For larger tanks with cone roofs the equation for stress in a cone under either an internal or external pressure can be derived as shown in Chapter 6 (see Eq, 6.139). The maximum stress will exist at the greatest diameter of the cone and will be:

The stress as calculated by Eq. 4.2 will be controlling only in the case of thick cones used with pressure vessels of limited diameter. In the case of large-diameter conical roofs such as those used for storage tanks, the controlling factor is elastic instability. The theoretical critical compressive stress that causes failure of a curved plate by wrinkling is given by Eq, 2.24.
icritical =

0.6E-

i r

(2.24)

where E

= modulus of elasticity

i= r=
icritical =

of material, pounds per square inch thickness, inches radius of curvature, inches (see Fig. 4.8) theoretical critical stress at which failure by wrinkling occurs, pounds per square inch

The safe compressive stress that can be carried without wrinkling was investigated by Wilson and Newmark (43) in a series of experimental tests. As a result of these tests and others (44). it was found that the safe compressive stress that can be imposed on a steel cylindrical shell without failure by wrinkling is one twelfth of the theoretical critical stress and can be expressed for r as follows:
iaUowable=

1.5(106) ~
r

=<

!yield 3

point

(2.25)

Equation

2.25 can be modified for use with a conical roof

D
r r

= diameter of tank, feet = radius of curvature of cone

at periphery, inches of cone element with horizontal

=~ sin 8 8 = angle

is
w-here p
=

= 2t

pd sin

(4.2)

internal or external pressure pounds per square inch gage

'V" Fig.

4.8.

Radius of curvature of conical roofs.

Design of Bottoms

nd Roofs for Flat-bottomed

Cylindrical

Yes

I(

Fig. 4.9. structural (Courtesy America.)

Field photograph showing support far a tank roof. of Aluminum Compony of

by referring to Fig. 4.8 and substituting for r as follows:

f allow. = 1.5(106) t sin 8 6D

or sin 8 =
fallow.D

(4.3) sin 8
=

___!!__

1000t

VP76
=

(4.5) 7.65 lb per

250,OOOt

(4.4)

If live load = 25 lb per sq ft and dead load sq ft (for % 6-in. roof plates) P
=

32.65 lb per sq ft

It is very important to recognize that the allowable compressive stress, fallow., is not the conventional allowable stress for the material but is the safe stress that can be applied without danger of failure by wrinkling. The compressive stress induced by live and dead loads on the roof must not exceed the allowable compressive stress, fallow .• Equation 4.2 can be used to calculate the compressive stress induced by the roof loads, or Eq. 4.2 can be substituted into Eq, 4.4 as follows: sin 8

If one substitutes for Pin Eq. 4.5, min sin 8

.E:

1000t

,132.65/6 =
/

4301

.J!_

(4.6)

(P/144)(12D) D (2t sin 8) 250,OOOt

Equation 4.6 is the equation specified by API Standard 12 C for self-supporting conical roofs. It should be emphasized that the derivation of the constant is based- upon selected roof loads and % 6-in. roof-plate thicknesses and should be modified for other conditions. 4.3b Conical Roofs with Structural Support. When the design calls for a conical roof with structural support, a

Roof Design slope, or pitch, of the roof of a %-in. rise per 12 in. is recommended. The roof plates may be ridged in order to decrease the number of rafters required. Roof plates should not be attached to the rafters. Roof plates of lap design should have a minimum lap of 1 in. when tack welded; moreover, if a continuous full-fillet weld is used on all seams, it is necessary to weld only the top side of the roof. For steel construction a minimum thickness of ~16 in. is recommended for the roof plates. Figure 4.9 is a photograph showing assembly of structural support for a tank roof before installation of roof plates. Storage tanks and other large vessels with conical roofs· usually are designed with no attempt made to prevent the roof plates from flexing. In such a design the rafters are spaced sufficiently close to each other to prevent overstressing of the outer fiber of the roof plates as a result of flexure. The roof plates are assumed to act as flat continuous beams with a uniform roof load. The rafters and girders are assumed to act as uniformly loaded beams with free ends. Roof design involves the consideration of bending and shear in the roof plates, rafters, and girders. Column action in the rafters of self-supporting roofs and in the columns of roofs having supports must be also considered. A brief discussion of these relationships follows. UNIFORMLY LOADEDBEAMSWITHFREE ENDS. Referring to Fig. 4.10, consider any point, e, between supports R1 and R2 in the beam uniformly loaded with w pounds per linear inch. The forces acting on the beam to the left of point x produce a bending moment, M, which can be evaluated by a summation of the moments at z. For a uniformly loaded beam freely supported at the ends, R1 = R2; therefore R1
=-

65

By substituting

= 1/2 into Eq. 4.7, we obtain:

Mmax =

wP

(for a uniformly loaded beam freely supported) (4.8)

To determine the deflection of the beam Eq. 2.41 is substituted into Eq. 4.7, and the equation: is integrated. I d2 . M = El _]'_ = w x _ ~ dx2 2 therefore El dy dx
2

= wlx
4

2 _

wx 6

+ C1

(4.9)

To evaluate the constant of integration C1 apply the boundary condition dy/dx = 0 where x = 1/2. Therefore

C1
By substituting ohtainr Ely Since y

=,--

wl3 24 again, we

in Eq. 4.7 and integrating wlx3 12 wx4 24 w[3x 24

=-

--

--

+ C2

(4.10)

= 0 where z = 0,

By substituting
y

and solving Eq. 4.10 for y, we obtain:

wi
2

;1 (W;;3 __
4

~~4

w;:x)

The force R, produces a clockwise or positive moment equal to R1x, and the uniform load to the left of x results in a force, wx, which produces a counterclockwise or negative moment equal to -wx(x/2), or M
=

The maximum deflection occurs at the center of the span where x = 1/2; therefore
y = --

wlx _ wx 2 2

5wl 384El

• . maxImum de flection

(4.11)

(4.7)

To obtain the location of the maximum bending moment, let

dM = 0
dx therefore

.!i (WIX _ WX2)

dx
2 2
or

UNIFORMLY LOADED CONTINUOUS BEAM. A uniformly loaded continuous beam having a large number of equally spaced supports reacts the same as a simple uniformly loaded beam with fixed ends. Consider the uniformly loaded beam shown in Fig. 4.11. The beam is assumed to be a section of a continuous beam with a large number of equally spaced supports I distances apart. A bending moment Mo exists over the supports .

wi
2 therefore

- - wx

=0

=-

I 2

;4 Fig. 4.10.

Uniformly loaded

beam with free ends.

...

---..,.,..--66 Design of Bottoms and Roofs for Flat-bottomed Cylindrical Vessels maximum bending moment occurs over the supports; as defined by Eq. 4.14. . The maximum deflection of the beam by inspection is observed to occur at x = lj2 and may be obtained by substituting Eq. 4.14 into Eq. 4.13, integrating, and evaluating the new constant of integration, C2•
Fig. 4.11. Uniformly loaded continuous beam.

Ely

=-

wl2x2
-

24

+ wlx - -24" + C -12 wx

0 at x = 0

Taking the summation of moments at a distance x from the support R1. we obtain:

but y therefore Substituting

y

M = Mo
But

+Rx
1

-I-

wx (~)

C2 = 0 and solving for y where x

l/2, we obtain:
(4.17)

wi Rl =2 for a beam with clamped ends, and

Eld2y

= - --wi'

384EI

. . = maXImum de flection

dx2 Substituting

=M

(2.14)

for M and R1. we obtain: M

EI ~y

dx2

= M0

+ wlx
2
2

_ wx

(4.12)

COLUMN ACTION. Slender' structural members under axial compression tend to deflect. This deflection .results in a bending stress superimposed on the compressive stress. Referring to Fig. 2.4 of Chapter 2, we find that the axial force, P, causes a deflection of y in the column of length, l, and cross-sectional area, a. The bending moment, M (equal to the force, P, times the lever arm, y) induces a bending stress equal to Me/I, which must be added to the compressive stress, P /a, or

Integrating,

we obtain:
dx EI dy

= Mox

+ wlx - wx + C - 2 -6
to evaluate

By definition 1= ar2 therefore (where r

=

Applying boundary C1. we obtain:

dy

conditions

the constant

radius of gyration)

=0

dx Tberefore
EI dy = dx

when x

=0

1= I!_
a

(1

+ ye) r2

(4.18)

The column may be compared to a uniformly loaded beam freely supported, in which by Eqs. 4.12, 4.8, and 4.11,

Moz

+ wlx

wx

(4.13)

1=

Me

T;

= 8;

wl2

5wl4 384EI

Also,
dy = 0

dx therefore

when x = 1

Solving for the product yc for the uniformly loaded beam, we obtain:
,y c= 5wl 384EI
4

(811) wl2

= 512j

48E

=C

12
1

(4.19)

Mo =

(maximum moment for uniformly loaded beam with fixed ends) (4.14)

where C1 = constant. , For a column the product yc is assumed to vary as 12, as in the case of a beam:

Substituting

into Eq. 4.12, we obtain:

EId2y dx2
=-

ye = C

212

M= x

wl2

wlx -+---

wx2

12

(4.15)

Substituting the quantity solving for P/a, we obtain:

C212 for yc in Eq. 4.1~ and

Tke bending moment l/2 is:

at the center of the span where

wl2 (M)l/2 =-

1 ;;-= 1 + C2(l2/r2)

(4.20)

24

(4.16)

Comparison

of Eq. 4.14 with Eq. 4.16 indicates that the

where C2 is a constant depending upon the material, the method of loading, and the method of support. No method is known for calculating from theory the value of

Roof Design

67

the constant C2• The constant is usually determined by experiment. Rankine experimentally evaluated the constants for round- and square-ended columns and found C2 to be 1/18,000 and 1/36,000, respectively (29). Since any slight displacement of a fixed-end or square-end column either laterally or axially will result in an unknown eccentric loading, columns are usually designed as round-ended members. For values of l/l' between 60 and 200, the American Institute of Steel Construction (102) recommends the following equation for steel columns: P
a

For a rectangular

beam,
Z =-

bt2

where b

= width of beam, inches t = thickness of beam, inches

= 1.0
Z

Thus, for the case where b

in., (4.24)

=-

t2
6

+ (P/18,OOOr

18,000

2)

(4.21)

Substituting obtain:

Eq. 4.24 and Eq. 4.21 into Eq. 2.10, we

For columns having values of l/r between 0 and 60, a column formula is not used, but a maximum value of IS,OOO for compressive stress, P/ a, for steel columns is specified. For values of l/l' greater than 200, Euler's column formula is used. (See Table 2.1, Chapter 2.) Self-supporting roofs have rafters under combined compressive load and bending load. Such rafters may be considered to act as beams under column loading. The constants J and C2 of Eq. 4.20 have' been specified as given in Eq. 4.22 by the American Institute of Steel Construction (102) for steel beams under column action.

J
or

pl2
= 2t2

1=

V2i1P

(4.25)

For an allowable stress, f, of 18,000 psi (the maximum specified by API Standard 12 C for steel roofs) a roof-plate thickness of ~ 6 in., and a roof load of 32.65 lb per sq ft, that is, p = 0.227 psi, a substitution into Eq, 4.2S gives:

I=~

16'\j

/c2)(18000) 0.227

= 74.6 in.

P
a

+ (l2/2000b

20,000

2)

(4.22)

where I b

= unsupported
=

length, inches width of-compression flange, inches

The application of Eq. 4.22 is limited to conditions in which l exceeds ISb but is less than 40b. If the member is shorter than ISb, the rafter may be designed as a beam. Lateral stiffeners may be used to maintain I within the upper limit of 40b. The value of 20,000 specified in the numerator of Eq, 4.22 is permitted because the bending stress is maximum only at the outer fiber, and therefore the maximum combined compressive stress exists only at the outermost fiber on the top side of the rafter. The average compressive stress across the member will be less than 20,000 psi. " RAFTER SPACING. Consider a circumferential strip of roof plate 1 in. wide located at the outer periphery of the conical roof, and disregard the support offered by the shell. This strip is considered to be essentially a straight, flat, continuous, uniformly loaded beam. 'The controlling bending moment is equal to w12/12 and occurs over the supporting rafters. By Eq. 4.14,
Mmax

It is apparent from the above calculations for a :J1s-in. roof plate that rafter spacing in large-diameter cone-roof tanks should not exceed 6 ft unless heavier roof plates are used. API Standard 12 C specifies a maximum rafter spacing of 211' feet or 7S% in. on the outer perimeter of a ring of rafters and a maximum of SH ft on the inner perimeter (2). The minimum number of rafters adjacent to the shell is determined by dividing the shell circumference by the maximum rafter spacing. The actual number of rafters to be specified should be a multiple of the number of sides of the polygon of girders supporting the other' end of the rafters to provide a symmetrical layout; this is a further restriction. The minimum number of rafters to be used between two adjacent inner girders should be based on the perimeter of the outer polygon of girders. The length of one side of a polygon having sides of equal length is: L where L .N

= 2R sin 360
2N

(4.26)

= polygon-side length, feet

= =

12 =

-w12

-p(I)12 12

=~

_p12

R
(4.23)

number of sides of polygon radius of circle circumscribing of rafters,

polygon, feet will then be

where I

= =

length of beam (strip) between rafters, inches unit load, pounds per square inch = w when width = 1.0 in. the stress resulting from flexure by Eq. 2.10,

The minimum number equal to (12NL/I) or,

n, required

24NR n =--sml

. 360 2N

(4.27)

Introducing we obtain:

where n = minimum number of rafters l = maximum rafter spacing, inches The actual number of rafters to be specified should be a

1\1

(2.10)

68

Design of Bottoms and Roofs for Flat-bottomed

Cylindrical

Vessels

multiple of the number of sides of the polygon, N, to maintain a symmetrical layout, SELECTION OF RAFTERS AND GIRDERS. Rafters are designed as uniformly loaded beams with free ends. Each rafter is considered to support the roof plates and roof load over an area extending on either side of the rafter and bounded by the center line to the adjacent rafter. From Eq. 4.8 it is seen that the maximum bending moment for such a beam is equal to wl2/8 and occurs at the center of the beam. The maximum fiber stress is directly proportional to the square of the length of the beam, 12. Therefore, to avoid use of excessively heavy rafters the length of the rafter is usually limited to from 20 to 24 ft or less. This may be demonstrated by considering 20 ft, 0 in. rafters spaced 6 ft, 0 in. apart at the outer side and 4 ft, o in. apart at the inner side with a roof-design load of 0.25 psi. The mairmum bending moment is, by Eq. 4.8:
/

wl2 M=-

(0.25)(5 X 12)(20 X 12)2

8
=

8
108,000 in-lh M

Rewriting Eq. 2.10 as, z for

=-

1=

18,000 psi (assumed allowable value) we obtain:

Z=

108,000 in.-Ib _ 6 o· 3 . In. 18,000 lb per in.2

From item 1 of Appendix G a 7 x 2H in. channel weighing 9.8 lb per ft in which z = 6.0 in.3 may be selected to fulfill the requirements. For vessels of large diameter in which the rafter span is reduced by the use of girders, column supports must be used for each ring of girders and also at the center of the tank. Usually five or more straight girders are joined end-to-end to form a polygonal support for the ends of the rafters. The girders are designed in the same manner as the rafters. The girder load is considered to be a uniform load equal to the roof load plus the weight of therafters. The roof area contributing this girder load is the length of the girder times the distance on either side halfway to the next rafter support. The rafters form a series of concentrated loads on the girders, but for practical considerations the load may be treated as uniform whenever four or more rafters are supported on one girder. Self-supporting roofs have rafters under a combined compressive load and bending load. In such designs, if the unsupported length, I, exceeds 15b where b equals the width of the compression flange, the stress in pounds per square inch should not exceed the value calculated from Eq. 4.22. The laterally unsupported length of beams and girders should not exceed 40 times b (the width of the compression flange). The above restrictions, limiting beams to lengths with an lib ratio not greater than 40 and to stress not greater than permitted by the formula for lib ratios greater than 15, do not, apply to rafters which are in contact with the steel roof plating. It is assumed that under full-load conditions, friction between the roof sheets and the rafters will provide

adequate lateral support to the compression flanges of the rafters. ' SELECTION OF COLuc\INS. The design of beams and rafters for roofs is based on a safety factor of approximately 3, that is, one third the ultimate strength. For rolled-steel structural shapes of 55,000 psi ultimate tensile strength, an allowable tensile stress of 18,000 psi is recommended. The same value may be used for the maximum allowable compressive stress of rolled-steel sections if lateral deflection is prevented. In the case of columns, the lateral deflection should be considered, and the maximum allowable compressive stress should not exceed 15,000 psi. This stress can be computed by Eq, 4.21. The allowable compressive stress so calculated'is based on the gross section of the column (including area of weld) if the column consists of two or more sections welded . together. For main compression members the llr ratio, should not exceed 180, and the ratio for bracing and second-: ary compression members should be limited to 200 ... Supporting columns for roofs may be either of standard structural shapes or of pipe, depending upon preference in design. In the installation of columns, clip angles should be used on the tank bottoms to prevent any possible-lateral movement of the column bases. 4.3c Dome and Umbrella Roofs. A dome'roof is one formed to a spherical surface. At the beginning of the century, tanks with dome roofs were used for a great variety of services. Today they are seldom used for atmospheric storage as the more simple cone-roof tank is cheaper. Dome roofs are still. used extensively for cylindrical flatbottom storage vessels designed for low-pressure service. . The umbrella roof is formed so that any horizontal section through the roof is a regular polygon with as many sides as there are plates. Umbrella roofs are a compromise between cone roofs and dome roofs. Umbrella roofs have approximately the strength of dome roofs but are easier to install because the roof plates are curved in only one . direction. The equation for stress in a spherical thin-walled vessel can be developed in a manner similar to that used in developing Eq. 3.13 with the following result:

t=-

pd

41

+c

(4.28)

By comparing Eq. 4.28 with Eq. 3.14 it is apparent that for the same radius of curvature and the same shell thickness the spherical shape is twice as strong as the cylindrical shape. Thus, for equal strength, the radius of curvature of a sphere should be twice that of the cylinder. Therefore, it is customary to make the radius of curvature of a dome roof equal to twice the radius of the shell. The API Standard 12 C recommends this proportion and permits a 20 % variation in either direction, or R=D or.
Rmin

(4.29) 0.8D· 1.2D (4.30) (4:31)

= =

or
Rmax

Example D~sign 4.2, Design of Roof and Structural Supports for a 122-Ft Tank where R = radius of curvature of dome, feet D = radius of tank shell, feet ELASTICSTABILITY UMBRELLA OF ROOFS. The specifications of an umbrella roof are also determined by the roof loads and the elastic stability of the roof under load.: By following the same method of derivation as that used for Eq. 4.6 for self-supported conical roofs, Eq. 4.32 can be derived for umbrella roofs as follows, starting with Eq, 2.25: but R

·69

=L

12 12

!!. = ~65~ = 304

t or R The constant
=

304t

(4.34) in

Eq. 4.32 by
therefore 1.5 X 106t2

pr2 (as for conical roofs) r2

in Eq. 4:28 differs from the constant in Eq. 3.14 by a factor of 2.0; moreover, t and r occur in the P critical equation to the second power. 4.4 EXAMPLE DESIGN 4.2, DESIGN OF ROOF AND STRUCTURALSUPPORTS FOR A 122 FT., 0 IN. DIAMETER STORAGE TANK 48 FT, 0 IN. HIGH

vi ~This is to be expected since the constant

in Eq, 4.34 differs from the constant

p == 32.65 lb per sq ft = .227 psia .

1.5 X 10 t2 0.227 V6.6 X 106 r
6

=~

2.58 X 1000 R

=-=t t

12R (4.32)

= 215t

Equation 4.32 is recommended by API Standard 12 C for both umbrella roofs and dome roofs. When one applies the same factor of safety, 12, dome-shaped roofs have greater elastic stability than umbrella roofs. This may be shown as follows. ELASTIC STABILITYOF DOME ROOFS. For elastic stability of a thin sphere under external pressure, (42) 2Et2 (4.33)

For steel, p.

= 0.3, and E
p

30 X 106 psi; therefore

2 X 30 X 106t2 -:--,:====== r2 V3(1 - 0.32) V (36.3 X 106)/p

~=
t

Introducing an elastic-stability factor of safety of 12.0, the same safety factor that was included in Eq, 2.25, used in derivations of cone- and umbrella-roof operations, we obtain:

SELECTIONOF ROOF PLATES. A tank of this diameter requires column supports and rafters and girders. If the rafter lengths are to be limited to about 20-ft spans, two polygonal rings of girders plus one central column will be necessary. All roof plates will be cut from two standard sizes of plates carried in stock: plate size A, 72 in. wide x 25 ft, 3H in. long, and plate size B, 72 in. wide x 22 ft, 1H in. long. All plates will be % 6 in. thick (7.65 lb per sq ft). A study of various combinations of the above plates together with sketch-plate requirements indicates that an economical roof-plate layout will be like the one shown in Fig. 4.12. Such a layout results in a small amount of scrap in cutting out the sketch plates. RAFTERANDGIRDERSPACING. A suitable rafter-spacing layout based on the use of two polygonal groups of girders is next determined. Girders of approximately 26-ft in length will be used because spans greater than 30 ft require excessively heavy structural sections. Using a radius of 22 ft, 0 in. to circumscribe the inner polygon of girders requires a polygon of five sides. To maintain symmetry a decagon will be used for the outer polygon, as shown in Fig. 4.13. To determine the rafter spacing. Eq. 4.25 may be used. The design load is 25 lb per sq ft live load plus 7.65 lb per sq ft dead load (% 6, in. plate). The allowable design stress for the roof plates will be taken as 18,000 lb per sq ft. Therefore, by Eq. 4.25,

l- t
=

_ v'2i/iJ _

3 2 X 18,000 :j'/p - 16 '\j (25 7.65)/144

74.6 in., maximum rafter spacing

Assuming p = 0.227, the same as for conical and umbrella roofs, we obtain:

Also, maximum rafter spacing = 27rft = 75.4 in. The minimum number of rafters in the outer ring can be determined by dividing the circumference of the shell by the maximum rafter spacing and equals (27rr/l) or
nmin =

~ = 103 V3.03/.227 t ~ = 3650

i

2 X 3.14 X 61 X 12 74.6

= 61.5 rafters, minimum number

70

Design of Bottoms and Roofs for Flat-bottomed

Cylindrical Vessels

0 0
®

0 0 0

29'-5y,"

,! I

35'-3!j"

41'-2%"
~.'

41'-1%" 53'-0y,·

II I I~I
@

8'-1"" 8'-4!j" 9~" 7'-0%" 1--:..=-11----=-,\1

r:----:--l

¢.

8'-414". 8'-1%' 1,7'-0%'.1 \4-9"" ,: ,16'-8' :. Cui2 as, 72·.10.2*.25'-3",'

II

No,
30

36

II,
l'
10'-81,;"

@
12'-7""

.1

I
I

Cui2 PLS,72' • 10,2* • 25'-3""

II·

'11'

13'-6""

®
13'-6""

'1

®
10'-816'

. II.

.1

I1: 0 /0 10\01 1Jr;.)""

. 6'-4%'.11 8'-6!l" 16'-9"" 15'-I~" : cui 2 PLS.72" • 10.2'" • 25'-3"'"

I.

8'-611'

3'-1'" 11.6'-4[6' '11 6'-9"'"1

444 44 444 4 4 4 4 42 2. 4 4-

8ill of Materials Make From MOrK Descnat/on 3D PIs.72·XIO.2#X2S~3J4" PIs. 72"%10.2#%25'-31'+" A 36 PIs.72"X10.2" X22~1'12" 8 PIs. 72" XIO.l ' X 22~ IY2" 2 PIs 72"X10. #X'25'-3~" 2 C PIs. 7Z"XIo.2" X IZ~7s;," [) '2Pls 72"XlO2#X22'-I'!2" PIs 72"XI0.2"X I/-O'l~" 2Pls 72"Xla2#X2s'-3Y~" 2 PIs. 102 ' X Sketch PIs. 102 If % Sketch PIs. 102 " % Sketch PIs. 102 ., X Sketch PIs. 102 ;, X Sketch PIs PIs PIs PIs PIs 102 " 102 " 102 " 102" 102 It X Sketch X SKetch X Sketch X Sketch X Sketch X Sketch X Sketch X Sketch X Sketch X Sketch 12 7 /I
I~

J
J

2 PIs 72 "XI0.2' X2S'-3}+2 PIs 72·XI0.2" X25'-3J4" 2 PIs 72 "X10.2" X25'-3Y~" '2 PIs 72"XI02" X25'-3y",a '2 PIs 72" X10.2' X22'- tYz'

5
4 6

Cui2 ptS, 72' • 10,2* .25'-33(" 2'-7"'"

HI'

11' 2'

9
10.
fA

J
J

'1

II.
CuI2 PLp 72".10,2* .22'-1's·

®
l1'-O~'

.1

PIs 102 II PIs. 102 It PIs 102 II PIs 102 II PIs. 102 II

3
1.3

2 PIs 72"XID.2'X25'-3Y+"
2 PIs 72·XIO.2#X25'-3j~·

Ii

I'

9'-83("

ii'

12'-3)('

'1

II. CD .11. 1
10'-0'

@
l1'-ll~'

I .1
Fig. 4.12. 1Io0f plates for a 122-ft·diameter tank.

Shipping Weights 3D "A "PIs. 46,410 36 ~"Pls. 48, 744 12A-Sketches 8's'Skelches 18,56+ 10.,832

CuI2 PLS,72' • 10,2* .22'-1's"

Example Design 4.2, Design of Roof and Structural

Supports for a 122-Ft Tank

71

As indicated in Fig. 4.13, a 10-sided polygon, or decagon, will be used to support one end of these rafters. Therefore, the minimum number of rafters required must be a multiple of 10. Thus it may be seen that the minimum number of rafters between the outer shell and the decagon of girders is equal to 6.15 rafters per girder. An integral number of 7 rafters per decagon girder will be used, giving a total of 70 rafters. To check the spacing of these rafters on the girders Eq. 4.26 may ~ used.

2R sin 360 2N
(2)(10)

= (2)(39.75) sin ~ = 79.5 sin 18°

= (79.5)(0.309) = 24.6 ft
Average rafter spacing

= 24.6 ft/7 rafters = 3.51 ft

This spacing is less than the 5 H ft maximum allowable rafter spacing on the inner ring. The minimum number of rafters required between the decagon and the pentagon can be determined by using Eq. 4.27, or n=--sm-

Notes: All holes '~.' q, unless otherwise noted All caps to be welded to columns in shop

No. 5
10 10 10 40 70

DeSCription Girders lurp Girders lugs Ilaflers

8ill of Materials Mark GI A to Fine/.

Make From SI!J IS'@33.9* 12S'-9'ts' IBor S'I Ij,' X 33'- 9'1,' IOI!JIS'#133.9# I 24'-6j40' I Bar 5' 11;,' U6'-9Yz· IO'!" O·@II.S# I 17'-7Ja,' 10.!" 8'#111.5# 1 19'- 6'U,' 400£ S·fPlI.S# X Zt'- 7'116' 700h 8@/I.S# I 24'-711' lu 12"@20.7#154'-712· luI0·@15.3*XS3'-1I1j4· IJ. 6·X6'1~8·fO~2Yz· II'/, Jr¢lJ;4* S~ IZ'@ 20.7# X50'-101;4" SIEJ 9'@ /3.4# X SO'-3Y2" 6·X6"X~*XO'-2Y2·

24NR . 360 1 2N
(24)(10)(39.75) sin ~ (2) (10) 74.6

GZ
G tol il1&l III 112 113 114 CI

= --------------~~~~
= 127.8 sin 18°

= (127.8)(0.309)

39.5
5

The actual number of rafters to be used must be a mult.iple of the number of sides in a pentagon and the number of sides in a decagon. Therefore, 40 rafters will be specified. The minimum number of rafters within the pentagon may also be determined by use of Eq, 4.27, or (24)(5)(22.0) sin 360 (2)(5) 74.6

Column Channel Channel Clip Cop Columns Channel Channel Clip Cop

CIC C2

sa
C2C C3

5pIs.lr

X 20.4#X/~5"

10

cotumns
Channel Channel Clip Cop Splices Splices Column bases ChannelS" Gussels lugs

= --------------~~ = =

34·.7sin 36° (34.7)(0.588)

=

20.4

10 16

C3C GIS G2S CB

10>h12"@ 20.7'" X48'-7~,' 10>h 9"@ IJ.4I1X'I7'-II'Y16" tao 6"X 6- H'8" X 0'- 2 Y2" 10 pls.12"X ZQ4# X I'-Z" SPls 4"11;8" XI~OY8· IOPls.4"X1;8" I

r-os«:

The actual number to be specified should be a multiple of five. This means that 20 rafters would be a most convenient number if its use can be justified. An examination of Fig. 4)2 indicates that the maximum span between rafters is less than that determined by Eq. 4.27, which is conservative. Therefore, 20 rafters will be specified. SELECTION OF RAFTER SIZE. The rafter having the greatest span controls the size of the structural section required. The maximum rafter span is indicated as rafter R4 in Fig. 4.13 and has a length of 22 ft, 9~{6 in. The spacing between the rafters at the shell periphery is approxi-

70 40 60 60 J40 140 6432

l!

49t!J 8"@I/.S"'X2'-6· 32Pls.10"@10.2#1 1'-/0" 701} 6".f'l"X"/8"XO~4·

Boll lisl Rafters to cop rounfersink Girders to columns Girder sflices llaHus to girdlfrs Rafters to lugs Column. to bases

~4"I
X ~4" X ~.f" X ~.f' X

%"

2Jt'4' 2" 2"

1'1.,"
I~.f"

~»: X 172"
..."
tank.

Columns to bases
Roof-support assembly

3/4"XI3j

Fig. 4.13.

for a 122-ft-diameter

--_

72

r
2'

.,.- J:

nnn
~1~112" 13$' \1
...4::

Design of Bottoms and Roofs for Flat-bottomed

12

'

\j12"

1' \fr~r\r"
~12' ~12' I/-SK,W'
1'-5'"

Cylindrical

I~H If! 2r/

12"V
12'1/ l'-5Ji:
3' ...1!{,"

Vessels

" J

12" ~

12"11 12"11
1'-8,,·

Hr7r7
1'-10~,'

12"1/t3$~._ ; 13$~~ 14

llo"'V~ <2lS" 1'-10}('" llo" ~2'-0~." ~" ,. 2'-0 li," 3'-10%" '4 1,,"., 2r--

1'-8"" 5-6%"

3'-1 ~,'
8'-71Ys,"

1'_4111:," 1'_41~." 1' ...5"

.
2 1"

r~ ®
F
lo'

10'-1%" 11'-6"· 12'-U,!," 14'-4"

F @

W®

I T
I

15'-9" 17'-21f,'

20'-3%'

15"w@33,9fx25'-9%"

CD

9 ®

19

"'-0"

"ql~

23'-10",

"'_IO'r.~'
5'-1,"~

2'-O!-i,"

1%· '~5X'" ...

.

25'-10"" Inner-row girders make 5, mark Gl \13$" ~12'

..l\..

t "'_
~

2lS'

'ft"
2 ~'''~ ~ 1+2lS' 2' ~12'

\l'f,'
~12"

IJi'U llW
t
I

2'W
12'~

12',

12"~

' 12'V 12'V

2'-2'1',' 21'-4~,' L414'!

2f1 3r1
__..~2" 2lS"-j .',
'

114' I' 41(' 1-1'-10'i'~1

~r
I

_u.

-I

. .IL

1'-8%' 5'-5'

3'-5'i.' 8'-10'f,"

5,,·

2'-1 Jlli:,'

2'-11'~.' 15'-3~,'

5'" 15'-S,!,'

3'-5'1',' 19'-1111'

1'-104f," 3'-21(,'

9'-3111{ 12'-3'"

I -'22'-S~," 24'-6""
~ ~--··I
@)

1'-10 'f,'

I t 2"

2· .... ~r2lS'

.. ~'"

.I -

n Q)

Q)

V -+
I
24'-6'"

®0

<:0

2lS"--1
1("-

-;1'

Outer-row girders make 10, mark G2

2~.'

Inner-girder splices make 5. mark GIS 10PLS5' x "" x 5,,· 10 PLS5" x ". x 611,' . 10PLS 5'x""x7%' 10PLS5" x ". x 8~,' ) Girder lugs
/

0(

V')

Required mark Required mark Requiredmark ReqUired mark Required mark Required mark Required mark Required mark

/j

// ?

10PLS 5' x "'x 8'!,' 20 PLS5" x ,,'x 4' 20 PLS5" x ,,'x 4lS' 20 PLS5" x ,,'x 5'!,'

Outer-girder splices make 10. mark G2S

20 PLS5' x ". x 5111' Required mark 20 PLS5" x ". x 51'f,' Required mark 10PLS 5"x ,,'x 6'"
Required mark

Fig. 4.14.

Girder details for a 122-h-diameter

tank.

mately 5~ ft, and the rafter spacing at the decagon end of the R4 rafters is approximately 3~ ft. The design of the rafter is based upon the roof load plus the weight of the rafter: Since the rafter size and weight is unknown, a preliminary design based on the roof load onl y will be made and later will be checked with the weight of the rafter selected. For the preliminary calculation the rafter load will be

assumed to be uniform and will be taken as the load induced by a roof plate having an average width of 4~ ft. The live load is taken as 25 11>per sq ft, and the 7l6-in. roofplate weight is 7.65 lb per sq ft; this gives a total design load of 32.65 lb per sq ft. This corresponds to a load of 0.227 psi. Following the procedure presented in the section entitled "Selection of Rafters and Girders," we find that the uni-

Example Design 4.2, Design of R90f and Structural Supports for a 122-Ft Tank

73

formly loaded beam with this thin type of end support has a maximum bending moment as given by Eq. 4.8.
Mmax

=_
.

wl2 8

. (see item 1 of Appendix G) is a 15 x 3% in. channel weighing 33.9 lb per ft and having a section modulus of 41.7 in.", Checking the girder by including the weight of the girder, we find:

w = 0.227 psi X (4.5 X 12) 1 = 22.67 X 12 = 272 in.

M max

= 12.25 lb per in., roof load

w = 57.6

12 8

+ 33.9

= 60.4
= 656,000 in-lb

= -'--_ _:_:__'8

(12.25) (272) 2

M max = (60.4)(295)2 z=M

= 113,500 in-lb
z

656,000 18,000

36.5 in.3

= 113,500 = 6.28 in.?

18,000

Referring to item 1 of Appendix G, we find that the lightest American Standard channel section that can be specified is an 8-in. x 2H in. l1.5-lb-per-ft beam having a section modulus of 8.1 in.3. The weight of rafters should be included in the rafter load. This added load amounts to:

w = 11.5 lb per ft
12

0.958 lb per in.

The total load is 12.25 .96 or 13.21 lb per in. Recalculating Mmax, we obtain: M max

Therefore, the girder selected is satisfactory. SELECTIONOF COLUMNSIZE. A total of 16 columns will be required: 10 for the decagon, 5 for the pentagon plus 1 supporting the apex of the cone as shown in Fig.- 4.13. The roof area and its corresponding load increase per column as the distance from the tank center increases as a result of the roof-support layout. The total roof load supported by each column (C3 of Figs. 4.13 and 4.15) of the decagon is equal to the load per decagon girder plus the weight of the girder itself or is 60.4 lb per lin in. of girder length from the previous girderdesign calculation. Therefore, P
=

60.4 X 295

= 17,800 lb

(13.21~ (272)

122,000 in-lb

z = 122,000 = 6.78 in.3

18,000 Therefore, the rafter selected is satisfactory. SELECTION OF GIRDER SIZE. Consider first the girders (G2 of Figs. 4.13 and 4.14) of the decagon. which have a span of 24 ft, 6% in. The girders will be assumed to act as a uniformly loaded beams carrying the rafter loads. Each of these girders supports one end of 11 rafters. The maximum rafter loading is 13.211b per lin in. over an average rafter span of (61 - 22)/2 or 19.5 ft. Assuming that half of the total load carried on each rafter is supported by the girder, we can calculate the roof-plus-rafter load as follows: us-ra f ter I d oa R 00 f-p I

The minimum radius of gyration of the column section is a function of the length of the column under consideration. If the ratio of (l/r is not to exceed 180 and the length of the column is 48 ft, O}-l6 in. (576.5 in.) as shown in Fig. 4.15, then the minimum radius of gyration is: 1 r
=

180

1 r=180

= --

576.5 180

= 3.21 m.

= -'--_":"':--'-':-':"";__';':_
(295)(2)

(13.21)(19.5)(12)(11)

= 57.61b per lin in.

.
Mmax

=""8

wl2 (57.6) (295) 2 8

Referring to item 9 of Appendix G under the heading, "properties of sections consisting of two channels," one observes that the lightest channel combination which will provide a radius of gyration of 3.21 in. or more about both the x-x and y-y axis is the combination of a 9-in., 13.4-lb channel and a 12-in., 20.7-lb channel. This combination has a value of 3.41 in. in relation to the x-x axis and a value of 3.62 in. in relation to the y-y axis. This combination provides a cross-sectional area of 9.92 sq in. and has a weight of 34.1 lb per ft of combined section. The allowable compressive stress for the column may be calculated by use of Eq. 4.21 as follows: 18,000 18,000 (600)2 (18,000) (3,41)
---"C____:___--::

= 627,0'00 in-lh
z
=

M = 627,000 = 34.9 in.3

1+
_ -18,000 _ 66?0 _ 2.72
pSI

18,000

The lightesl channel which will provide this section modulus

74

Design of Bottoms and Roofs for Flat-bottomed

Cylindrical

Vessels

~-------,,'J.tL-,817x f---------.%OhOS

*L"OZ@"ZI x *L"OZ@.ZI

£.? "IOJJOI.'Yo,S-,817 ----------;., lO ·IOJJOI.Si6-.oS ----------;., lD 'IOJ .1f9-,17S JOI ----------;.; Pia", snonuuuoo .9 ~

.,.
2
E

ii

..
I: II

" ~

:: N N

~~__i
TI
1 1

t. / ~If:t:
I I I
1 1

-- + t~
Q.~
",<.)~

.2
II "U

.".
..

:&

I
1 1

I
1

I I I

I I I I I

"E~
~ E~

E .2

I:

0 u

I
1
1

I-!;:!I I ': ~ I l,g ,g I

I I I I I

i
1 1 1

~~
II>

11>..

"U I: II

(.)2

.:
II<

.. "
"

!;:! ,g

I~I~~I
1::::::1

1--

·..l('1
It.

1>< ><)

I"!:: 2: I "

.....ID

~
"

, ~

I '" " I'" I~

~ ..... ~

-, I
~:!

1 1 1

I
1

iii

,I® 100
1 1
1

...

I
1

I I
1 1

I
1

II:

til

'"

I
1

I I

I ,J

I I I I I

,
;;,.'!!.

'. ~
(1\

I®®I

I~~I I I iil I 1 ,g ,
1

1 J JI

I I

111
~

I I

I I
1

1 1
1

!. .,.

I" ~ I

'"

I + '. -,I ;::.. , iI N 1 <:))1I '" I II I I

'" " '" '"

'" '"

"

~ Q.", "'~ "E:aE

<.)"" <.)

.!.ui

c .. -2

~.:

11(-

~f!-r

I ~I I~1 I :!I

I I I:C .... liC..J

I I I
"----..J

1~1
1

~.t ~
'"

Example

Design 4.2, Design of Roof and Structural

Supports

for a 122-Ft Tank

75

The actual induced stress is: j=j

p

it does not support girders. Its length is 54 ft, O"i in. or 648 in. Therefore,
l 648 . r = - = - = 3.61 Hl, 180 180

= 17,800 + (50 X 34.1)

9.92 _ --19,505 _ 2270 pSI. 9.92

Therefore, the column combination is satisfactory, and the radius of gyration is controlling. The size of the pentagon column supports (C2 of Figs. 4.13 and 4.15) can be readily determined because it is recognized that here also the radius of gyration is controlling. These columns have a length of 50 ft, 4 in., or 604 in., as shown in Fig. 4.15. Therefore, r
=-

180

=-

604 • = 3.36 ffi. 180

Thus the same column combination called for at the decagon supports will also be specified here. The central column (Cl of Figs. 4.13 and 4.15) has a greater length because of the roof pitch and the fact that

Consider a combination of a 10-in. and a 12-in. channel. This combination has a radius of gyration of 3.83 in. in relation to the x-x axis and a radius of gyration of 3.52 in. in relation to the y-y axis; therefore the average radius of gyration is 3.68 in. In view of the low induced compressive stress on this column and the average radius of gyration of 3.68 in. as compared with the 3.61 in, required, its use in this application can be justified. It is customary for many tank fabricators to use built-up structural sections for columns as is done in this example. However, it is apparent that an appreciable saving in column material could be realized by using pipe in which the induced stress would be higher. For example, 10-in. schedule-l0 pipe having a radius of gyration of 3.74 in. and a cross-sectional area of 5.49 sq in. would be satisfactory. This area compares with the area of 10.50 sq in. for the 10-in., 15.3-lb and 12-in., 20.7-lb channel combination for the center column. Thus a material saving of 48 % can be realized. This gain may be partially offset by the greater cost of pipe.

J:
PROBLEMS

~I
I.

1. Using the dimensions given in Fig. 4.16, determine the required section modulus, z,'for rafters RA, RB, RG, and RD. 2. Using the dimensions given in Fig. 4.16, determine the required section modulus, z, for girders GA, GB, and GG. 3. Using the dimensions given in Fig. 4.16, determine the required radius of gyration for columns Gl, G2, G3, and G4. 4. Derive a relation comparable to Eq, 4.6 for use with aluminum roof plates.

Fig. 4.16. tank.

Rafter

and

girder

layout

for

150-ft-diameter

cone-roof

C H A P,T E R

PROPORTIONING

AND

HEAD SELECTION

FOR CYLINDRICAL VESSELS

/

WITH FORMED CLOSURES

he real need for the use of formed closures on cylindrical vessels arose with the development of the power steam boiler early in the nineteenth century. As a result of the frequent occurrence of boiler explosions, the British House of Commons in 1817 made the recommendation that the heads of cylindrical boilers be hemispherical (12). Since then a wide variety of formed closures termed "heads" have been developed, standardized, and extensively used in the fabrication of process pressure vessels. The development of the thermal cracking process in the petroleum industry during the period from 1915 to 1930 resulted in the construction of thousands of pressure vessels with formed heads operating in the range of from 100 to 400 psi. The heads of these early vessels usually were of the tori spherical-dish type with a small knuckle radius. The first formed heads were of a small size and were hand-forged by "bumping out" a flat plate. One of the ~arly American steel producers, Lukens Steel Company, m 1885 formed a 5-ft-diameter dished head by digging a hole in the ground to the approximate radius of the dish and bumping the heated plate into the depression by the use of mauls. Since then methods of-forming heads have been highly developed by the use of dies and forging and spinning techniques. Figure 5.1 shows a photograph of the world's largest flanging machine spinning a head with a 20 ft, 6 in. diameter. 5.1
GENERAL CONSIDERATIONS

5.1 a Development of Welded Construction. Theearly thermal-cracking plants of the petroleum industry used pressure vessels in which the formed heads were riveted to
76

the shell. These vessels had the fault of frequent leakage around the rivet heads. Attempts to correct this difficulty were made by means of fillet welding the plate edges and seal welding the rivet heads. These vessels often were not . satisfactory unless the fillet welds were made so large that the loads were carried by the fillet welds rather than by ~he rivets. When it was realized that the welds were carrymg the loads rather than the rivets, a large number of vessels for low-pressure service (walls less than 1 in. thick) were fabricated entirely by oxyacetylene welding. The limitations of the welding art at this time, in particular the brittleness of the bare electrode welds, made the construction of heavy-walled vessels impracticable. With the development of flux-coated electrodes ductile welds were possible. This development resulted in practical obsolescence of riveted-fabrication techniques for pressure-vessel service. 5.1 b Use of Formed Heads. Cylindrical vessels with formed heads are used for a wide variety of applications in which cylindrical tanks with flat bottoms cannot be used. These applications can be grouped into three classes: (1) functional use, (2) pressure consideration, and (3) size limitations. Processing equipment such as distillation columns desorption units, packed towers, evaporators, crystal1izers: and heat exchangers are essentially cylindrical vessels having formed heads plus other required functional parts. If the working pressure of the process vessel is to be other than atmospheric pressure, formed heads are usually used to close the vessel. In general, all cylindrical vessels requiring a working pressure in the vapor space of about 5 lb per sq in. gage or more are fabricated with formed heads. Large-diameter

Material

Specifications

Fig. 5.1.

World's largest flanging mochine spinning

heod 20 ft, 6 in. in outside diameter.

(Courtesy of Lukens Steel Compony.)

flat-bottomed, cone-roofed storage vessels are limited to u working pressure in the vapor space of only a few ounces. However, cylindrical vessels with flat bottoms and considerably smaller diameters may operate under allowable working pressure of several pounds per square inch if a domed or umbrella roof is used. Equipment designed to operate under less than atmospheric pressure will also require the use of formed heads. Small horizontal storage vessels supported off the ground are usually fabricated with formed heads although flat ends of heavy plate are sometimes used. 5.1c Vertical versus Horizontal Vessels. In general, the functional requirements of the vessel determine whether the vessel shall be vertical or horizontal. For example, distilling columns and packed towers, which utilize the force of gravity for phase separation, require vertical installation. Heat exchangers and storage vessels may be either vertical or horizontal. In the case of heat exchangers, the selection is often controlled by the routing of the fluids and heattransfer considerations. In the case of storage vessels, the installation location is important. If the vessel is to be installed outdoors, the wind loads on vertical vessels may impose the necessity of heavy foundations to prevent overturning. For t his reason, horizontal storage vessels are

usually more economical. However, other important considerations such as available floor space or ground area, head room, and maintenance requirements may be determining factors.
5.2 . MATERIAL SPECIFICATIONS

Vessels with formed heads are most commonly fabricated from low-carbon steel wherever corrosion and temperature considerations will permit its use because of the low cost, high strength, ease of fabrication, and general availability of mild steel. Low- and high-alloy steels and nonferrous metals are used for special services. The steels commonly used fall into two general classifications: (1) the steels specified by the ASME code for unfired pressure vessels (11), often referred to as "boiler-plate steels," of flange or firebox quality; and (2) structuralgrade steels, some of which are permitted by the above code in certain applications and which are widely used for the construction of storage vessels under specifications given in API Standard 12 C (2). The design of vessels in accordance with the ASME code for unfired pressure vessels is treated in Chapter 13, which includes a description of the materials and specifications. The discussion in this chapter will be restricted to those steels used in the fabrica-

- ~-,---------78 Proportioning and Head Selection for Cylindrical Vessels with Formed Closures ASTM-A-7, A-283, Grade C, and A-283, Grade D are the most commonly used plain carbon steels in the construction of storage vessels and are widely used for vessels with formed heads, especially the steel designated as ASTl\IA-283, Grade C. Steel A-283-54 is of the structural quality intended for general applications. It is available in four grades, A, B, C, and D, having minimum tensile strengths of 45,000, 50,000, 55,000 and 60,000 psi respectively, as given in Table 5.1. This steel is available" in thicknesses up to and including 2 in., but its use in vessels designed to code specification is limited to thicknesses up to, and including % in. Grades A and B are primarily used in severe cold-forming applications where high ductility is of prime importance and tensile strength is a minor consideration. On the other hand, Grade D does not have sufficient ductility for easy shell and head forming and is not as 'easily welded as Grade C. As a result, Grade C is the most widely used structural-quality plate steel for vessel construction. The major portion of all oil-storage tanks, elevated tanks, water standpipes, and other varieties of tanks of all descriptions, involving both dishing and rolling, are constructed of ASTM-A-283, Grade C. Steel A-7 is intended for use in the construction of bridges and buildings and for general structural purposes. It has physical properties identical to A-283, Grade D. These steels are the same whether made by the open-hearth or electric-furnace processes. However, steel A-7 is also made by the acid-Bessemer process, and steel made by this process is not recommended for vessel construction. Steel A-7 is available in all standard thicknesses, and its use is permitted in vessels designed to present code specifications and having shell thicknesses up to and including U in., providing the steel has properties equivalent to A-283, Grade D. Steel ASTM-A-113-55 is a structural steel used for the construction of locomotives and railroad cars except where firebox boiler plate is required. It is made by either the open-hearth or the electric-furnace process and is available in nearly all standard thicknesses. This steel is made in 3 grades, A, B, and C. Steel A-113-55, Grade B has properties approximately midway between those of steels A-28354, Grades C and B, as shown in Table 5.L Note that the grade specifications for tensile strength for the A-1l3- steels run in the reverse order of the grade specifications for A-283 steels. There is no particular advantage to using this steel in preference to A-283 steels except when it is more readily available; It may be used for vessels designed to present code specifications with the same limitations as for A-283 grade steel. Steel ASTM-A-131-55 is an improved structural steel intended primarily for use in ship-construction. Formerly, the specifications for this steel were essentially the same as for A-7 and A-283, Grade D. To improve the quality of ship-hull steels, the specification was changed in 1950 in order to include an increase in quality specifications with increasing thicknesses. This logical requirement of increased quality with increased thickness warrants consideration of this steel as a material of construction for heavy-vessel fabrication. For this steel there is a limitation on the maximum percentage of carbon and a range of from 0.60 % to 0.90 % manganese for all plates thicker than H in.

tion of vessels with formed ends not requiring fabrication in accordance with these codes. 5.2a Comparison of Specifications for Structural- and Boiler-quality Steel Plates. Structural-quality steel rather than boiler-plate-quality steel is used in the fabrication of many vessels with formed heads because of economic considerations and its availability. Both types of steel are a vailable in th e "killed" and the "semikilled" or rimmed quality. A" killed" steel is one completely deoxidized by the addition of aluminum, silicon, or manganese at the time of the casting of the ingot. The purpose of killing is to minimize the interaction of carbon and oxygen and to reduce the formation of blow holes. A completely killed steel requires "hot capping," more time in the soaking pit, and more time for the ingot heating. .. Hot capping" is the use of an insulated mold on top of the ingot mold to hold a molten reservoir of metal for feeding the ingot as it shrinks on solidifying. A partially killed or rimmed steel is a partially deoxidized steel. An ingot of rimmed steel has a high-purity,low-carbon steel rim from which it obtains its name. Fully killed structural steels have no advantages over boiler-plate steel because of their high cost and limited availability. One of the major differences between boiler-plate steel and structural-plate steel is the "quality" control dictated by the number and severity of test requirements. As far as chemical requirements are concerned, the principal difference expressed by ladle analysis is the more restrictivelimit placed on phosphorus and sulfur for boiler-plate steels. The thickness tolerances are the same for boiler-plate steels and structural steels when plates are ordered to a given thickness. The physical tests are the same for both steels except for the number of tests and the stipulated location for test specimens. Sirudural-qualiiy plate steels require only two tension and two bend tests from each heat of metal which may contain oyer 100 tons. Flange-qualily boiler-plate steel requires one tension and one. bend test from each plate rolled. Firebox-quality boiler-plate steel requires two tension tests and one bend and one homogeneity test from each plate as rolled. There are also minor differences in the methods permitted for repairing surface defects in the slabs prior to rolling. Boiler-plate steel such as SA-285 flange quality and firebox quality had mill quality extras of $0.40 and $0.50 per ! 100 lh respectively as of January 1956 (see Appendix: C). Other boiler-plate steels such as SA-212 and SA-201 had mill quality extras of from $1.20 to $1.55 per 100 lb, depending upon thickness and grade. Killedsteels had mill extras of $0.65 per 100 lb .. The ~e of structural-grade steels results in the minimum of quality-extra charges, and the use of these steels is justified whenever permissible. In selecting steels for pressure-vessel fabrication to satisfy code requirements, Chapter 13 should be consulted. 5.2b Types of Structural-steel Plates. The most widely available / types of plain-carbon structural-steel plates are listed (67) in ASTM-A6-54T. Those most suitable for vessel construction are A-7, A-113, A-131 and A-283. Specification ASTM-A6-54T gives the general requirements such as permissible variations in dimensions and weight, methods of testing, correcting of defects, and rejection (67).

Proportioning of Vessels with Formed Heads Table 5.l. Tensile Strength, psi 45,000 50,000 55,000 60,000 60,000 to to to to to 55,000 60,000 60,000 72,000 72,000 Min Yield Point, psi 24,000 27,000 30,000 33,000 33,000 32,000 32,000 32,000 1955 ASTM Steel Specifications (67) Max Min % Thickness Min % Elong., in., Elong., in., Available, 2 in. 8 in. in. 2 2 2 2 15 ~ and less H to 1 1 and over 27 25 23 21 21 30 28 27 24 24 Max % P (ladle) (basic) 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04

79

Steel A-283-54 Grade A Grade B Grade C Grade D A-7-55T A-131-55* Grade A Grade B Grade C A-1l3-54 Grade A Grade B Grade C

Max % C (ladle) no no no no no spec. spec. spec. spec. spec.

Max % S (ladle) (basic) 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05

58,000 to 71,000 58,000 to 71,000 58,000 to 71,000

21 24 (see Ref. 67) (see Ref. 67) 21 24 26 24 38 no spec.

no spec. 0.23 0.25 no spec. no spec. no spec.

33,000 60,000 to 72,000 27,000 50,000 to 62,000 26,000 48,000 to 58,000 -....... * See text for limitations.

Also, for plates having a thickness of 1 in. or more, a requirement of 0.15 % to 0.30 % silicon is specified. In addition, it is stipulated that this steel be manufactured to have an inherent fine-grained structure. This steel is available in a wide range of thicknesses and is of higher quality than A-7 but presently is not permitted in the construction of vessels designed to meet unfired-pres sure-vessel codes. The additional quality requirements for heavier plates of thiR steel will increase its cost and may thereby eliminate any savings from using it instead of boiler-plate steels. Other structural-quality steels listed in ASTM designation A-6-54T are A-8, A-94, A-284, and A-242. Steel A-8 is a 3.0% to 4.0% nickel steel containing a maximum of 0.43 % carbon and having a tensile strength of from 90,000 to 115,000 psi. It is intended for use in main stress-carrying structural members. The nickel addition results in finer, stronger, and tougher pearlite than is found in plain carbon steel and appreciably increases the yield point, fatigue limit, and impact strength. The difficulties of welding this steel plus the cost extras for nickel addition precludes its use for vessel construction. Steel A-94 is a structural silicon steel containing a maximum of 0.40 % carbon and a minimum of 0.20 % silicon and having a tensile strength of from 80,000 to 95,000 psi and a minimum yield point of 45,000 psi. This steel may be eliminated from consideration for vessel construction On the basis of welding difficulties and the cost extras for fully killed steel. Steel A-284 is a low- and intermediate-strength carbon-silicon steel containing from 0.10% to 0.30% silicon and having tensile strengths of from 50,000 to 60,000 psi depending upon the grade. The steel is coarse-grained and requires heat treating for grain refinement. The presence of the silicon tends to dissociate carbides to form soft graphite thereby weakening the welded joints. For these reasons and because this steel is a fully killed steel and therefore involves cost extras, it is not economical to use it for vessel construction. Steel A-242 is a low-alloy structural steel intended primarily for use as a stress-carrying material of structural members when saving in weight and atmospheric-corrosion

resistance are important. Thicknesses are limited to not under % 6 in. and not over 2 in. It contains a maximum of 1.25 % manganese and a maximum of 0.20 % carbon. This steel has a yield point of 50,000 psi for thicknesses of from %6 to' % in., 45,000 for thicknesses of % to 1H in. and 40,000 for thicknesses of 1H to 2-in. in comparison to a yield point of 30,000 psi for A-283, Grade C. For the plates IH-in. thick and less this represents an increase of 50 % or more in yield strength. Using the same design factor of safety based on yield point results in a proportional decrease in metal thickness required to resist a given load. In designs in which stress rather than elastic stability or brittle fracture is controlling, the use of this steel rather than a plain carbon steel such as A-283, Grade C may result in a saving. See Table 3.2 for specifications for this steel and Chapter 3 for further discussion of its use. 5.3 PROPORTIONING OF VESSELS WITH FORMED HEADS

In general, the cost of a vessel may be considered to be proportional to the weight of the steel used in its construction. It would therefore appear that for storing a fluid under uniform pressure a vessel having the minimum surface area and thickness per unit volume would be the most economical. A spherical vessel has the minimum surface area per unit volume and the minimum shell thickness for a given pressure and volume. If the cost of fabrication were not a prime consideration, the most economical shape for a vessel would therefore appear to be a sphere. However, the fabrication costs of spherical vessels are so great that their use is limited to special applications. Cylindrical vessels are more easily fabricated, in the majority of cases are considerably simpler to erect, are readily shipped, and are therefore more widely used in the process industries. For a simple cylindrical vessel with formed heads, the optimum ratio of length to diameter, L/ D, is a function of the cost per unit area of the shell and the formed heads. More complex vessels such as distillation columns, heat exchangers, and evaporators have additional parts such as

80

Proportioning

and Head Selection for Cylindrical

Vessels with Formed Closures Solving for x2 we obtain:

Y;--

--------t-------- --it
x ---: a ------~

x2
b

= 4b2 _ 4y2 = 4(b2 _ y2)

Differential volume, dll Integrating we obtain:

=

TTl II"'"'~-yI
j_L-

I
~----------------~
D -----------i
I

1111.

A dy

11'X2dy

1----------------Fig. 5.2.

L~-----

Dimensions for a 2: 1 ellipsoidal dished head.

trays in distillation columns and tube bundles in heat exchangers which must also be considered in determining the optimum proportions. The proportioning of a simple vessel may be based either on the cost per pound of the material or the cost pe'r unit area of the material. In Chapter 3 the proportioning of flat-bottomed, cylindrical, cone-roofed tanks was based on the cost per unit area because land and foundation costs, which are important for such vessels, can best be considered on a unit-area basis. In addition, the cost of coned roofs and flat bottoms are relatively constant on a unit-area basis for large-diameter tanks. However, cylindrical tanks with formed ends for various pressure services have wide variations in thickness and therefore vary in cost per unit area. The cost of land area and foundations is usually a minor consideration for such vessels. Therefore, it is more advantageous to consider the cost of shell and heads in terms of unit weight rather than in terms of unit area. 5.30 Equations for Optimum with Elliptical Dished Heads.
I

The volume of an equivalent cylinder is: V

= 11'a2H

where H = length of cylinder Equating we obtain:

H= -

=-

D
6

(S.2)

Thus the volume of two ellipsoidal heads having a majorto-min or-axis ratio of 2.0 is: Vh =

(11'~2)

(f)

2=

11'~3

Proportions

of Vessels

Therefore, the total volume contained in the vessel is:

VOLUMERELATIONSHIPS. A cylindrical vessel closed at both ends with elliptical dished heads has a volume equal to the volume of the cylindrical section plus twice the volume contained in one of the heads. The volume contained in a head can be expressed in terms of a cylinder of equivalent volume having the same inside diameter as the cylindrical section of the head. Figure S.2 is a cross section of an ellipsoidal head having a 2: 1 major-to-minor-axis ratio. The equations for the volume relationships for a 2: 1 ellipsoidal head (103) are as follows. The equation of an ellipse is: (S.I) For a 2: 1 ellipsoidal dished head

where L = length of the vessel, tangent line to tangent line, between heads, feet. Solving for L, we obtain: .

a = 2b
Substituting we obtain:

COSTRELATIONSHIPS.The diameter of a circular plate required for forming an ellipsoidal head is approximately 22 % greater than the internal diameter of the finished vessel (103). Also, the cost of the formed heads is approximately 50% greater than the cost of the steel from which they are formed. This increase in cost results from cost extras for circular plates and the cost of forming and machining. Let
C8

I.S

CS

= cost of fabricated head, dollars per pound t = thickness of head and shell, inches p = density of steel, pounds per cubic foot

cost of fabricated shell;: dollars per pound

The cost of the shell section of the fabricated vessel is: Expanding we obtain:
csp

(11' DL

12

_!__) =

C
8

p1f

_!__ (
12

7rD2

4V _

Q)

Selection of Optimum Plate Dimensions

81

and the cost of two elliptical dished heads is 2 X 1.5c.p [~ (1.22D)2 ~] 4 12 or the total cost of the vessel is:

For vessels fabricated from plates from 2 in. to 6 in. in thickness, the thickness extra will modify the cost per unit weight. In this range of thickness the cost of the vessel may be estimated as varying approximately inversely with DY< (103). Or
C
8

= =

C8{Ylr t - V - D2 12 7rD 3
C
8

[4

+ i(1.22D) 2]
Substituting

=-

DH

Cs"

into Eq. 5.3, we obtain:

{Ylr_i_ [1.275 .!:: _ D2 12 D 3 [1.275 .!::

+ 1.115D2]

c/'k

[l.~!V

+ 0.782D2.75]
to zero to obtain the mini-

= C.{Ylr~

12

+ 0.782D2]

Differentiating and equating mum, we find that dC -dD 8.60D3 D3

But according to Eq. 3.14,

==

'4 --

11.

t
Substituting we obtain:

= pd = pD

275V Db/4

+ T(O
11

782)Dl.75 = 0

2/

24/

1.275 V 0.148V for V, we obtain: D3

C = C8{YlrpD
8

[1.275.!:: 288/ D

+ 0.782D2]
(5.3)

Substituting

= c k[1.275 V
where k
=

+ 0.782D3]

0.148 [ 7r~2 L

+ 11"~3]

{Ylrp 288f

D = 0.116L L 0.961

+ 0.039D
(5.5)

PROPORTIONING. The cost of the shell is not a constant but is a function of the weight of the vessel, which in turn is a function of the pressure and diameter. For vessels having a shell plate thickness of up to 2 in., the cost of the vessel may be estimated as varying approximately inversely with D'I.I (l03). Or

= 0.116 = 8.28D "'" 8D

Substituting

in Eq. 5.3, we obtain: C

=

c. 'k [1.275V D'I.I

+ 0.782D%]

Holding V constant, differentiating, and equating to zero in order to obtain the minimum, we find that dC dD .6.25D3
=
_1

"ll"

1.275 V
m'J"ll"

+ 8(0:782)D%.

= 1.275 V
for V we find that D3

D3 = 0.204V Substituting

= 0.204 [7rD2 L
4

+ 7rD3]
12

D = 0.160L

+ 0.053D
"'" 6D (5.4)

Or use Lj D = 8 for vessels with plate thickness of from 2 in. to 6 in. DIAMETER AND LENGTH LIMITATIONS. The selection of the proportions of a vessel may be influenced by other factors such as the maximum diameter or length that can be shipped by railroad flatcar. In general, the maximum diameter that can be shipped on most railroad lines is 13 ft, 6 in. Larger diameters may be shipped by rail but require special routing of the shipment. If water transportation is available between the fabrication shop and the erection site, large-diameter equipment may be shipped by barge or floated to the site. Two other alternatives are: (1) shop forming and partial fabrication by welding in sections with final fabrication in the field or (2) field assembly of plates cut and formed in the shop. The length of a vessel is not as critical as the diameter with respect to railroad shipping limitations because more than one flatcar may be used. Figure 5.3 shows an oil-refinery fractionation column loaded on three flatcars, supported on the two end flatcars with no load supported on the middle car. This permits the cars to negotiate a curve with the vessel pivoting on the end cars. Other considerations such as the selection of plate widths and plate lengths to minimize the number of welded joints may influence the proportions of the vessel. (See the following section.) 5.4 SELECTION OF OPTIMUM PLATE DIMENSIONS

L = 0.947 D = 5.93D 0.160 Or use L/ D

=

6 for vessels with plate thickness up to 2 in.

PLATE WIDTH. The cylindrical shells of vessels with formed heads may be fabricated by rolling and welding one or more plates together. A choice exists as to the plate

82

Proportioning

and Head Selection for Cylindrical

Vessels with Formed Closures

Fig. 5.3.

Oil-refinery fractionating tower ready for shipment on three flat cars.

(Courtesy of C. F. Braun & Co.)

widths and number of plates to be used. Usually a circumferential weld and sometimes a longitudinal weld may be avoided by using a larger plate width. Plates having widths in excess of 90 in. bear a cost extra which increases with increasing width. The most economical design is often one in which a wider plate is used, providing that a welded joiut is thereby eliminated and the cost saved by eliminating such a joint exceeds the extra cost of wider plates. An example of th.e reduction in cost that may be realized by the selection 'Of a plate size that will eliminate a welded joint is given by W. G. Theisinger (104) in regard to a purchase order involving 20 vessels 48 in. in diameter and 20 vessels

54 in. in diameter. The shells for these vessels might have been ordered as follows: 1. Two-plate shells a. 48-in.-diameter vessels 20 plates 157H X 87 X 1% in. 20 plates 157H X 85 X 1% in. b. 54-in.-diameter vessels 20 plates 176% X 91 X 11%6 in. 20 plates176% X 93 X 11%6 in. 2. Single-plate shells a. 48-in.-diameter vessels 20 plates 171 X 157~2 X 1% in. b. 54-in.-diameter vessels 20 plates 183 X 176% X 11716 in. For two-plate shell construction the extra fabrication costs were estimated to average $436.00 per shell or a total of $17,440.00 for the 40 vessels. With single-plate shell construction the extra for the 40 wide plates at width extras of $1.25 per 100 lb for the 157H-in.-wide plate and $1.50 per 100 lb for the 176%-in.-wide plate and the overweight allowance totaled $9,853.00; therefore, a net saving of $7,587.00 was realized by the purchaser by using singleplate shell construction. In addition, the fabricating time was reduced by 5800 man-hours, and this resulted in quicker delivery. It should be pointed out that these figures are for prices existing in 1944 and are not representative of current prices. When a vessel shell may be fabricated by one- or twopiece construction, the selection may be made by simply estimating the costs for each design and selecting the design giving the lesser cost. However, for larger vessels in which the shell' must be fabricated from many plates. the above

200
.c
lit GI

f
.c

.5100

80
/'

.,
~
I "

IP'

.!! 60 «I Q. 50

;
~
I !il

II 40 !il

30

.g

20
10 /' 0.1

).'

0.2 0.3 0.4 0.6 0.81.0 2 Ce cost extra, dollars per 100 Ib

Fig. 5.4.

Width extras for corban-steel

plates as of 1953.

Selection Table 5.2. Average Extra Fabrication Cost, Cw (104) Dollars per Foot as of 1944 (Based on 133 % Shop Burden) Code Welded Unclassified welds for N number will be:

of Optimum Plate Dimensions of plates (excluding

83

7rDCU/. The total cost of all of the circumferential

shell the head welds)

Gage, in.

~i ~i
~~ % 1
% %

~i

3.75 4.50 5.40 6.38 6.98 7.50 8.10 8.74 9.30 9.83 10.43 11.10

2.50 3.00 3.60 4.25 4.65 5.00 5.40 5.83 6.20 6.55 6.95 7.40 7.85 8.40 8.95 9.50 10.10 10.75 1l.50 12.35 13.00 13.70 14.35 15.20 15.90 16.60 17.25 18.00 18.85 19.40 20.10 20.90

(N -:- 1)7rDCw = (~ where l = length of shell, inches D = shell diameter, feet The additional cost of using plates given by the equation:

1)

7rDCw

(5.7)

wider than 90 in. is

7rDltP) ( 144

Ce .'\ = ( 7rDlt490 ) 100"( 144 X 100

[_!__ ]
145

(in _ 90)1.23 (5.8)

The total extras for using plate widths wider than 90 in. plus the costs for all the circumferential joints exclusive of head joints is given by the sum of Eqs. 5.7and 5.8as follows:

1% 1%
2

Us

11.78 12.60 13.43 14.25 15.15 16.13 17.25 18.53

C = 7rD [ (~ - 1)

c; + 0.000235lt(w

90)1.23] (5.9)

Differentiating the cost, C, with respect to plate width, w, and equating to zero to obtain the minimum, we find that dC = 7rD -C2w dw w

+ 0.000235t1.23(w90)0.23

90)0.23= 0 (5.10)

2% 2%
3

2~i

19.50 20.55 21.53 22.80 23.85 24.90 25.88 27.00

Cw

=~

(w _

3460

3% 3% 3%
4

28.28 29.10 30.15 31.35

Solving Eq. 5.10 for w gives the optimum width of plate to give minimum fabrication cost for the shell as a function of joint fabrication cost, Cw, and shell thickness, t. This equation is plotted in Fig. 5.5 for convenience. Since 1953 the width extra has been combined with the thickness extra (see Appendix C). Therefore Fig. 5.5 is useful only for first approximations. PLATE THICKNESS. Plates having thicknesses of from ~-2 in. to 1 in. are available from mills at base cost with no thickness extras. To avoid extras for plates thicker than 1 in., a higher-strength steel often may be used to advantage. This is of particular importance in connection with

irocedure is not so simple since a number of designs may )e possible. To determine the optimum number of plates, .he plate width resulting in the minimum cost for the fabri:ated shell can be evaluated mathematically .. The width extras for plain-carbon-steel plates as of 1953 rre shown in Fig. 5.4,in which the cost extra in dollars per 00 lb is plotted against w - 90 where w is the plate width n inches. The equation of the line given in Fig. 5.4 is:

40

~ .5 30
::J ... ... Q)

~ :v ~

8.; 20
o ."

;
17
1/

1 C =e 145
rhere Ce
=

~ ~ 10 ]5 en 8

tV

VV

i>

V i-

j...-

i-

..... 1---

(w -

90)1.23

(5.6)

~~

dl-~

II ~ (.)6

w=

dollars per 100 lh plate width, inches

5 4 3 90

II

The cost of circumferential welding, Cw, including the ost of preparation of the joint, is usually expressed in erms of dollars per foot of weld and is given in Table 5.2. 'he fabrication cost per circumferential weld will be

100 110 120 130 140 150 160 170 180 190 200
w

= optimum

plate width, inches construction based on

Fig. 5.5.

Optimum plate

widths for vessel-shell

width extras as of 1953.

84

Proportioning and Head Selection ·for Cylindrical

Vessels with Formed Closures

Fig. 5.6. Forming dished heads by "drawing" in a press. Detail a: Taking circul.ar blank plate from furnace for forming. thick for a vessel 60 in. in inside diameter in a 1000-ton press. (Courtesy of C. F. Braun & Co.)

Detail b: Forming of head

1X 6 in.

vessels designed to meet code requirements and is considered in detail in Chapter 13. The 1957 practice in steel pricing combined thickness extras with width extras (see Appendix C). PLATE LENGTH. Plates having lengths between 8 and 50 ft are available from mills at no length extra. If possible, the plate lengths selected for a vessel should be within these limits. Warehouses usually do not stock plates longer than 40 ft, and this length is usually carried only in plate thicknesses of % in. or less and plate widths of 72 in. or less. The plates of heavier gage (up to and including 3 in.) and greater width are usually carried in 20-ft lengths at the warehouses. The maximum plate length, thickness, and width that can be handled by the shop fabricating the vessel may impose a limitation on the size of plate that can be handled. 5.5 COMMON TYPES OF FORMED HEADS AND THEIR SELECTION

Although the cost of heads formed from flat plates involves the additional cost of forming, the use of formed heads as closures is usually more economical than the use of flat plates as closures except for closures of small diameter. This can be shown by comparing the thickness required for closures of flat plates with that of various types of formed heads. Figure 5.7 illustrates various types of the more common formed heads where

= head thickness, inches icr = inside-corner radius, inches sf = straight flange, inches
t r

= radius of dish, inches

=

00
b
a

outside diameter, inches

= depth of dish (inside), inches

Nearly all formed heads are fabricated from a single circular flat plate by spinning, as shown in Fig. 5.1, or by "drawing" with dies in a press, as shown in Fig. 5.6. Detail a of Fig. 5.6 shows a single blank plate being removed from a furnace for forming to a head, and detail b shows the plate in a press during the forming operation.

ID/2 s OA

= = =

inside radius, inches slope of cone, degrees overall dimension, inches

H = diameter of flat spot, inches

CommonTypes of Formed Heads and Their Selection

85

The inside depth of dish and overall dimension, OA, may be determined by use of the dimensional relationships for flanged and dished heads given in Fig. 5.S. For purposes of welding heads to the shells of vessels, various styles of machined edges can be supplied on the formed head by the manufacturer. Standard machining styles for heads supplied by one manufacturer are shown in Fig. 5.9. It should be noted that in styles C6 and C7 the dimension t (the head thickness) must exceed dimension s (the shell thickness) by at least -h in. and in styles DS, D9 and DI0 the dimension t must exceed dimension s by ! in. Table 5.3 gives the cost extras for the various standard machining styles and applies to all types of formed heads. Quantity differentials (1955) must be applied to the cost extras given in Table 5.3 as follows for types A and B only: list plus 90% for 1 to 4 heads, list plus 50% for 11 to 50 heads. All other styles, diameters, and gages are list plus 90%. 5.5a Flanged-only Heads. The formed head most economical to fabricate is that produced by simply forming a flangewith a radius on a flat plate. This head is identified as a "flanged-only head" and is illustrated in detail a of

Fig. 5.7. The radius of a flanged-only head decreases somewhat the abruptness in change of shape at the junction of the flat head and the cylinder. The resulting gradual change in shape reduces local stresses. The flanged-only head finds its widest application in closing the ends of horizontal cylindrical storage vessels at atmospheric pressure. 'These vessels typically store fuel oil, kerosene, and miscellaneous liquids having low vapor pressures. Flanged-only heads may be u~ for the bottom heads of vertical cylindrical vessels that rest on concrete slabs and do not have diameters in excess of 20 ft. Table 5.4 gives the straight-flange length and inside-cornerradius for such heads as functions of head thickness. These heads are fabricated on the basis of using the outside diameter as the nominal diameter. Head diameters based on the outside diameter are available in increments of 2 in. from 12 to 42 in., in increments of 6 in. from 42 to 144 in., and in increments of 12 in. from 144 to 240 in. A head with a 246-in. outside diameter is also available. During spinning of the head'S,thinning out of the plate occurs at the corner radius. Therefore, for heads having an outside diameter of under 150in., plate thicknesses must

f
Fig. 57. Various types of more common formed heads: (a) flanged' only, (b) flanged ond and shallow dished, dished, (e) conical
(c)

00 (a)
~nSide depth of dish

r reJ1
(e)

iF
,

II~i I
OD
(b)

..

flonged dished dished (g)

standard

(d)

ASME ond API-ASME code flonged and (torispherical), (ellipsoidol), flanged ond elliptiea I dished
k-

__l
-+--ID __
(f) (c) Inside depth of dish

(f) hemisphericol, (Courtesy of Lukens

[torlcenlccll ll 05). Steel Co.)

~Jt

sf

~---~OD--~'I

=fl

OA

~----------+-OD--------~
(d)

86

Proportioning

and Head Selection for Cylindrical

Vessels with Formed Closures

Table 5.3. Cost Extras for Standard Machining Styles for Heads (105) (Courtesy of Lukens Steel Company) Outside Diameter Style A 24" and under Gage
1'2
IL"

36" 48" 60"

$2.00 $4.50 $5.50 $7.00 $8.50 $10.00 $11.00 $12.00 $14.00 $18.00 $19.00 $21.00 $28.00 $35.00 $2.50 $5.00 $6.50 $8.50 $9.50 $11.50 $12.50 $13.50 $15.00 $20.00 $21.00 $22.50 $30.00 $39.00 $4.00 $6.50 $8.00 $10.50 $12.00 $13.50 $14.50 $15.50 $17.00 $21.50 $22.50 $24.00 $35.00 $46.00

%" $3.50 $5.00 $6.50 $8.50 $9.50 $11.50 $12.50 $13.50 $15.00 $19.50 $20.50 $22.50 $30.00 $37.00 $4.00 $6.00 $7.50 $10.00 $11.00 $12.50 $14.00 $15.00 $17.00 $21.00 $22.50 $24.50 $32.00 $41.00 $5.00 $7.50 $9.50 $12.00 $13.00 $15.00 $17.00 $18.00 $20.00 $25.00 $26.50 $28.00 $40.00 $52.00

I" $4.00 $5.50 $7.00 $9.00 $10.00 $12.00 $13.00 $15.00 $17.00 $20.50 $22.50 $24.,50 $32.00 $39.00 $4.50 $6.50 $8.00 $10.50 $11.50 $13.50 ,$15.00 $17.00 $19.00 $23.00 $25.00 $27.00 $35.00 $43.00 $5.50 $8.00 $10.00 $13.00 $14.00 $16.00 $17 50 $19.00 $21.00 $26.00 $27.50 $29.00 $45.00 $61.00

1H" $5.00 $7.50 $9.00 $11.50 $13.00 $15.00 $17.00 $18.00 $20.00 $21.50 $23.00 $25.00 $35.00 $45.00 $6.00 $8.50 $10.50 $13.50 $14.50 $17.50 $19.00 $20.50 $22.50 $24.00 $25.50 $28.00 $38.00 $48.00 $7.00 $10.50 $13.00 $17.00 $18.00 $20.50 $22.50 $24.00 $25.00 $27.00 $28.50 $30.00 $47.00 $65.00

1H" $5.50 $8.00 $9.50 $12.00 $13.50 $15.50 $17.50 $19.00 $20.50 $22.00 $24.00 $25.50 $37.00 $48.00 $6.50 $9.00 $11.00 $14.00 $15.00 $18.00 $20.00 $21.50 $23.00 $24.50 $26.00 $28.50 $41.00 $53.00 $7.50 $11.00 $13.50 $17.50 $19.00 $21.50 $23.00 $24 .. 50 $26.00 $27.50 $29.00 $31.00 $50.00 $70.00

1%" $9.00 $10.00 $12.50 $14.00 $16.00 $18.00 $19.50 $21.00 $23.00 $24.50 $27.00 $39.00 $51.00

2" $9.50 $10.50 $13.00 $14.50 $16.50 $18.50 $20.00 $22.00 $23.50 $25.50 $28.50 $41.00 $54.00

2H" $11.00 $14.00 $15.50 $17.50 $19.50 $21.00 $23.50

:c-

3" $12.00 $15.00 $16.50 $19.00 $21.00 $22.50 $25.00 $26.50 $29.50 $33.00 $46.00 '$60.00

72" 84" 96" 108" 120" 132" 144" 160" 176" 192"
Style B or C

$25.00 $28.00 $31.00 $43.00 $56.00

24" and under 36" 48" 60" 72" 84" 96" 108" 120" 132" 144" 160" 176" 192" 24" and under 36" 48" 60" 72" 84" 96" 108" 120" 132" 144" 160" 176" 192"
Style D

$10.00 $11.50 $15.00 $16.50 $19.50 $21.50 $23.50 $25.00 $26.50 $29.00 $32.00 $45.00 $58.00

$11.50 $12.50 $16.50 $18.00 $20.50 $23.00 $25.50 $27.00 $28.50 $32.00 $37.00 $50.00 $63.00

$15.00 $20.50 $22.00 $25.50 $28.00 $30.00 $33.00 $36.00 $39.00 $44.00 $57.00 $70.00

$17.00 $23.00 $25.50 $29.00 $33.00 $37.00 $39.50 $42.00 $44.00 $50.00 $62.00 $75.00

$12.50 $14.00 $19.00 $21.00 $23.50 $25.50 $27.00 $29.00 $31.00 $33.00 $35.50 $55.00 $75.00

$15.00 $16.00 $21.00 $22.50 $25.50 $28.00 $30.00 $32.00 $34.50 $37.00 $40.00 $60.00 $80.00

$19.00 $26.00 $29.50 $34.00 $37.00 $40.00 $43.00 $46.50 $49.00 $52.00 $68.00 $85.00

$21.50 $29.50 $35.00 $38.00 $41.50 $45.00 $49.00 $52.00 $55.50 $61.00 $77.00 $93.00

be increased by M 6 in. for plates up to 1 in. in thickness and 78 in. for plates 1 to 2 in. in thickness if the minimum plate
thickness is to be maintained throughout the corners. The manufacturer's catalog should be consulted for greater thicknesses and diameters and for blank weight and forming costs.

5.5b Flanged Standard Dished and Flanged Shallow The pressure rating of a flanged-only head Dished Heads. can be increased if the flat portion is dished. Such heads, not designed to code specifications, are formed from a flat plate into a dished shape consisting of two radii: the" crown" . radius or radius of dish and the inside-corner radius. some-

Common Types of formed Heads and Their Selection Dimensions of Standard flanged-only Heads for All Diameters (Courtesy of Lukens Steel Company) Gage Standard Straight Inside-corner Flange (in.) Radius (in.) (Thickness) Table 5.4.

87

~----------------~OD------------~
b = depth
of dish '

sf

tcr

%s

% %
H

%"s

H6
% ~8

1 1~8 1H 1% 1H 1% 2

1~~-2 1H-2H IH-3 1H-3 1H-3~~ 1~~-3H U~-3H 1~~-3H U~-4 1~~-4 1H-4H l~~-4H 1H-4H 1~~-4H 1H-4H 1~~-4~~

~s %
l~lS

1H 1%"6 1H 1% 2~" 2% 3 3% 3% 4H 4~ 5H 6

,....t=--_-_-a-=--=--=--=-'_D:t---f-----,..j L
r

c
Fig. 5.8. Dimensional relationshi~s for flanged and dished heads. a

= ID _
2 _ (AS)2

b = r _ V(SC)2

times referred to as the "knuckle" radius. If the radius of dish is greater than the shell outside diameter, the head is known as a" flanged and shallow dished head." If the radius of dish is equal to or less than the outside diameter, the head is known as a "flanged and standard dished head." A flanged and shallow dished head is shown in Fig. 5.7, detail b, and a flanged and standard dished head is shown in Fig. 5.7, detail c. These heads are fabricated on the basis of using the outside diameter as the nominal diameter. Head diameters based on the outside diameter are avalable in increments of 2 in. from 12 to 42 in., in increments of 6 in. from 42 to 144 in., and in increments of 12 in. from 144 to 240 in. A 246-in.-outside-diameter head is also available. It should be emphasized that because of the high localized stresses due to the small inside-corner radius, the use of flanged and shallow dished heads and' flanged and standard dished heads is not permitted in vessels which must meet pressure-code requirements. Typical applications of these heads occur in the construe-

ID AS = _ - (ier) 2
BC = r _ (ier) AC = V'--(S-C--;;)2:-_-(A-S-O;)2

OA = I

+ b + sf

tion of vertical process vessels for low pressures, of horizontal cylindrical storage tanks for volatile fluids such as naphtha, gasoline, and kerosene, and of large-diameter storage tanks in which the vapor pressure and hydrostatic pressure is too great for the practical use of flanged-only heads. Vessels with flanged and shallow dished heads are primarily used for horizontal storage tanks. Table 5.5 gives the dimensions of flanged and shallow dished heads. Table 5.6 gives the dimensions of flanged and standard dished heads except for the radius of dish. The radius of dish varies with thickness and diameter, and

2
Style A

3
Style B Fig. 5.9. Standard

6
Style C

9
Style 0

10

machining styles for heads (105).

(Courtesy of Lukens Steel Company.)

88

Proportioning Table 5.5.

and Head Selection

for Cylindrical Vessels with Formed Closures economical to use an elliptical flanged and dished head. These heads are used principally for vessels designed to meet the ASME codes for unfired pressure vessels. In general, these heads are used either for horizontal or vertical vessels fora great variety 'Of process equipment within the pressure ranges specified above. For pressures in the range of 150 lb per sq in. gage and for higher pressures, a cost comparison should be made between the code flanged and dished heads and the code elliptical dished heads. The optimum choice based on total cost varies with pressure, diameter, thickness, and material of construction. Table 5.7gives the inside-corner radius and radius of dish for code flanged and dished heads. Table 5.8 gives the straight-flange length for different head thicknesses of flanged and dished heads. For purposes of cost estimation it is necessary to know"the blank weight in order to obtain the cost of the steel used and the cost of forming the head at the fabrication plant. The approximate blank diameter may be determined by use of the following relationships: diameter
=

Dimensions of Flanged and Shallow Dished Heads in Inches (See Fig. 5.7.) (Courtesy of Buffalo Tank Company)

OD Gage

~4: ?-i6

lcr sf 3~~ ~~ % 3~~

%

316 H
OD Gage

}s

66 r 120 120 H~ 120 120 6 120 6 108

72 r 120 120 120 120 120 ll4 r 197 197 197 197 197

76 r 120 120 120 120 120 120 r 300 197 300 197 300 197 300 197 300 197

84 r 120 120 120 120 120 126 r 300 197 300 197 300 197 300 197

90 r 197 197 197 197 197 132 r 300 300 300 300 300

96 r 197 197 197 197 197 138 r 300 300 300 300 300

102 r 197 197 197 197 197 144 r

1300

~4:

icr sf ~~ 3H %
%

r 197 197 197 197 197

~6
%

3~~ 4~~ 6 6

300 300 300 300

316
~~ %

% 1 1~(

OD

+ 42 + 2sf + -licr
(for gages under 1 in.)

OD

(5.12)

. diameter

= OD

OD. + 24 + 2sf + itcr + t (for gages 1 in. and over)

the manufacturer's catalog should be consulted for this dimension, blank weight, and forming costs. 5.5c Flanged and Dished Heads (Torispherical) to ASME Code. The pressure rating of flanged and dished heads can be increased by decreasing the local stresses which occur in the inside corner of the head. This may be accomplished by forming the head so that the inside-corner radius is made at least equal to three times the metal thickness; for code construction, the radius should in no case he less than 6 % of the inside diameter. Also, the radius of dish may be made equal to or less than the diameter of the head. Figure 5.7,detail d shows a sketch of a cross section of a flanged and dished head meeting the ASME Code, in which is identified as a "torispherical" head. These heads are fabricated on the basis of using the outside diameter as the nominal diameter. Head diameters based on the outside diameter are available in increments of 2 in. from 12 to 42 in., in increments of 6 in. from 42 to 144 in., and in increments of 12 in. from 144 to 240 in. Heads having outside diameter of 210 in. and 246 in. are also available. The volume in .cubic feet of heads having icr equal to 6% of the outside diameter (not including the straightflange portion) is approximately equal to:

(5.13)

where OD
sf

icr = inside-corner radius, inches

= outside diameter of dish, inches

straight-flange length, inches

= gage thickness, inches

Table 5.6.

Dimensions< of Flanged and Standard Dished Heads (Courtesy of Lukens Steel Company) Standard Straight . Flange (in.)
sf

Thickness

(in.)

Inside-corner Radius (in.)

icr
~6

t l!16
~6

~~
%

~~ % %
1

316

}s

v
where d, V

= 0.000049d;s

(5.ll)

1H 1H 1% 1%
2

inside diameter of vessel, inches volume of torispherical dished head to straight flange, cubic feet

H~

1%

I%.

Heads of this type are used for pressure vessels in the general range of from 15 to about 200 lb per sq in. gage. These heads may be used for higher pressures; however, for pressures over 200 lb per sq in. gage it may be more

2H 2~2 2%

1%-2 1H-2~~ H~-3 U2-3 1%-3~'2 U'2-3% 1%-3>'2 1%-3~'2 H'2-4 1%-4 H'2-4~'2 H'2-4~'2 U2-4% U2-4~'2 1%-4~'2 H~-4~'2 1%-4~'2 1%-4~~ 1%-4~2' 1%-4~'2 H'2-4H U2-4~'2

%
1~6

1%6 U2 1% 2H 2% 3%
3%,

1>8

4>8
H2 4}s 5~4: 5%

6 6%, 9

7H 8>4: