Chapter 23 pp177 196 Revised

Document Sample
Chapter 23 pp177 196 Revised Powered By Docstoc
					CHAPTER


23

          SECTION VIII, DIVISION 3—
ALTERNATIVE RULES FOR CONSTRUCTION
          OF HIGH-PRESSURE VESSELS



J. Robert Sims, Jr.
23.1      INTRODUCTION
   This chapter provides a commentary on the ASME Boiler and         Formatted: Right: 3.99"
Pressure Vessel Code, Section VIII, Division 3. It is intended to
be used as a companion to Division 3 by Manufacturers and Users
of high-pressure vessels, and will also provide guidance to
Inspectors, materials suppliers, and others. The material is
generally presented in the same order in which it appears in the
Code. Comments are not given about each paragraph, but
paragraph numbers are referenced as appropriate. The comments
apply to the 2004 2007 edition.
   Division 3 was developed by the ASME Special Working
Group (SWG) on High Pressure Vessels of Subcommittee VIII
(currently the Subgroup on High Pressure Vessels).
Subcommittee VIII is responsible for developing and maintaining
all of the Boiler and Pressure Vessel Section VIII Codes,
including Divisions 1, 2, and 3. Recommendations from the SWG
SG-HPV and Subcommittee VIII are submitted to the Boiler and
Pressure Vessel Committee, which is the “official” ANSI
Standards Committee, for ballot. In a re-organization that is
scheduled to take place in 2009, Subcommittee VIII will become
a standards committee.
   The Code permits the use of either U.S. Customary or SI units
at the User’s option, but one system or the other must should be
used consistently for all phases of construction. However, the
Code permits exceptions based on practical considerations. In the
text of Division 3, the U.S. Customary unit is shown first,
followed by the SI unit in parenthesis. This article follows that
precedent in the text, but the figures have been presented only in
SI units to avoid the clutter associated with dual units.
   The comments herein are the personal opinions of the author.
They should not be considered interpretations of Division 3 or
representations of the opinions of the Special Working Gubgroup
on High Pressure Vessels or any other ASME Committee.


23.2      FOREWORD AND POLICY STATEMENTS
  The Foreword and Policy Statements at the beginning of             Formatted: Right: 3.99"
Division 3 are the same as those in the other Sections of the
Boiler and Pressure Vessel Code. Note particularly the statement
that “It is not intended that this section be used as a design
                                                                         Paragraph KG-102 limits the scope of application of Division 3
                                                                         only to pressure containers for fluids. There are no size- or
J. Robert Sims, Jr.                                                      application-specific limitations such as are found in Division 1,
                                                                         except that the application of Division 3 for direct fired vessels is
                                                                         limited to those applications outside of the scope of the BPV,
handbook; rather, engineering judgment must be employed in the           Section I. However, the User is cautioned that individual legal
selection of those sets of Code rules suitable to any specific           jurisdictions may impose additional restrictions on the application
service or need.” This is reinforced later by the following              of Division 3.
statement: “The Code is not a handbook and cannot replace
                                                                                                                                                 Formatted: Font: (Default) IGADL B+
education, experience, and use of engineering judgement.”                                                                                        Helvetica, 12 pt, Font color: Black
Although this chapter provides a commentary that is intended to             Division 3 can be applied to stationary vessels in fixed
aid individuals involved in the construction of high-pressure            locations, to vessels that are re-located from site to site between     Formatted: Default, Left, Indent: First line:
vessels, it cannot be a substitute for experience and judgment.          pressurizations and to vessels installed in transport vehicles, but     0.13"
                                                                         operation and maintenance control shall be retained during the
                                                                                                                                                 Formatted: Indent: First line: 0.13", Space
                                                                         useful life of the pressure vessel by the User who prepares the         After: 0 pt
                                                                         User’s Design Specification.
23.3      PART KG — GENERAL                                                 Paragraph KG-110 and its dependent paragraphs KG-111
                                                                         through KG-117 describe the geometric limits of Division 3,
          REQUIREMENTS                                                   which are generally similar to Divisions 1 and 2. It is a common
23.3.1    Article KG-1 — Scope and Jurisdiction                          practice to construct the jacket of a high-pressure vessel to
                                                                         Division 1. In this case, the Division 1 rules apply after the
   Article KG-1, paragraph KG-101, describes the scope of                welding pad or first weld of the jacket. Welds directly to the shell
application of Division 3. Division 3 is limited to metallic vessels     of a Division 3 vessel must meet all of the Division 3
except as provided in Code Case 2390-3, which is titled                  requirements. This is done because a crack in an attachment weld
Composite Reinforced Pressure Vessels. and Code Case 2579-1,             could propagate into the pressure boundary. In the case of a
which is titled Composite Reinforced Pressure Vessels for                welding-end connection, the geometric scope of Division 3 ends
Gaseous H2 Service. It is anticipated that this these Code Cases         at the weld prep. This weld prep could be machined as an integral
                                                                                                                                                 Formatted: Subscript
will be incorporated into Division 3 at a future date. The scope of      part of the shell of a vessel such that a crack in the weld could
Division 3 does not contain a lower limit on the design pressure,        propagate into the vessel shell. However, it is anticipated that this
but notes that it is generally intended to be applied for design         weld will be made using the rules in the High Pressure Piping
pressures above 10,000 psi (70 MPa). This limit is subjective            Code, ASME B31.3, Chapter IX, or another Code with
since it is possible to build a satisfactory vessel using the rules in   appropriate welding requirements.
Section VIII, Division 2, for design pressures up totowel above
2010,000 psi (140 70 MPa). Laboratory scale vessels have been              23.3.1.3 Limitations on Use of Materials Paragraph KG-142
built to Division 2 for design pressures over 30,000 psi (210            requires that standard pressure parts such as flanges and fittings
MPa). However, many of the rules in Divisions 1 and 2 are based          be made from materials listed in Division 3. It is essential that
on thin-shell theory and therefore do not provide an optimum             materials have good ductility and excellent toughness because of
vessel design for the thick shells that are typically required for       the higher stresses permitted in Division 3 construction compared
vessels with a design pressure over ~10,000 psi (70 MPa). The            to Divisions 1 or 2. Therefore, numerous restrictions on the
thin shell theory that forms the basis for many Division 1 and 2         application of materials are found throughout the document.
rules gives increasingly inaccurate results as the diameter ratio is
increased. Division 3 requires that elastic-plastic analysis be used        23.3.1.4 Use of U.S. Customary and SI Units The User is
for cylindrical and spherical shells that have diameter ratios above     required to specify either U.S. Customary or SI units in the User’s
1.5 because the application of the linear-elastic stress analysis        Design Specification (see 23.3.3.1). One of these systems must
rules, which are similar to those in Division 2, could give a            should be used for all phases of construction. Mixing of units is
significantly non-conservative result for thicker shells. This is one    not recommended because of the potential for confusion, but the
of the reasons that Division 1 states that “deviations from and          use of alternative units is permitted to comply with local custom
additions to these rules usually are necessary to meet the               at the installation site or to accommodate components that are
requirements of design principles and construction practices” for        customarily specified in a different system of units. However, all
design pressures above 3,000 psi (21 MPa). Therefore, the author         variables must be expressed in a single system of units within an
recommends that Division 3 be used for the construction of               individual equation. In addition, the specific units to be used in
vessels with thick shells. In addition, the author recommends that       equations are specified in Mandatory Appendix 7 or in the text.
shell elements not be used for vessels with a diameter ratio
greater than about 1.25 because shell elements will not accurately       23.3.2 Article KG-2 — Organization of Division 3
characterize the non-linear through-thickness stress distribution           Article KG-2 describes the organization of Division 3. There
that is characteristic of thick shells.                                  are eight parts, representing eight phases of construction. In
                                                                         addition, there are six seven Mandatory and eight ten Non-
   23.3.1.1 Thick-versus Thin-Shell Theories Except for the              Mandatory Appendices. In general, the technical requirements in
linear-elastic stress analysis approach in KD-220 240 through            Division 3 are included in the eight parts (main body of the
KD-235247, the rules in Division 3 are based on thick shell              Code). This differs from Division 21, which has significant
theories. These theories also work well for thin shells, so Division     technical requirements in Mandatory Appendices. Requirements
3 could be used for very low pressure, thin shell vessels.               within the eight parts are performance-based in some cases, but
However, this will not be economically feasible in most cases,           provide detailed (prescriptive) methodology in others as
because of the more stringent requirements for material                  appropriate. Several of the Non-Mnonmandatory aAppendices
toughness, fabrication and examination in Division 3.                    contain detailed technical approaches that can be used, but the
                                                                         designer is free to use other approaches if the performance-based
  23.3.1.2 Geometric and Application Limits in the Scope                 requirements are met. The parts are:
         (1) KG — General Requirements
         (2) KM — Material Requirements
         (3) KD — Design Requirements
                                                                  Non-Mandatory Appendices are:
         (4) KF — Fabrication Requirements
         (5) KR — Pressure Relief Devices                                   (1) A — Guide for Preparing Manufacturer’s Data
         (6) KE — Examination Requirements                        Reports
         (7) KT — Testing Requirements                                      (2) B — Requalification
         (8) KS — Marking, Stamping, Reports, and Records                   (3) C — Guide to Information Appearing on Certificate
                                                                  of Authorization
Mandatory Appendices are:                                                   (4) D — Fracture Mechanics Calculations
                                                                            (5) E—Construction Details
         (1) Nomenclature
                                                                            (6) F — Approval of New Materials Under the ASME
         (2) Quality Control System
                                                                  Boiler and Pressure Vessel Code
         (3) Submittal of Technical Inquiries to the Boiler and
                                                                            (7) G — Design Rules for Clamp Connections
Pressure Vessel Committee
                                                                            (8) H — Openings and Their Reinforcement
         (4) Acceptance of Testing Laboratories and Authorized
                                                                            (9) I — Guidance for the Use of U.S. Customary and SI
Observers for Capacity Certification of Pressure Relief Devices
                                                                  Units in the ASME Boiler and Pressure Vessel Code.
         (5) Adhesive Attachment of Nameplates
                                                                            (10) J – Stress Concentration Factors For Crossbores In   Formatted: Bullets and Numbering
         (6) Rounded Indications Charts Acceptance Standard for
                                                                  Closed-End Cylinders And Square Blocks
Radiographically Determined Rounded Indications in Welds
         (7) Standard Units for Use in Equations.
23.3.3 Article KG-3 — Responsibilities and Duties                     paragraph KG-311):
  Article KG-3 describes the responsibilities and duties of the                 (1) Division 3 requires significantly more detailed
User, Manufacturer, and Inspector.                                    information than Division 2. In particular, a detailed description
   23.3.3.1 User’s Design Specification As in Division 2, the         of the contained fluid is required.
User or his Designated Agent is required to prepare the User’s                  (2) The Minimum Design Metal Temperature (MDMT)
Design Specification (paragraph KG-310). However, the                 is defined differently than that in Divisions 1 or 2. The User must
information that must be provided is much more extensive than         specify the lowest (i.e., coldest) metal temperature that
that required by Division 2. In preparing the Design Specification,
the User should particularly note the following information (see
     can be anticipated when the primary stress at any location in          before-burst manner, the Manufacturer or designer can then
     the vessel exceeds 6 ksi (41.4 MPa). This is based on the              accept this rather than perform a fracture mechanics analysis.
     assumption that the lowest stress that will result in a running                  (4) Paragraph KG-311.11 places responsibility for the
     brittle fracture in the most commonly used ferritic alloys is          design, construction and installation of the overpressure
     generally in the range of 6–8 ksi (41–55 MPa), and is                  protection system on the User. This is typically what is done with
     consistent with the ASME B31.3 Process Piping Code.                    both low and high-pressure vessels in the process industries, so it
     Divisions 1 and 2 elected to use a percentage of the allowable         has been recognized explicitly in Division 3.
     stress rather than an absolute value. The selected percentages,                  (5) Paragraph 311.12 describes some additional
     35% for Division 1 and 30% for Division 2, result in a stress          requirements for the User’s Design Specification. Comments on
     in the 6–8 ksi (41–55 MPa) range for most of the materials             some of these are as follows:
     listed. However, Division 3 lists materials that have an                     Laminations in plate materials can result from the steel
     “effective allowable stress” over three times higher than              making process. These discontinuities are a concern primarily
     Division 2 1 in some cases. Since the resistance to brittle            when the plate is subjected to loads that result in bending stresses
     fracture does not increase as the material strength increases,         or if hydrogen charging due to the service environment is
     the absolute limit was selected.                                       possible. Laminations can also exist in forgings, but they are
            (3) The User must state whether leak-before-burst               much less common. If the User believes that laminations may be
(LBB) behavior has been demonstrated by experience with                     harmful, he or she should specify additional inspection to detect
similar vessels. Leak-before-burst behavior describes a failure             them. The typical technique is straight beam ultrasonic
mode in which a fatigue or environmental crack propagates                   examination to SA-435 for plate and SA-388 for forgings.
completely through the thickness of a component, resulting in a                   As noted previously, the User is required to specify the
leak, before the crack reaches a critical size and initiates a fast         system of units to be used (U.S. Customary or SI).
fracture (burst). If the vessel is expected to exhibit leak-before-
burst behavior, a “traditional” S-N fatigue analysis can be                 1         The User is also required to state any additional
performed in lieu of a more complex fracture mechanics analysis.            requirements for seals and bolting for closures. The details of the
Calculations to determine whether leak-before-burst is predicted            design of seals are not included in Division 3.
can be done using fracture mechanics. Fracture mechanics                    2         Requirements for pressure testing that are not
analysis in Division 3 uses a linear-elastic approach, except that          specifically addressed in Division 3 should be specified. For
Code Cases 2390-3 and 2579-1 require the more robust approach               example, if the User wants the vessel to be cleaned and dried, this
in API-579-1/ASME FFS-1, which uses a failure assessment                    should be stated.
diagram (FAD) to determine the combination of crack tip stress              3         Division 3 requires that the Manufacturer furnish an
intensity and reference stress in the remaining ligament that will          extensive list of records to the User (see Division 3 paragraph
result in fracture. The SG-HPV is currently considering                     KS-320). If the User wishes additional records or reports, this
incorporating this approach into the Code for all fracfure                  should be stated.
mechanics analyses. Although both of theseis approaches
produces good results for the calculation of fatigue crack growth,            (6) For vessels constructed to the 2007 Edition, Aa Registered
it they can be excessively conservative when it is used to predict                Professional Engineer registered in one or more of the
the critical crack size. In particular, high fluid pressure acting                states of the United States of America or the provinces of
within a crack has been demonstrated to reduce constraint by                      Canada and experienced in high-pressure vessel design
promoting crack tip yielding. This will significantly increase the                must certify the User’s Design Specification (UDS).
critical crack size, but there is insufficient basis in the literature to         However, starting with the 2008 Addenda, engineers with
quantify the effect. Therefore, Division 3 permits the User to                    equivalent certification in countries outside of the United
document actual field experience with fatigue failures that have                  States or Canada may certify the UDS.
occurred in similar vessels. If the failures occurred in a leak-
23.3.4    Article KG-4 — General Rules for Inspection               subcontractors who possess a valid U or U2 Certificate may
   Article KG-4 describes the rules for inspection, the             perform welding (Paragraph KG-420c).
Manufacturer’s responsibilities, and the Inspector’s duties. Note
that while the Manufacturer must have a valid U3 Certificate,
23.4       PART     KM                    —                           application, the outside diameter will be over 18 times the inside
           MATERIAL                                                   diameter and the weight will be about 507,000 kg per meter
                                                                      (341,000 lbs. per foot) of vessel length. Doubling the yield
           REQUIREMENTS                                               strength to 414 MPa (60,000 psi) will reduce the outside diameter
   It is necessary to use high strength materials for high-pressure   by a factor of about 4.3 and the weight by a factor of almost 20.
vessels. The reason for this is demonstrated in the following         Doubling the yield strength again to 818 MPa (120,000 psi) will
example: A Manufacturer has been requested to build a                 reduce the outside diameter by an additional factor of about 2.1
cylindrical vessel with a 19.7 in. (500 mm) inside diameter for a     and the weight by a factor of about
design pressure of 58,000 psi (400 MPa). Table 23.1 and Figs.         5.3. As a general rule of thumb, it is desirable to limit the design
                                                                                         1

23.1 and                                                              pressure to about 2 of the yield strength of the material, because
23.2 show the required wall thickness as a function of the yield      the weight will increase rapidly at higher design pressures.
strength of the material using the minimum wall thickness
equation in paragraph KD-251.1.                                                                                                              Formatted: Font: (Default) IGADL B+
   It can be seen that if the manufacturer attempts to use a                                                                                 Helvetica, 12 pt, Font color: Black
material with a yield strength of 207 MPa (30,000 psi) for this
                                                                                                                                             Formatted: Default, Left, Line spacing: single
                                                                            TABLE 23.1 CYLINDRICAL SHELL WEIGHT AS
                                                                            A FUNCTION OF MATERIAL YIELD STRENGTH



                                                                      Yield          Required       Outside                  Weight
                                                                      Strength       Diameter       Diameter         (kg per meter)
                                                                      (MPa psi)       Ratio       (mm inches)         (lbs. per foot)
                                                                      20730,000        18.2         9,090358       507,000341,000
                                                                      27640,000        8.81         4,400173        118,00079,100
                                                                      41460,000        4.26         2,13084.0        26,40017,800
                                                                      55280,000        2.97         1,48058.4         12,0008,070
                                                                      689100,000       2.39         1,19047.0           7,2304,860
                                                                      827120,000       2.07         1,03040.7           5,0203,370
                                                                      965140,000       1.86          93036.7            3,7902,550




                                                                      range of yield strength from ~690 MPa (100,000 psi) to ~1,240
   Since the stress level in high-pressure vessels is of necessity    MPa (180,000 psi). However, the higher yield strength grades do
much higher than it is in lower pressure vessels, the toughness of    not have sufficient toughness to be used for the primary pressure
the material used must also be much higher to provide a reason-       boundary. There are many other high-strength materials available,
able level of flaw tolerance. The toughness requirements in           but it is difficult to achieve the necessary combination of high
Division 3, which will be discussed in more detail in the follow-     strength and high toughness if the yield strength is above about
ing paragraphs, effectively place an upper limit on the strength of   965 MPa (140,000 psi). In addition, the susceptibility of the
the material, since material toughness tends to decrease as the       material to environmental cracking in water and moist air
strength increases.                                                   environments is a major concern as the yield strength increases
   The most common material for monoblock and shrink-fit              above about 830 MPa (120,000 psi).
layered high-pressure vessels is SA-723. This high-strength, low-        Charpy V-notch toughness testing is required for essentially all
alloy material is available in several classes and grades with a      materials that will be used in a Division 3 vessel. In addition,
toughness tests are required for welding procedure qualifications   absorbed are significantly higher than in Divisions 1 or 2.
and production weld tests. The acceptance criteria for the energy
23.4.1     Article   KM-1       —               General               that the toughness criteria be met in the minimum property
           Requirements for Materials                                 direction. Longitudinal specimens are permitted only if the
   Article KM-1 defines the materials that are permitted for          geometry of the material makes it impracticable to obtain
Division 3 construction. Three levels of requirements are includ-     transverse specimens. If longitudinal specimens are used, the
ed, although they are not explicitly called levels in the Article.    minimum required value of energy absorbed is about twice that
The highest level of requirements includes all primary pressure       required for transverse specimens. The notch in the transverse
boundary materials; all of the rules of Division 3 must be met for    specimen may be oriented so that the crack will propagate in
these materials. The lowest level of requirements covers integral     either the through-thickness or longitudinal directions.
cladding and weld overlay materials that may be of any weldable
quality material. Also at the lowest level are protective liner       23.4.3     Article   KM-3         —             Supplementary
materials that may be any metallic or nonmetallic material that is               Requirements for Bolting
suitable for the service conditions. The intermediate level             Article KM-3 gives special requirements for bolting. As is the
includes inner layers of layered, wire- or strip-wound vessels. The   case for other materials in Division 3, the stresses in bolting are
requirements for inner layers are discussed under Article KD-8.       permitted to be significantly higher than in Divisions 1 and 2.
                                                                      Therefore, some additional requirements for bolting are needed.
23.4.2     Article   KM-2     —   Mechanical
           Property Test Requirements                                 23.4.4     Article KM-4 — Material Design Data
  Article KM-2 defines the testing requirements. The primary            Article KM-4 lists permitted materials and provides references
emphasis is on Charpy V-notch testing, since the other test           to the tables of material properties in the Boiler and Pressure
requirements are adequately covered in the Material                   Vessel Code Section II, Part D. A few lower-strength materials,
Specifications given in Section II. The orientation of the Charpy     such as carbon steels and annealed austenitic and nickel alloys,
specimen is important because most product forms exhibit much         are listed, along with the high-strength, low-alloy steels that are
higher toughness if the major axis of the specimen is oriented        used for most high-pressure vessel applications.
parallel to the direction of major working (longitudinal
specimen). The intention of the rules in Article KM-2 is to require
23.5      PART   KD    —                                            the material yields in tension during one portion of the load cycle
          DESIGN                                                    and then yields in compression during another portion. Therefore,
                                                                    many of the Division 3 design rules are based on elastic–plastic
          REQUIREMENTS                                              approaches.
   The most significant differences between Division 3 and             Because of the high stresses in Division 3 vessels, fatigue is a
Divisions 1 and 2 are in the design requirements, although the      more common mode of failure than in typical Division 1 or 2
complete of Division 2 that was published in 2007 brought it into   vessels. Extensive experience has shown that fatigue performance
much closer alignment with Division 3. A typical high-pressure      can be improved significantly if a favorable compressive mean
vessel will have a significant amount of plastic deformation        stress is introduced at locations of high cyclic stresses. It is com-
during pressure testing. In some cases, large areas of the vessel   mon practice to apply a process known as autofrettage to create
may not shake down to fully elastic behavior during operation.      the compressive mean stress. This process, which will be dis-
For example, the material of the entire bore surface of a very      cussed in a later section, often requires exposing the vessel to a
thick wall vessel may exhibit a stable hysteresis loop, in which    pressure significantly higher than the normal hydrotest pressure.
23.5.1     Article KD-1—General Design                                  design metal temperature (paragraph KD-112) is given a
           Requirements                                                 somewhat more detailed treatment in Division 3 than in Divisions
   23.5.1.1 Fatigue Analysis Required Paragraph KD-100                  1 and 2. The Division 1 and 2 concept of using the maximum
emphasizes that a fatigue analysis is required in all cases except      average through thickness metal temperature as the design
for small laboratory scale vessels that are to be operated with         temperature to establish the yield strength to be used in the static
supplementary protective devices, such as a barricade or                strength analysis is retrained for simplicity. However, if a large
containment. Because fatigue is one of the most common failure          temperature gradient is anticipated, with which results in a
modes for high-pressure vessels, the designer should use                significant variation in the yield strength of the material through
particular care in the design to avoid areas of local stress            the thickness, it would be prudent to consider the effect of this on
concentration, such as notches and other discontinuities.               the static strength of the vessel.
                                                                           Paragraph KD-112 does require that the maximum temperature
   23.5.1.2 Protective Inner Liners and Prestressed Inner               at any location be limited to the maximum value listed in the
Layers Paragraphs KD-103 and KD-800 provide additional                  yield strength tables. This was done because the upper
guidance on protective liners and prestressed inner layers. As          temperature limit for some materials is based on metallurgical
described in paragraph KD-103, a protective liner provides a            considerations, such as temper embrittlement. In addition, since
barrier to chemical and mechanical damage. It may be made from          rules for time-dependent behavior have not been included in
any material — either metallic or nonmetallic — but is not con-         Division 3, it is prudent to maintain the temperature at all
sidered in the static strength calculations. However, Division 3        locations within the vessel below the value at which time-
does require that the liner be considered when calculating the          dependent properties would control the design.
stresses in the rest of the vessel for the fatigue and fracture            For purposes of fracture mechanics analysis, the minimum
mechanics analyses.                                                     metal temperature should be determined at the postulated crack
   On the other hand, prestressed inner layers, as described in         tip location. The MDMT that is stamped on the nameplate is the
paragraph KD-800, are considered to be an integral part of the          lowest temperature at any location in the vessel. Therefore, it is
vessel and are considered in all phases of the design analysis.         possible for the crack tip to be located in an area that will be
They are exempt from the more stringent Charpy impact values            warmer than the MDMT.
required in Part KM, provided that the values required by the              Paragraph KD-113 contains a special provision for upset condi-
appropriate Material Specifications in Section II are met and that      tions. In some cases, for example the production of polyethylene,
fracture of the inner liner will not result in rupture of the vessel.   a runaway reaction can cause overheating of the inner surface of a
To meet this requirement, it is necessary to show that the outer        vessel. The duration of the overtemperature condition is typically
layers will exhibit a leak-before-burst mode of failure as              very short because protective devices act to depressure the
described in paragraphs KD-141 and KD-810, and that they will           reactor, thus reducing the process fluid temperature. The portion
not fail by plastic collapse. The intent is to contain the fragments    of the wall thickness that is heated above the maximum
of the inner layers if they should fail in a brittle manner. Since      temperature listed in the yield strength tables must be ignored for
these layers are in compression, brittle fracture is not likely, but    the calculation of the static strength of the vessel. This effectively
the exemption from the impact test rules is granted only if the         increases the inside diameter and reduces the wall thickness for
fragments will be contained.                                            the purpose of the minimum wall thickness calculations. The
                                                                        effects of the upset on fatigue and fracture must be considered.
   23.5.1.3 Design Loads Paragraph KD-110 describes the loads           For example, if the hot material at the inside surface yields, the
that must be considered in the design of the vessel. The intent is      favorable compressive mean stress from autofrettage can be
to require consideration of all loadings that cause significant         replaced by a tensile mean stress. In addition, the high local strain
stress in the vessel. In contrast to Divisions 1 and 2, residual        range can result in crack initiation.
stresses are considered to be loadings. (Residual stresses are             If the runaway reaction upset is considered to be a remote pos-
discussed in more detail in paragraph KD-132.) Although residual        sibility, it may be reasonable to do the design analysis, including
stresses do not affect the static strength of a vessel, they do         the fatigue life calculations, without considering this event. In this
change the mean stress and therefore can have a significant effect      case, if the event does occur in service, it would be necessary to
on the fatigue life. These mean stresses are considered in the          develop a fitness-for-service plan, which may include examina-
fatigue and fracture mechanics calculations in Division 3. It is        tion of the affected areas for cracks, re-autofrettage, or reanalysis
common practice in high-pressure vessel design to use shrink            based on the current and future predicted operating conditions.
fitting, strip or wire winding, or autofrettage to induce residual
                                                                           23.5.1.5 Failure Modes Addressed in Division 3 The
compressive stresses on the inside surfaces of the vessel. Since
                                                                        following failure modes are explicitly addressed in Division 3:
these surfaces are typically exposed to the highest range of stress
in operation, the favorable compressive mean stress can give a                     (1) Plastic collapse or through-thickness yielding. The
significant increase in the fatigue life.                               actual burst pressure of a vessel made from ductile, high-tough-
   High residual tensile stresses are detrimental if environmental      ness material is higher than the pressure that causes yielding
cracking is a concern. The shrink fit and autofrettage processes        throughout the wall thickness. However, the through-thickness
typically introduce tensile residual stress on the outer surfaces to    yield pressure was selected as the failure pressure in a static
balance the favorable compressive stress at the bore. Although          analysis for simplicity. Calculation of the true failure pressure
Division 3 does not provide detailed design rules to prevent stress     requires knowledge of the detailed stress–strain curve of the
corrosion cracking or other forms of environmental degradation,         material and exposure of a vessel to pressures above the through-
these must be considered by the designer as required in paragraph       thickness yield pressure can result in excessive deformation.
KD-114. Environmental cracking has been the initiating event in                    (2) Local yielding that could produce excessive
several high-pressure vessel failures. In particular, the designer      distortion or leakage. If a linear–elastic analysis is performed, it is
and User should preferably avoid direct contact between aqueous         necessary to limit the through-thickness linearized stress range at
solutions and high-strength, low-alloy steels. If contact is neces-     any location to two times the yield strength of the material to
sary for a specific application, the User should ensure that appro-     ensure that excessive distortion does not occur. If an elastic-
priate water treatment is used.                                         plastic analysis is performed, distortion can be determined
                                                                        directly from the analysis, and limited as necessary for the
  23.5.1.4 Design Temperature Because of the very thick walls           specific application. The “twice yield” criterion is intended to
characteristic of high-pressure vessels, the specification of the       result in shakedown to elastic behavior, such that the linearized
stress will cycle from minus yield (compressive) to plus yield        but can occur in specialty components such as thermowells. There
(tensile) without undergoing additional plastic deformation.          fore, rules for external pressure design are provided.
          (3) Leak caused by through-wall cracking (leak-before-
burst). This is a common failure mode in vessels that have high          23.5.1.6 Theories of Failure for Static Strength Analysis For
cyclic stresses at the inside surface. Crack initiation and           simplicity in performing linear-elastic analyses, the maximum
propagation is typically ductile if the fluid pressure enters the     shear stress theory of failure is used (see paragraph KD-131).
crack at a pressure that represents a significant fraction of the     This theory, also known as the Tresca theory , assumes that
yield strength of the material. The compressive stress from the       yielding will occur at any point in a structure when the difference
fluid pressure combines with the tensile stress from the crack        between the algebraically largest and the algebraically smallest
opening to promote crack tip yielding and blunting, thus reducing     principal stresses equals or exceeds the yield strength of the
the tendency for fast fracture. However, if a crack starts on the     material. The Tresca theory ignores the third (intermediate)
outside surface (e.g., due to stress corrosion), brittle facture is   principal stress. The more accurate, but more complex, Von
more likely.                                                          Mises yield criterion considers all three principal stresses. The
          (4) Fast fracture. Although the minimum acceptable          difference in the applied load to cause yielding between the two
toughness of the materials in Division 3 is much higher than in       approaches varies from 0 to a maximum of about 15%, with the
Divisions 1 and 2, the high stresses and large wall thickness         Tresca theory always giving an equal or more conservative result.
typical of high-pressure vessels make fast fracture a significant     The elastic–plastic finite element analysis programs typically use
concern. A fracture mechanics analysis is required for cases          the Von Mises theory , so this has been adopted for plastic
where leak-before-burst behavior is not predicted.                    analysis in Division 3.
          (5) Buckling. High external pressures are not common,
   Note that the limits on design pressures in cylindrical and        behavior is not predicted. The reason for this requirement is
spherical shells given in paragraph KD-250 are based on a design      described in the next paragraph.
margin of 1.732 on the collapse pressure as determined using             A traditional “S-N” fatigue analysis, as described in Article
elastic-plastic analysis and Von Mises theory. For a closed end       KD-3, is based on the assumption that the fatigue life of complex
cylindrical shell, the corresponding design margin on collapse        structures can be predicted by calculating the highest range of
pressure using the Tresca theory would be 1.5. It should also be      stress at any location, then by determining an allowable number
noted that the collapse pressure of an open ended cylinder, where     of design cycles from laboratory data obtained on strain-cycled,
the pressure end load acting on the closures is carried by an         smooth, polished bars. If there are no significant fabrication flaws
external yoke or frame, is lower than the collapse pressure of a      in the structure, this approach will usually produce a conservative
closed end cylinder by about 15%.                                     result. However, the fatigue life of a smooth, polished bar test
                                                                      specimen is the sum of the number of cycles required to initiate a
   23.5.1.7 Fatigue Evaluation Requirements and Leak-                 crack and the number of cycles to propagate the crack through the
Before-Burst Mode of Failure It is obviously desirable for a          thickness of the specimen. If a metallurgical or fabrication flaw
pressure vessel to have a combination of material toughness and       exists at the high stress location in a pressure vessel structure,
stress distribution that would ensure that a fatigue or environmen-   initiation has already occurred, so the fatigue life of the structure
tal crack can propagate through the wall (leak) without reaching a    will be only the number of cycles required for propagation to
crack tip stress intensity high enough to result in a catastrophic    failure. This could theoretically give a non-conservative fatigue
fast fracture (burst). However, this is not practicable for most      life, so the more accurate fracture mechanics approach was
high-pressure vessels, except in the special case discussed           selected for cases where a catastrophic fast fracture could result.
previously when high fluid pressure acts on the crack tip to          However, the probability that the S-N approach will produce a
promote crack tip yielding and blunting. Therefore, paragraph         non-conservative result in machined, non-welded structures in
KD-140 requires that a fracture mechanics analysis in accordance      practice is very low for the following reasons:
with Article KD-4 be performed in cases where leak-before-burst
          (1) The probability that a crack will be located in the     cyclic stresses exist on the surface, so highly sensitive NDE         Formatted: Font: Times New Roman, 10 pt
region of highest stress is very low.                                 methods such as wet fluorescent magnetic particle (WFMT)
          (2) The number of cycles required to propagate a            and dye penetrant (PT) can be used. It should be noted that,
crack to failure in a pressure vessel structure after initiation is   for the reasons described above, it is possible for the fatigue
almost always greater than the number of cycles required for          life calculated using the fracture mechanics approach (Article
propagation of a crack through the laboratory specimen                KD-4) to be greater than that calculated using the S-N
because                                                               approach (Article KD-3), particularly in low cycle fatigue (e.g.
             (a) the general stress field at the crack tip in a       a design life less than about 10,000 cycles). However, in high        Formatted: Indent: Left: 0.5", No bullets or
          pressure vessel usually drops off as the crack grows        cycle fatigue (e.g. a design life greater than about 100,000          numbering
          beyond the highly stressed surface                          cycles), the fracture mechanics approach will almost always           Formatted: Font: Times New Roman, 10 pt
             (b) the maximum permitted crack size in the              give a shorter life unless a very large stress intensification
                                                                                                                                            Formatted: Bullets and Numbering
          pressure vessel structure is usually greater than the       factor is used with the S-N approach.
          thickness of the laboratory fatigue specimen                   The SG-HPV has an agenda item to consider adopting the
                                                                      “structural stress” method for fatigue analysis of welded
  (c) the S-N approach uses a design margin of 2 on the stress
                                                                      structures that was included in the 2007 Edition of Section
range and 20 on the number of cycles. Since the fracture
                                                                      VIII, Division 2. This method is based on a large database of
mechanics approach is more accurate, a margin of 4 on the
                                                                      fatigue test results on welded structures. It gives results similar
critical crack size or 2 on the number of cycles is required.
                                                                      to those of the fracture mechanics approach.                          Formatted: Font: Times New Roman, 10 pt
Note that in a typical analysis, the margin of
                                                                         Leak-before-burst behavior can be predicted by using
                                                                                                                                            Formatted: Font: Times New Roman, 10 pt
                                                                      fracture mechanics to show that the critical crack size for the
          4 on the critical crack size is equivalent to a margin      initiation of fast fracture is greater than the wall thickness of
          on the stress range well below 2                            the component. Alternatively, since there is experience in the
      (d) the design fatigue curves for high-strength, low-alloy      industry with through-wall cracks in service that exhibited
          steel in Article KD-3 were modified to reflect the          leak-before-burst behavior even though analysis predicted fast
          results of cyclic pressure fatigue tests conducted on       fracture, it is acceptable to apply that experience to predict
          small cylindrical pressure vessels. In some cases, the      leak-before-burst behavior for a new vessel. This experience-
          bore surfaces of these vessels were only roughly            based approach could be used in cases where fluid pressure
          polished and contained small fabrication flaws. In          acting within a crack promotes crack tip ductility as described
          most cases, an examination for flaws was not                previously. Since accounting for this effect is difficult by
          conducted prior to testing.                                 using generally accepted analytical approaches, experience
   A fracture mechanics analysis starts with the assumption           with vessels that have stable though-wall cracks can be used.
that a flaw exists at the high cyclic stress location. The size of    The User should be careful to ensure that the design, size,
this assumed flaw is the largest that reasonably could be             material properties, and operating conditions of the new vessel
present without detection by the nondestructive examination           will be comparable to those of the cracked vessel providing
(NDE) technique that will be used. In most cases, the high            the experience basis.
23.5.2 Article KD-2 — Basic Design Requirements                           following paragraphs will focus on the following two common
  This article provides the rules for a static stress analysis for a      analysis approaches:
high-pressure vessel. Supplementary rules for specific types of             (1) Linear-elastic finite element analysis with stress lineariza
construction and for fatigue analysis are provided in later articles.           tion and categorization. The basics of finite element
   23.5.2.1 Analysis Techniques Division 3 is primarily a design-               analysis will not be discussed here since many excellent
by-analysis Code. There are several approaches to the analysis                  references are available. In a linear-elastic analysis, the
that are permitted as described in the following paragraphs. As an              material is assumed to have a linear relationship between
alternative to analysis, a series of Non-Mandatory Appendices                   stress and strain according to Hooke’s law. Since the
provide design rules for some common geometries. For simple                     material response is linear, the load can be applied in a
structures, it is possible to use these rules, together with classical          single step and the mathematical matrix can be solved
stress analysis techniques, to complete the required static and                 without extensive iteration. However, extensive post-
fatigue analyses without the need for a numerical (e.g., finite                 processing, involving linearization of the stresses through
element) analysis. However, in most cases, it is anticipated that               the thickness of the structure and categorization of the
finite element or other type of numerical analysis (e.g., boundary              stresses into primary, secondary, and peak, as well as
integral) will be performed. Note also that Division 3 places an                membrane and bending, is typically needed to address the
upper limit on the design pressure for cylindrical and spherical                failure modes of concern.
shells in paragraph KD-250. The discussion in the forthcoming




                                                                          Council and SWGSG-HPV members has shown that this
  (2) Elastic-plastic finite element analysis using ideal elastic,        technique can produce a non-conservative result in some cases. In
      perfectly plastic material properties, and large displace-          particular, if stresses significantly above the yield strength of the
      ment theory. In this approach, the load is applied in               material are included in the average primary stress, the result will
      increments. Locations within the structure that reach yield         over-predict the collapse pressure.
      as defined by the Von Mises criterion are permitted to                  As one example, plots of the through-thickness stress
      strain as necessary to satisfy equilibrium without an               distributions for a shell with a diameter ratio of 6.0 are shown in
      increase in stress (perfect plasticity). This method requires       Figs. 23.3 and
      extensive iteration because of the nonlinear material               23.4. This large diameter ratio is not common in high-pressure
      behavior and nonlinear geometry, since the deformation of           vessels, but was selected to illustrate the point. The individual
      the component due to the strain induced in the material is          stress components have been linearized as shown in Fig. 23.3.
      considered. The SG-HPV has an agenda item to consider               The axial or longitudinal stress component is not shown because
      adopting the elastic-plastic analysis methods in the 2007           it is assumed to be constant through the thickness. The actual and
      Edition of Section VIII, Division 2, which uses true                linearized Tresca and Von Mises stress distributions are shown in
      stress/true strain material properties to consider strain           Fig. 23.4. It can be seen that the actual Tresca stress at the bore
      hardening.                                                          surface is about 2.5 times the yield strength of the material.
                                                                          Tresca stresses above yield exist through the initial 10% of the
                                                                          wall thickness. The design pressure determined by averaging the
   23.5.2.2 Linear-Elastic Stress Analysis One approach to                stress through the thickness to obtain a general primary mem-
                                                                                            2

stress analysis in Division 3 is a linear elastic analysis using stress   brane stress of 3 yield is about 1.9 times the yield strength of the
linearization and categorization. Linearization and categorization        material. This is 37% higher than the design pressure determined
of stresses are discussed in the chapter on Section VIII, Division        using the through-thickness yield pressure equation in Division 3
2, so this discussion will not be repeated here. The linear-elastic       with a margin of 1.5. In the case of an open-ended cylinder,
approach used was adopted for consistency with Division 2, but it         where the end closures are supported by external frames, the
has been limited to shells with a diameter ratio (outside diameter        design pressure calculated using the linear-elastic approach is
divided by inside diameter) of 1.5 or less in Division 3. Although        58% higher than the design pressure using the Tresca plastic
linear-elastic analysis has been used extensively for high-pressure       collapse formulation.
vessels in the past, recent work by the Pressure Vessel Research              For a cylinder with a diameter ratio of 2.0, the linear-elastic
calculation for the open-end cylinder is non-conservative by             confined to a very small area in the vicinity of a local stress
about 8%; for the conventional closed-end cylinder, it is                concentration, their effect on the collapse pressure of this region
conservative by about 7%. The non-conservatism for the open-             of the vessel will be small. However, in the vicinity of
end cylinder is reduced to about 3% for a diameter ratio of 1.5.         intersecting bores (openings) or other discontinuities, the
Simple cylinders have been used in the example, but the problem          inclusion of stresses above yield could give a significantly non-
with linear-elastic analysis exists in the general case whenever         conservative result. Therefore, elastic-plastic analyses should be
local stresses above yield exist in a structure. If these stresses are   conducted on high-pressure vessels if possible.




                                                                                    (2) The shape of the stress–strain curve can vary
   One other aspect of linear-elastic analysis that must be              significantly depending on the location and direction of the test
considered is the limit on secondary stresses. As discussed in the       specimen.
Chapter on Division 2, secondary stresses will equilibrate through                  (3) Theat use of the minimum specified yield strength,
the structure and therefore do not contribute to plastic collapse of     with no credit for strain hardening, will givegave a conservative
the vessel. Secondary stresses in Division 3 are limited to two          plastic collapse pressure.
times the minimum specified yield strength of the material. It is                   (4) The predicted plastic strain to be used in the fatigue
assumed that this will result in shakedown to linear-elastic             analysis will also be conservative.
behavior after a few load cycles. This is important to validate the                 (5) The plastic collapse results will be comparable to the
use of linear–elastic analysis for the fatigue calculation, and to       values calculated in paragraph KD-250220, which are also on an
ensure that incremental plastic deformation does not occur on            ideal elastic, perfectly plastic basis. This is an important
each load cycle. Since the effects of local notches and stress           validation step.
concentrations are not considered in the calculation of secondary                   (6) The plastic collapse pressure for the most common
stress, it is recognized that there may still be some cyclic             material, SA-723, will not be excessively conservative because
plasticity in these areas. However, a greater concern is that the        the yield to tensile strength ratio is high (e.g., 0.9).
Bauschinger effect has not been considered. The implications of             However, the work that went into the development of the rules        Formatted: Indent: First line: 0.13"
this are discussed in the section on fatigue analysis.                   for elastic-plastic analysis in Section VIII, Division 2 resulted in
                                                                         true stress-true strain curves for most commonly used materials,
   23.5.2.3 Elastic–Plastic Stress Analysis Paragraph KD-240             and validation of elastic-plastic analysis methodology using those
230 provides rules for elastic-plastic analysis. Although the title      curves. Therefore, the SG-HPV has an agenda item to consider
of this paragraph is “Limit Analysis,” tThe intent of the rules in       adopting the Division 2 rules.
this paragraph is for the designer to perform a finite element
analysis using elastic, perfectly plastic material properties with          23.5.2.4 Strain Limits in Elastic-Plastic Analysis For some
                                                                         geometric shapes, the resistance of a structure to plastic collapse
nonlinear geometry (i.e. large displacement theory) enabled.
Division 3 permits the use of a bilinear stress strain curve that        will increase significantly as the structure deforms under pressure.
will produce a strength that is not more than 5% above the               For example, a flat head will assume a shape similar to a segment
                                                                         of a sphere as it deforms plastically under a pressure load. Plastic
minimum specified yield strength at the maximum strain achieved
in the analysis. The purpose of this is to achieve stability in the      deformation will begin to occur at a relatively low pressure, then
finite element solution. The use of non-linear geometry is               as the head assumes the more favorable hemispherical shape, the
                                                                         bending stress is converted to a membrane stress. The final failure
essential to produce realistic strains and deformations in the
structure. Although using the actual stress–strain curve of the          pressure may be several times the pressure to cause the initial
material rather than perfectly plastic material behavior would           plastic deformation. However, the local plastic strain—for
                                                                         example, at the point where the head is attached to the shell—
produce a more accurate result, the latter was selected when
Division 3 was first published in 1997 for the following reasons:        may be very high. Therefore, it is necessary to limit the strain to a
                                                                         value that the actual material could withstand without microvoid
          (1) The shape of the stress–strain curve is was not            formation and incipient cracking. Experience with cold metal–
available in the Code.                                                   forming operations and standard tensile test specimens has
indicated that a strain limit of 5% will produce a conservative            be toward the lower end.
result in almost all cases. However, it is well known that much               If the sum of the three principal stresses is three times the total
larger strains can be accommodated if the area has high                    equivalent stress at a specific location, the strain limit at that
hydrostatic (triaxial) compression. Conversely, failure can occur          location will be in the general range of 5% to 30%, with ferritic
at zero a very low value of plastic strain in a highly constrained         alloys again being toward the lower end. Very high values of
area with high hydrostatic tension. The Pressure Vessel Research           triaxial tension can produce strain limits well under 1%. If the
Council has initiated a project to developed appropriate strain            sum of the three principal stresses is essentially zero at a specific
limits as a function of the state of triaxial stress (hydrostatic          location, the strain limit at that location will be in the general
stress) that will be considered forwere incorporation incorporated         range of 55% to almost 200%. High values of hydrostatic
into the 2007 Editions of Division 2 and Division 3 when the               compression produce strain limits that can be over 1,000%.
results are available.                                                        Experience with application of the strain limit damage criterion
   As a practical matter, calculation of the strain limit damage as        indicates that even small areas of local stress concentration (i.e.
required by either Division 2 or Division 3 should be done by              notches or sharp grooves) can have high triaxial stresses, resulting
using an application for finite element analysis programs that will        in low strain limits. Since this can limit the maximum calculated
calculate the strain limit damage at each location in the model at         design pressure for the vessel, it is important to avoid details that
each load step in the analysis using the equations in the Code. The        produce high constraint.
application should sum the damage at each location over the                   As a part of the agenda item mentioned earlier in this chapter to
entire load range for comparison with the damage limit in the              adopt the Division 2 elastic-plastic analysis methodology, the
Code. It is important to use “brick” elements for the analysis,            design margin for the strain limit will be reviewed to establish the
since shell elements cannot accurately capture the strains in areas        appropriate level of conservatism.
of local stress and strain concentration. In addition, the shell              It is also necessary for the designer to limit the amount of
element formulation in some FEA programs does not include the              deformation of the structure depending on the application.
radial stress due to pressure acting on the surface in the
calculation of the von Mises stress. The radial pressure stress is            23.4.2.5 External Pressure For very thick cylindrical shells,
an important component when the fluid pressure is a significant            failure under external pressure loads occurs by a plastic collapse
fraction of the yield strength of the material.                            mechanism (i.e., the bore becomes uniformly smaller) rather than
   The strain limit can be very low if the triaxial stress (sum of the     the buckling failure mode that is common for thin shells.                Formatted: Indent: First line: 0.13", Space
three principal stresses) is tensile (negative hydrostatic stress),        Paragraph KD-252 provides an equation to check for buckling in           After: 0 pt
with a magnitude significantly greater than the yield strength of          thinner shells, but it also requires that the maximum external
the material. Conversely, the strain limit can be very high (i.e.          design pressure be determined using the internal pressure
large strains are acceptable) if the triaxial stress is negative (i.e. a   equations. Note that this applies only for the “closed-end
compressive stress field or a positive hydrostatic stress). For            cylinder” case, where the external pressure is applied to the end
example, if the triaxial stress at a specific location in a pressure       of the cylinder. If the cylinder is configured with packing glands
vessel is equal to the total equivalent stress (von Mises stress) at       at both ends, and if the pressure end load is carried by an external
that location, the strain limit at that location will be in the general    yoke or frame, (the “open end cylinder” case), plastic collapse
range of 25% to 75%, depending on the alloy. Austenitic alloys             will occur at a much lower pressure.
will be toward the higher end of the range, and ferritic alloys will
23.5.3 Article KD-3 — Fatigue Evaluation
   This article gives rules for the traditional “S-N approach” for        Because the interaction of these parameters is complex, the
determining the design fatigue life of a high-pressure vessel. The     selection of “fatigue-sensitive” points involves a judgement by
alternative fracture mechanics approach is documented in Article       the designer. As a minimum, points should be selected for
KD-4. As discussed earlier in this chapter, the S-N approach in        analysis based on the following:
Article KD-3 may be used only for vessels that are expected to                    (1) The point or points within the structure that will be
fail in a leak-before-burst mode. Also, in cases where this mode is    exposed to the highest range of strain (stress) during each of the
anticipated, an experimental approach to fatigue may be used           operating or upset cycles. For example, one point may have the
instead of the rules in KD-3, as provided in paragraph KD-1260.        highest strain range during a complete start-up– shutdown cycle
                                                                       that occurs once a month, but a different point may have the
   23.5.3.1 General and Theory Paragraphs KD-301 and KD302             highest strain range due to pressure fluctuations that occur several
give a general discussion of the S-N approach to fatigue eval-         times per second. It is important to determine all of the conditions
uation and briefly summarize the theory used. The approach is          or loadings that result in cyclic strains on the structure as well as
generally similar to that used in the 2004 Edition of Division 2       the highest strain range at any point during each of these
and in paragraph 5.5.3 “Fatigue Assessment – Elastic Stress            conditions. Note that a pre-service hydrotest is a part of the
Analysis and Equivalent Stresses” in the 2007 Edition of Division      fabrication process and is not considered to be an operational
2., except thatHowever, Division 3 includes a separate fatigue         cycle. However, if periodic overpressure testing is contemplated
curve and a mean stress correction method for non-welded               during service, its effects on the fatigue life of the vessel should
components made from high-strength, low alloy steel. Both              be considered. Because of the potential for the initiation and
Divisions 2 and 3 assume that fatigue life is controlled primarily     propagation of fatigue cracks, in-service hydrotests should be
by the range of alternating stress intensity (i.e., range of shear     minimized, but they are occasionally required by some
stress) at a point. The fatigue evaluation process begins with the     jurisdictions. In determining the points at which a detailed
selection of “fatigue-sensitive” points within the structure. The      analysis should be performed, local strain concentration and
following parameters are considered to affect the fatigue life at      surface finish effects should be considered.
each point in the structure in the KD-3 analysis:                                 (2) The point or points within the structure that will be
   (1) Range of local strain. Note that paragraph KD-302 refers to     exposed to the highest mean stress during each of the operating
       alternating stress intensity rather than strain. However, the   cycles (non-welded structures only).
       fatigue data are based on strain-controlled tests, so it is                (3) Intermediate points that may have a lower fatigue
       more accurate theoretically to think in terms of a strain       life than the points selected in items (1) and (2), because of the
       range. Division 3 has followed the lead of Division 2 by        combined effect of several operating cycles and the interaction of
       using stress range throughout the methodology description       the stress range and mean stress. This selection will typically
       and in the fatigue curves for the convenience of the User,      involve considerable experience and engineering judgment. If
       to avoid the step of converting stress from an analysis to      there is any question about a point, it should be selected for
       strain.                                                         analysis.
          (2) Mean stress at the same point. In this case, stress is
the appropriate term. However, the mean stress is used only in the        23.5.3.2 Stress Analysis for Fatigue All service loadings
fatigue evaluation of non-welded structures, because the SWG           should be considered, including any periodic in-service hydrotests
SG-HPV decided to use the precedents in Division 2 for welded          that may be contemplated. In addition, for non-welded
construction. In Divisions 2 and 3, the fatigue curves for welded      construction, residual stresses from fabrication should be
construction consider the maximum or “worst case” effect of            determined to allow calculation of the mean stress during each
mean stress. One reason for this is the difficulty of determining      fatigue cycle. The residual stress from forming operations is
the mean stress distribution in a weldment. In addition, the           difficult to calculate in most cases. However, if heat treatment or
SWGSG-HPV wanted to be consistent with the Division 2                  a mechanical stress relief process such as autofrettage is used, it is
approach for welded structures that hads been used successfully        usually possible to estimate the resulting residual stress with
for many years. However, as mentioned earlier in this chapter, the     sufficient accuracy for the fatigue calculation regardless of the
2007 Edition of Division 2 includes a new “structural stress’          magnitude of residual stress after forming. Procedures for calcu-
method for welded structures that is under consideration for           lating the residual stress in plain cylinders as a result of
inclusion in Division 3.                                               autofrettage are given in Article KD-5. For more complex
          (3) Number of cycles associated with each operating or       geometries, nonlinear finite element analysis can be used to
upset cycle.                                                           predict residual stress.
   In all cases, tThe Bauschinger effect must should be considered      if the residual stresses have been included in the analysis, addi-
in the calculation of residual stress. This effect, which exists to     tional plastic deformation will not occur due to service loads. If
some extent in most materials, is particularly significant in the       high thermal or other stresses do result in stresses above yield, it
SA-723 alloys commonly used for high-pressure vessel                    will be necessary to do an elastic–plastic shakedown analysis to
construction. The Bauschinger effect can be summarized as               predict the effects of plasticity during the first few load cycles.
follows:                                                                Division 3 does permit using Paragraph 23.5.3.3(1) in lieu of the
                                                                        elastic–plastic calculation, since this will always produce a con-
           (1) If a material is stressed beyond its yield strength in
                                                                        servative result.
tension and undergoes plastic deformation, the subsequent yield
                                                                           Although Division 3 requires consideration of the Bauschinger
strength in compression will be reduced. If the material is
                                                                        effect in the calculation of residual stress, it is silent on the need
subsequently loaded in compression such that plastic deformation
occurs, it will “strain harden” back to its original compressive        to do this for the calculation of the alternating stress intensity. It
yield strength. However, the plastic strain required to achieve the     has been common practice to ignore the Bauschinger effect in this
original strength may be very high.                                     calculation, and to use the 0.2% offset yield strength to determine
           (2) The amount of the reduction in compressive yield         whether the range of secondary stress is less than two times the
strength increases with increasing tensile plastic strain, but can be   yield strength to validate the fatigue calculation. However, the
60% or more.                                                            Bauschinger effect can result in significant cyclic plasticity at
           (3) The process also applies in reverse. In other words,     stress ranges well below two times the 0.2% offset yield strength.
the yield strength in tension will be reduced as a result of plastic    This effect is minimized in practice by the design margins and
flow in compression.                                                    other conservatism in the Division 2 and Division 3 approaches.

   For the purpose of calculating residual stress, it is necessary to      23.5.3.4 Calculation of Equivalent Alternating Stress
predict the change in yield strength as well as the shape of the        Intensity The equation in paragraph KD-312.4 is an empirical
resulting stress–strain curve. There are several papers on this         approach for calculating an equivalent alternating stress intensity
topic that can be used.                                                 from the calculated alternating stress intensity and the mean
   The calculation of operating stresses (paragraph KD-311.2) can       stress. This should be determined for each load cycle at each
be done using closed-form solutions for simple geometries, but is       “fatigue sensitive” point.
most commonly done using finite element analysis. Local notches
or areas of stress concentration can be considered in one of two           23.5.3.5 Change of Principal Stress Axes If the directions of
ways:                                                                   the principal stresses change during a load cycle, an iterative
                                                                        process must be used to determine the maximum range of shear
          (1) Modeling the notch in the FEA program. This               stress, as described in paragraph KD-313.
requires a very fine mesh in the vicinity of the notch because of
the steep stress gradients. The mesh density that is necessary to         23.5.3.6 Calculating the Number of Design Cycles The pro-
get an accurate result is a function of the element type and the        cess for calculating the number of design cycles is straightforward
nature of the stress field. For example, Hhigher order multinode        as described in paragraph KD-320. Two alternatives to reading
elements, with nonlinear interpolation, can be larger than 4 node       values from the fatigue curves are the following:
linear elements. However, stress and strain gradients across the
element should be no greater than 10%–20% for reasonable                         (1) using tabulated values with interpolation; and
engineering accuracy.                                                            (2) using closed form curve fit equations
          (2) Using a relatively coarse FEA model or closed form
solutions with strain concentration factors from handbooks. If a           23.5.3.7 Surface Finish Correction Factor Another difference
conservative value for the strain concentration factor will give a      between Division 3 and Division 2 is the surface finish correction
calculated fatigue life that is acceptable for the application, this    factor, or “roughness factor.” This factor corrects the range of
approach will be much less time consuming. Note that strain             stress to allow for the stress concentration effects of a pressure
concentration factors may be larger than stress concentration           vessel surface compared to the polished surface of the smooth bar
factors used with a linear analysis.                                    specimens used to generate the fatigue curves. As of this writing,
                                                                        this factor is independent of the range of stress, but it logically
   In all cases, the maximum values of the three principal stresses     should be higher for high-cycle fatigue than for low-cycle fatigue
should be determined, and then the maximum range of shear               since the roughness of a surface is significant only in the crack
stress calculated as shown in paragraph KD-312.2. A maximum             initiation phase. The SWGSG-HPV has opened an agenda item to
range of shear stress should be determined for each load cycle at       reevaluate the surface roughness correction factor.
each “fatigue-sensitive” point. Up to this point, the methodology
is essentially the same as that in Division 2. However, for non-           23.5.3.8 Combining the Results from Several Load Cycles
welded materials it is necessary to apply a mean stress correction.     The combined effect of several load cycles is determined using
                                                                        “Miner’s Rule” (paragraph KD-330). In this approach, also called
   23.5.3.3 Mean Stress Correction Since the shear stress has no        the “life fraction” approach, each load cycle is assumed to con-
sign, it is not appropriate to calculate a “mean shear stress.”         sume a fraction of the fatigue life of the structure. This fraction is
Therefore, the maximum stress on the plane that is normal to the        calculated as the ratio of the expected number of cycles of the
plane of the maximum shear stress is defined as the mean stress         load under consideration during the life of the vessel to the
for the purpose of the calculation. If the structure shakes down        maximum number of cycles permitted for the equivalent
after the first few load cycles, without incremental deformation,       alternating stress intensity that results from the load.
one of the following two conditions will apply:                            This approach has been successful in Division 2 applications.
          (1) If the material remains within the linear-elastic range   However, there is a possibility that this approach could produce a
during the entire cycle, the mean normal stress is the average of       non-conservative result in a case where a load producing a very
the algebraically minimum and maximum normal stresses.                  high equivalent alternating stress intensity is combined with a
          (2) If the material yields in both tension and                load producing an equivalent alternating stress intensity that gives
compression during each cycle (i.e., a stable hysteresis loop has       a fatigue life in excess of about 100,000 cycles. This occurs
been established), then the mean normal stress is zero.                 because the high equivalent alternating stress intensity may initi-
                                                                        ate a macroscopic fatigue crack within the first few cycles. This
  In most cases, if a high level of autofrettage has been used, and     may represent only a small fraction of the life for that load cycle,
because low-cycle fatigue is dominated by the number of cycles             Therefore, if a designer believes that this could be a problem in
to propagate the crack to failure rather than initiation. Conversely,   a specific application, the fracture mechanics approach in Article
high-cycle fatigue life is dominated by the initiation phase. After     KD-4 or an experimental approach (KD-1260) should be used.
a macroscopic fatigue crack forms, the remaining life to failure        However, since the fracture mechanics approach requires an
may be less than 10% of the total design life. Therefore, if a crack    assumed initial crack size, it may give an excessively
forms during the high-stress cycle, it may reduce the fatigue life      conservative result for loads that cycle more than about 100,000
during the low-stress cycle by more than the life fraction              times during the life of the vessel if the assumed initial crack is
approach would predict.                                                 not present. (See discussion of Article KD-4 below.)
23.5.4     Article KD-4—Fracture Mechanics                                 be determined from Charpy V-notch data using the correlation in
           Evaluation                                                      paragraph KD-600. Correlations are also provided for JIc and
    This article gives rules for a fracture mechanics approach for         CTOD. However, in most casesWhile it is preferable to determine
determining the design fatigue life. A fracture mechanics analysis         KIc directly by testing, experience indicates that use of the Charpy
is required if it cannot be shown that the vessel will fail in a leak-     correlations is satisfactory in most cases.
before-burst mode. Since there are many approaches to the details
of a fracture mechanics analysis, only the essential requirements             23.5.4.2 Initial Crack Size For cases of low-cycle fatigue,
are given in KD-4. Detailed guidance on one approach that can be           where most of the life is consumed by crack propagation, it is
used is given in Nonm-Mandatory Appendix D. A commentary                   anticipated that the initial crack size (paragraph KD-411) will be
on Appendix D is provided later in this chapter. Other                     the maximum acceptable indication from Part KE, Examination
approaches that are commonly used are described in API-579-1               Requirements. In high-pressure vessels, surface flaws are of the
ASME FFS-1, 2007 Edition. For example, the use of API-579-1                most concern, since the steep stress gradients illustrated in Fig.
ASME FFS-1 is required by Code Cases 2390-3, which is titled               23.3 typically result in much higher stresses on the surface. This
Composite Reinforced Pressure Vessels and 2579-1, which is                 is particularly true if there are notches or other areas of local
titled Composite Reinforced Pressure Vessels for Gaseous H2                stress concentration. Therefore, starting crack sizes are typically
Service. The SG-HPV has an agenda item to consider requiring               based on the acceptance criteria for dye penetrant or magnetic
the use of API-579-1 ASME FFS-1.                                           particle examination in KE-232.2. These criteria permit a crack-
                                                                                             3

    The designer should recognize that it is difficult to predict the      like flaw up to 16 in. (4.8 mm) in length for material 2 in. (50 mm)
critical crack size in a high-pressure vessel. Even if a good stress       or greater in thickness. Smaller lengths are permitted for thinner
intensity solution can be found for the complex geometry that              material. Paragraph KD-411(b) requires that a surface crack be
often exists, the effect of biaxial and triaxial loads and the             considered to have an aspect ratio of 1:3, so the maximum depth
                                                                                     1
resulting hydrostatic stress state at the crack tip can be difficult to    will be 16 in.
quantify. For example, if a fatigue crack initiates on an internal         (1.6 mm).
surface, the fluid pressure will act directly on the crack tip. If the        If the vessel will be in high-cycle fatigue service (e.g., greater
pressure is within the same order of magnitude as the yield                than 100,000 significant load cycles), it may be necessary to
strength of the material, the resulting radial compressive stress          specify a smaller initial crack size to obtain an acceptable life
acting at the crack tip will promote crack tip yielding and                from the calculation. If this is done, the following requirements
blunting. This can increase the apparent fracture toughness                apply:
significantly. Conversely, if a vessel has a deep notch in an area
of high bending stress or multiaxial tensile loads, the resulting                    (1) The length to be used must be specified in the User’s
high triaxial or hydrostatic tensile stresses can produce a level of       Design Specification
constraint at the crack tip that is greater than the plane strain                    (2) It must be demonstrated that the NDE method to be
condition.                                                                 used will reliably detect a flaw of the size specified.
    Because of these concerns, it is important that the designer use
experience with vessels that have similar geometry and loads                  For very high-cycle fatigue applications, it may be necessary to
when determining the critical crack size. If a finite element analy-       specify initial crack lengths as small as 0.010 in. A specialized
sis is performed, the hydrostatic stress should be examined at all         examination methodology will be needed to reliability detect
locations where the local stress intensity is high.                        cracks this small. For example, some specialized eddy current
    23.5.4.1 General and Theory Paragraph KD-401 gives an                  techniques have been used. However, it is necessary to apply the
overview of the process, which is based on linear-elastic fracture         specialized examination methodology only in acreas where the
mechanics. In general, the following values are needed as input            highest stresses exist (e.g., in areas of local stress concentration).
for the analysis:                                                          A larger critical crack size can be assumed in other areas where
                                                                           the cyclic stresses are lower.
          (1) Initial or starting crack size.
          (2) Time history of service stresses at each        “fatigue-       23.5.4.3 Stress Calculations To calculate crack growth rates, it      Formatted: Justified
sensitive” location. The stress distribution through the thickness         is necessary to calculate the range of stress at the location of the
along the path of crack propagation will also be needed.                   crack tip. Therefore, the distribution of stress in the direction of
          (3) Residual stresses.                                           crack propagation must be known. The direction of crack
          (4) Critical crack depth. This is defined in Subparagraph        propagation usually can be determined by examining the                   Formatted: Justified
KD- 401 (c) as the crack depth at which the calculated stress              distribution of stress in the through-thickness direction as well as
intensity factor for the crack equals the critical stress intensity, KIc   along the surface. For complex geometries, engineering judgment
for the material. If the MDMT of the vessel is high enough that            and experience must be used.
the material will exhibit “upper shelf toughness values,” KIc may
   Experience has shown that cracks in the radial–axial plane in            23.5.4.4 Allowable Final Crack Depth Vessel failure is
the wall of a thick cylinder tend to propagate in the through-thick-     assumed to occur when a fatigue crack reaches the critical crack
ness direction with an aspect ratio of about 3:1 when the loading        size as discussed in the General and Theory paragraph or when
is due to cyclic pressure only. Longer cracks can result if thermal      the crack has penetrated the full thickness of the vessel wall or of
or other loadings are also present. In contrast, cracks in the root of   an individual layer in the case of a layered vessel. The following
threads in threaded end closures tend to propagate in the circum-        two design margins must be considered when calculating the
ferential direction until they extend for the full 360 deg., then        number of design cycles:
propagate through the thickness.
         (1) The number of cycles to propagate a crack from the   assumed initial size to 25% of the critical crack depth or 25% of the   Formatted: Font: Times New Roman, 9 pt
assumed initial size to the critical crack depth divided by 2     section thickness, whichever is less.
         (2) The number of cycles to propagate a crack from the
   For layered vessels, the allowable final crack depth permitted is   as follows:
different for the inner layer, intermediate layers, and outer layer
          (1) For the inner layer, the life can be determined by                    (2) For the intermediate layers, the crack must be limited to   Formatted: Font: Times New Roman, 9 pt
assuming that a crack propagates completely through the layer if the       25% of the layer thickness.
remaining layers can contain the pressure. The collapse pressure of                 (3) For the outermost layer, the crack must be limited to
the remaining layers must be at least 20% higher than the design           25% of the layer thickness or 25% of the critical crack depth,
pressure (i.e., a design margin of 1.2). Since the design margin for the   whichever is less.
unflawed vessel is 1.732, most layered vessels will meet this
requirement.
   23.5.4.5 Calculation of Stress Intensity Factors It should be         calculation of stress intensity factors is given in Appendix D.
noted first that the term “stress intensity factor” as used in Article
KD-4 and in Appendix D refers to the classic fracture mechanics             23.5.4.6 Calculation of Crack Growth Rates The crack
crack tip parameter; it should not be confused with the term             growth rate is a function of the range of stress intensity factor, K,
“stress intensity” used elsewhere in Division 3, which refers to         and a “mean stress intensity factor” correction, fRK as described in      Formatted: Subscript
two times the shear stress with no crack present. Paragraph KD-          paragraph KD-430. The mean stress intensity factor correction is
420 gives a general description of the process for calculating the       a function of the stress intensity factor ratio, RK, which is the ratio   Formatted: Subscript
minimum and maximum values for the stress intensity factor, as           of the minimum and maximum values of the stress intensity
well as the stress intensity factor due to residual stress. If the       factor, including the stress intensity factor due to residual stress.
residual stress normal to the crack is compressive, a negative           If the minimum normal stress at the crack tip during the cycle
value for the stress intensity factor due to residual stress will        under consideration is zero, and if there are no residual stresses,
result. Although this has no physical meaning because of crack           no correction is needed (fRK =1.0). Otherwise, guidance on the            Formatted: Subscript
closure, the negative value can be superimposed with positive            calculation of the correction for mean stress is given in Appendix
values that result from service loadings to account for the effects      D and will be discussed later.
of mean stress, as described below. More specific guidance on the
   Paragraph KD-430 also provides an approach for calculating a           up cycle first. Then the larger crack size that results is the starting      Formatted: Font: Times New Roman, 9 pt
threshold value for ΔK below which it is assumed that crack growth        point for the 2,000 cycles of pressure fluctuations. The incremental
                                                                                                                                                       Formatted: Font: Italic
will be zero. This threshold and the crack growth rate constants given    growth that results from these cycles is added to the crack size
in Division 3 are based on data taken in laboratory air environments.     resulting from the start-up cycle. The increment of crack growth             Formatted: Font: (Default) Times New
The designer should consider the possibility of stress corrosion          during the next cycle is then calculated based on the larger crack size      Roman, 9 pt
cracking and corrosion fatigue in aggressive service environments.        that results.
The crack growth rate in aggressive environments may be                      It is necessary to calculate a small increment of crack growth,
significantly higher than the growth rate in laboratory air.              recalculate the crack tip stress intensity factor, then calculate the next
                                                                          increment of growth. Division 3 requires an iterative approach, using
   23.5.4.7 Calculation of Number of Design Cycles In contrast to         smaller and smaller increments of crack growth to demonstrate that
the S-N approach to fatigue in KD-3, the fracture mechanics               the increment is small enough. However, a starting increment can be
approach requires that the load cycles be applied in the sequence in      determined by calculating the change in crack depth necessary to             Formatted: Font: (Default) Times New
which they are expected to occur in service. For example, if a vessel     change the range of the crack tip stress intensity factor, ΔK by about       Roman, 9 pt
will be subjected to a start-up cycle, then to 2,000 cycles of pressure   1%. If desired, a conservative approach is to calculate the crack            Formatted: Font: (Default) Times New
fluctuations of 20% of the design pressure, then to a shutdown cycle,     growth rate for each increment using the size of the crack at the end        Roman, 9 pt
the increment of crack growth should be calculated for the one start-     of the increment rather than the beginning.
23.5.5 Article KD-5 — Design Using Autofrettage                         structure for maximum fatigue life. However, the designer should
   Autofrettage is a process that is not explicitly considered in       use caution when combining autofrettage with shrink fitting, wire
Divisions 1 and 2. The purpose of autofrettage is to introduce          wrapping, or other processes that produce residual compressive
favorable residual compressive mean stresses into highly stressed       stresses. For example, if an inner layer is first autofrettaged and
“fatigue-sensitive” areas, the result of which is a dramatic            then shrink-fitted or wire-wound, the designer must should
increase in fatigue life. As with much of the technology used in        consider the nonlinear aspects of the Bauschinger effect. The
the construction of high-pressure vessels, the origins of the           inner layer may yield in compression during the shrink fitting or
autofrettage process can be traced to the manufacture of cannon         wire winding operation depending on how much it was yielded in
and large gun barrels. These barrels had a tendency to split open       tension during autofrettage. It is difficult to assess the amount of
after a number of rounds had been fired. Early producers of these       strain hardening that occurs when the inner layer re-yields in
weapons learned that shrinking iron or steel bands or hoops onto        compression and to accurately determine the actual residual
the outside surface of cannon barrels increased the life                compressive stress distribution in the assembled vessel.
significantly. Later, it was discovered that expanding the bore of a       The phenomenon of mean stress reduction due to local yielding
gun barrel by pulling or pushing a slightly larger object through       under a pressure load is not limited to the bore surface of thick
it, or by subjecting it to a high internal pressure, gave similar       cylinders; essentially any area of a pressure vessel that yields in
results. Literally translated from French, autofrettage means “self-    tension under pressure loading will have compressive residual
hooping.”                                                               stresses when the pressure is removed. This includes areas with
   23.5.5.1 General and Theory As explained in paragraph KD-            high peak stresses in both thick- and thin-wall vessels, such as the
501, the distribution of stress through the thickness of a thick wall   inside corners of nozzle to shell intersections. However, the
cylinder or sphere is highly non-uniform. Even if the wall              closed form solutions in Article KD-5 apply only to plain mono-
thickness could be increased to infinity, the total range of Tresca     block cylinders. For other cases, the determination of the value of
stress at the bore surface of a machined and polished cylindrical       mean stress to use in the fatigue calculations in Articles KD-3 and
shell would be twice the range of pressure. Therefore, the stress       KD-4 should be done by elastic–plastic finite element analysis
amplitude to be used in the fatigue analysis per Article KD-3           (FEA). When FEA is used, it is necessary to include tthe
would be equal to the range of pressure. Notches, structural            Bauschinger effect should be included in the material model.
discontinuities, and surface finish effects will increase the stress       Experimental techniques could also be used to determine resid-
range further, a phenomenon that results in an inherent limitation      ual stresses, but in most cases these are more expensive and less
on fatigue life. However, this phenomenon also provides the             accurate than FEA. For example, if strain gages are used, it may
opportunity to increase the fatigue life by introducing favorable       be necessary to install them on the surface that is in contact with
compressive mean stresses.                                              the pressurizing fluid, since this area is often where the highest
   As the internal pressure within a thick cylinder is increased, the   stresses exist. This requires high-pressure electrical feed throughs
stress at the bore surface reaches the yield strength of the material   to read the signal. Also, it is difficult to locate the strain gages
while the material at the outside surface is still well within the      exactly at the most highly stressed location in areas of high peak
elastic range (see Figs. 23.3 and 23.4). Further increases in pres-     stress such as notches and other discontinuities. If the area of high
sure result in plastic flow of the inner portion of the cylinder,       stress is small with a steep gradient, as is commonly the case, the
while the outer portion remains elastic. Note that plastic collapse     strain gage will give an average rather than a true peak stress.
does not occur until the outermost fibers reach the yield strength
of the material. When the pressure is released, the material of the        23.5.5.2 Residual Stress Calculations Paragraph KD-530
inner layers that has been permanently elongated is placed into         provides a “cookbook” procedure for calculating the residual
compression in the tangential or hoop direction as the outer por-       stress. To improve the accuracy of the result, the Von Mises yield
tion contracts elastically.                                             criterion is used rather than Tresca. Paragraph KD-522.2 gives a
   Conceptually, autofrettage is similar to the situation that exists   procedure to correct for the Bauschinger effect . The Bauschinger
when two concentric cylinders are shrink-fitted together. Since         effect is described in the section of this chapter called “Stress
the outside diameter of the inner cylinder is slightly larger than      Analysis for Fatigue.” The approach in paragraph KD-522.2 is
the bore diameter of the outer cylinder, the outer cylinder is          largely empirical, but it is based on both theoretical and experi-
heated (or the inner cylinder cooled) for assembly. As the outer        mental work performed primarily on high-strength, low-alloy
cylinder cools, it places the inner cylinder into tangential            steels similar to SA-723. The approach is believed to be conserva-
compression. The permanent elongation of the “inner layers” of a        tive for other materials, so it is required for all residual stress cal-
monoblock cylinder during the autofrettage process makes them           culations involving plain cylinders. However, for more complex
“larger” than the “outer layers.”                                       geometries, the cyclic stress–strain behavior of the material
   Note that autofrettage can be used in combination with shrink        should be determined, and the results used in a finite element
fitting, wire wrapping, or other processes that can produce favor-      analysis to obtain a more accurate indication of the residual stress.
able compressive stresses to optimize the distribution within the
23.5.6     Article KD-6 — Design Requirements for                        requires a very fine mesh at the root of the threads to accurately
           Openings, Closures, Heads, Bolting, and                       characterize the peak stresses. It is not necessary to perform a
           Seals                                                         fatigue calculation on ANSI standard nuts.
   This article contains some specific requirements for openings,
                                                                            23.5.6.2 Threaded End Closures Paragraphs KD-630 and
closures, heads, bolting, and seals. In most cases, only general
                                                                         KD-650 give limitations on single-threaded end closures and
requirements and a few specific limitations are provided, since it
                                                                         quick-actuating closures respectively. A large variety of threaded
is anticipated that these components will be designed by analysis
                                                                         and quick-actuating closures have been used for vessels that must
in most cases. However, closed form solutions are provided for
                                                                         be opened frequently. For example, the hot isostatic pressing
some cases in Non-Mandatory Appendix H.
                                                                         process requires that the vessel end closure be opened each
                                                                         processing cycle to remove the pressed and sintered item and
   23.5.6.1 Threaded Connections and Bolting Paragraphs KD-              replace it with the unsintered powder. Since this process occurs
610 through KD-621 give limitations on threaded connections
                                                                         several times per day (in some cases several times per hour), a
and bolting. The requirements for length of engagement in para-
                                                                         quick acting closure is needed to maximize production. In many
graph KD-615 626 are intended to provide a conservative                  cases, a closure with an interrupted square, acme, or buttress
engagement length for tapped holes for those cases where the
                                                                         thread form is used, so that the closure can be opened quickly by
threads must be as strong as the stud itself (i.e., the stud will fail
                                                                         rotating it a fraction of a turn and lifting. In other cases, a
before the threads). However, as written, it applies to all tapped       complete single-threaded design is used, which requires multiple
holes without exception. Note also that paragraph KD-616 627
                                                                         rotations to open. In either case, the stresses in the threads in the
requires a fatigue calculation for all threaded connections. It is
                                                                         vessel shell are typically high, and several cases of vessel failure
common practice to do this by using a stress intensification factor      have occurred due to cracks that started in the female threads.
of 4.0 on the nominal range of stress in the stud. Alternatively, a
finite element analysis of each fastener can be done, but this
   In most cases, the stress gradient from the root of the first           in performing this analysis it is common practice to assume that
female thread to the outside surface of the shell is even more             the initial flaw is semi-elliptical with a 3:1 length to depth aspect
nonuniform than for the cylindrical shell under pressure.                  ratio. The calculations should calculate crack growth in both the
Therefore, there is a tendency for a crack to propagate in the             circumferential and through thickness directions for each
circumferential direction completely around the vessel before it           increment of crack growth, and the new aspect ratio should be
propagates through the entire wall thickness. This 360 deg. crack          used for subsequent increments. This will establish whether the
can then propagate in the through-thickness direction until it             crack will propagate more rapidly in the circumferential direction.
reaches a critical size, resulting in fast fracture. If the material has      Paragraphs KD-651, KD-652, KD-652.1, and KD-653 contain
high toughness, the failure may be ductile tearing in the                  requirements that are intended to ensure that quick-actuating clo-
remaining ligament. In either case, the failure can be catastrophic,       sures are fully engaged before pressurization, and are not opened
since the end of the vessel will separate. Because of this, the            until the vessel has been fully depressurized. Several accidents
“leak-before-burst” provisions do not apply to threaded end                have occurred with lower pressure autoclaves due to the failure of
closures, so paragraph KD-631.4 requires a fracture mechanics              partially engaged quick-actuating closures.
analysis in accordance with Article KD-4 in all cases. However,
23.5.7     Article KD-7 — Design Requirements                               The axial or longitudinal stresses in a layered vessel should be
           for Attachments, Supports, and Heating                        transmitted through welds between layers rather than relying on
           and Cooling Jackets                                           friction. However, it is often possible to design a multilayer
   This article contains some specific requirements for                  shrink fit vessel such that the axial load can be carried through
attachments, supports, and external heating and cooling jackets.         only one or two of the layers. Note that the longitudinal stress due
In general, all welds attaching nonpressure parts to pressure            to pressure in a thick-wall cylinder can be significantly less than
boundary parts are required to be continuous, full penetration           one-half of the hoop stress. For example, a cylindrical shell with
welds. Fillet-welded and partial penetration-welded attachments          an outside diameter to inside diameter ratio of 2.0, made from
are not permitted, because cracks that may initiate at the heel or       SA723 Class 2 material (yield strength 120 ksi [830 MPa]), will
toe of the fillet can propagate into the pressure boundary.              have a design pressure of about 55 ksi (380 MPa) according to the
                                                                         equation in KD-251221.1. The longitudinal stress due to the
  23.5.7.1 Attachments Paragraphs KD-710 through KD-730                  pressure load only will be about 18 ksi (130 MPa), or about 15%
give requirements for attachments. Except for minor, non–load-           of the yield strength. In contrast, a thin wall vessel made from the
bearing attachments of a small size that are discussed in para-          same material, with the design pressure limited by the same
graph KD-712, attachments must be made of a listed material and          equation, will have a longitudinal stress of about 33% of the yield
must be analyzed in the same way as pressure boundary                    strength.
components.
                                                                            23.5.8.2 Residual Stresses in Shrink-Fit Cylindrical Shells
   23.5.7.2 Supports Paragraph KD-740 gives design require               Paragraphs KD-811 and KD-812 give a standard textbook
ments for supports. This paragraph is very general, and is               procedure for calculating the residual stresses in two or more
intended primarily to clarify that all loadings must be considered.      layer shrink-fit shells. It is possible to autofrettage one or more of
                                                                         the layers of a shrink-fit cylinder prior to assembly. However, this
   23.5.7.3 Heating and Cooling Jackets paragraph KD-750                 is not common in practice. If the inner layer is relatively thin, it
gives requirements for jackets. In most cases, heating and cooling       will not be possible to achieve significant residual compressive
jackets operate at low pressure, so they are normally constructed        stresses, because the stress distribution is relatively uniform. If
to the requirements of Division 1. However, welds that attach the        the inner layer is relatively thick, it is likely that the overall vessel
jacket to the pressure boundary must meet all of the requirements        will have a large enough outside diameter to inside diameter ratio
of Division 3.                                                           that yield level residual stress can be achieved in the inner
                                                                         cylinder due to shrink-fitting alone. This level of compressive
23.5.8     Article  KD-8      —     Special                Design        stress will actually be higher than could be achieved by
           Requirements for Layered Vessels                              autofrettage of a monoblock vessel with similar overall
  This article contains requirements for three types of layered          proportions because the shrink-fit vessel can be constructed such
vessels:                                                                 that it is not subject to the Bauschinger effect (e.g. if yielding as a
                                                                         result of the shrink fit is avoided). If the amount of interference
          (1) Non-welded construction, in which forged cylinders
are shrink-fitted together in two or more concentric layers.             between the inner and outer cylinders is calculated correctly, the
          (2) Construction similar to that described in item (1), but    bore of the inner cylinder will can be compressed close to the
in which each layer is rolled from plate, welded, and then               yield point without actually yielding. Since the Bauschinger effect
machined prior to shrink fitting.                                        is triggered only if plastic flow occurs, the full linear-elastic range
                                                                         of twice yield is available for service loadings. If the inner
                                                                                                                                                     Formatted: Indent: First line: 0.5"
          (3) Concentrically wrapped and welded layered vessels          cylinder had been autofrettaged prior to shrink fitting, additional
                                                                         compressive plastic flow of the bore surface, and therefore a               Formatted: Indent: Left: 0", First line: 0.5"
in which the inner layer is either conventionally rolled and long
seam-welded or of seamless construction. Successive layers are           higher interference, may have been required to achieve the same
then fitted outside, and the long seam for each layer is welded in       level of residual stress as shrink fitting alone.
place. The shrinkage of the long seam introduces a tangential               It may be tempting for the designer to consider autofrettage of
compressive stress into the previous layers. This technique can          a thick outer layer. The inside surface of the outer layer (i.e., the
also be used for spherical vessels and hemispherical heads.              location of contact between layers) will have high residual tensile
                                                                         stress. These high residual tensile stresses may be OK in many
   Depending on the application, the inner layer of these vessels        cases because cyclic stresses are typically much lower at this
may be of corrosion-resistant material. The contribution of this         location than at the inner cylinder bore. However, if the interface
material to the static and fatigue strength of the vessel is consid-     has been identified as a fatigue-sensitive region, introducing some
ered in the design analysis. It is also possible to fit a loose liner,   level of residual compressive stress into the outer cylinder using
as discussed earlier.                                                    autofrettage prior to assembly could increase the fatigue life
                                                                         significantly. The problem with this approach is that heating of
   23.5.8.1 Design Criteria and Residual Stresses For any of the         the outer cylinder for shrink fitting could relax some of the
three types of layered vessel construction, the static strength is       residual stress. In general, if heating to less than about 700°F is
considered to be the same as a monoblock vessel that has the             required, an estimate of the loss in residual stress can be
same overall thickness if                                                calculated using the ratio of yield strength at shrink-fit
          (1) the material of each of the layers has the same yield      temperature to the room temperature yield strength. Obviously, it
strength as the material of the monoblock vessel; and                    is necessary to heat the outer cylinder uniformly to ensure that
          (2) there are no significant gaps between the layers.          through-thickness thermal stresses do not result in yielding.
                                                                            An alternative assembly technique in which the inner cylinder
   Although a classical limit analysis would show that the static        is cooled for assembly could be considered. However, this is
strength under internal pressure loading should be essentially the       difficult in practice because moisture can condense on the inner
same even if there are large gaps between layers, the concern is         cylinder during assembly. The technique could be used if the
that the amount of plastic deformation that may be necessary to          parts are small enough to be assembled in a vacuum chamber or a
close the gaps may result in cracking. Also, if gaps exist after the     totally dry environment.
vessel is placed in service, the cyclic stresses in an individual           If a combination of autofrettage and shrink-fitting is used,
layer will be much higher, resulting in a shorter fatigue life than      paragraph KD-811.3 requires combining the stresses from these
would otherwisw be the case.                                             operations. Superposition can be used if the resulting stresses are
below yield, but note that superposition should always be done by      pressurized, it can have a significant effect on the range of stress
combining the stress components. Tresca or Von Mises stresses          in the inner layers. It is difficult to locate or characterize these
should not be added directly. (See also Subsection 23.5.5.1).          gaps using nondestructive examination except at the ends of the
                                                                       shell or at the location of weep holes. Therefore, Division 3
   23.5.8.3 Concentrically Wrapped and Welded Layered                  requires that the magnitude of the gaps be estimated during
Vessels Paragraph KD-820 and its dependent paragraphs give             hydrotest by measuring the actual circumferential expansion of
rules for concentrically wrapped and welded layered vessels. This      the shell between each pair of circumferential seams and
technique has been used for many years for Division 1 and              comparing it to the theoretical value that would occcuroccur if
Division 2 construction. In many cases, the heads of these vessels     there were no gaps. The ratio of measured to theoretical
are of single wall construction, but layered heads can also be         expansion is then used to calculate a “gap correction factor.” This
produced. Nozzles are typically of single wall-forged                  factor is then used to increase the principal stresses throughout
construction. The layers must should be relatively thin to achieve     the thickness of the vessel. Note that, in contrast to Division 2,
good fit-up, so paragraph KD-821(a) limits the outer diameter to       both the static and fatigue analyses are affected by the gap
inner diameter ratio to 1.1.                                           correction factor. The significantly higher allowable stresses of
   Although a significant amount of residual compression is            Division 3 make it important to be sure that the design approach
induced in the inner layers of these vessels due to the shrinkage of   is conservative.
the longitudinal weld as each successive layer is applied, the            A similar approach for gap correction is provided in paragraphs
magnitude of the residual stress is hard to predict. This is due to    KD-824 and KD-825 for spherical shells and hemispherical
the effects of friction between layers as the weld cools and to the    heads.
residual stresses in the welds themselves. Therefore, Division 3
does not permit the beneficial effects of the residual compressive        23.5.8.4 Openings and Supports for Concentrically
stress to be considered in the fatigue analysis.                       Wrapped and Welded Layered Vessels Paragraph KD-840
   Since the plates that form the layers are not machined after        gives rules for openings in concentrically wrapped and welded
forming, it is possible for small gaps between layers to exist at      layered vessels. Openings must be integrally reinforced.
some locations. Although this may not affect the plastic collapse         Figure KD-850 in Division 3 gives some acceptable support
pressure of the shell, since the gaps tend to close as the vessel is   deetails. These are similar to those in Division 2.
23.5.9     Article  KD-9     —    Special   Design                     the same equation to limit the wall thickness in both cases for
           Requirements for Wire-Wound Vessels and                     simplicity.
           Frames                                                         The external frames that carry the pressure end load can be of
   This article contains requirements for a specialized type of con-   either “conventional” or wire-wound construction. Only perfor-
struction that is found in neither Division 1 nor Division 2. A        mance requirements for frames are given, and detailed design is
wire-wound vessel is constructed by helically winding a square or      left to the designer.
rectangular cross section wire onto a cylindrical core that has
                                                                          23.5.9.1 Residual Stresses in Wire-Wound Vessels Paragraph
been fabricated by conventional means. The wire is maintained
                                                                       KD-911 gives equations that can be used to calculate the residual
under tension as the winding proceeds so that the inner core is
                                                                       stress in the core layer and in each wire-wound layer based on the
placed into compression. Since the edges of the wire are not
                                                                       tension used during winding and the geometry of the shell.
welded together, wire-wound vessels can support axial or
longitudinal loads only through the inner core layer. Since this          23.5.9.2 Derivation of a Design Fatigue Curve in Wire-
layer is normally very thin, essentially all wire-wound vessels use    Wound Vessels Paragraph KD-932 gives requirements for
an external frame to carry the axial pressure load. Since most of      fatigue testing of the wire used for winding and for the
these vessels are used in applications such as isostatic pressing or   determination of design factors to be used in the fatigue analysis.
food processing where the vessel must be opened frequently, the        The procedure is based on the assumption that a few widely
frame is normally designed to permit rapid withdrawal of one end       spaced fatigue failures in individual wires will not result in failure
plug. Although the theoretical collapse pressure of a cylindrical      of the vessel itself because friction between successive layers will
shell that does not carry the axial pressure load through the shell    prevent the broken wire from “unraveling.”
is up to about 15% lower than one that does, it was decided to use
23.5.10 Article KD-10 — Special Requirements for
        Vessels in High Pressure Gaseous Hydrogen
        Transport and Storage ServiceSpecial Design
        Requirements for Interlocking Strip-Wound
        Vessels
   Article KD-10, and the associated fabrication requirements in
Article KF-10, have been deleted from the 2004 edition of
Division 3. These articles had covered vessels that were cons-
tructed by helically winding a profiled strip onto a core in such a
way that successive layers interlocked. In contrast to wire wound
vessels, the interlocking strip permited the vessel to carry end
loads. These articles were deleted because the SWG was not
aware of any current applications for this method was introduced
in the 2007 Edition to provide requirements for vessels to support
the emerging “hydrogen economy”. Pressures, up to 15,000 psi
(100 MPa), are expected to be necessary for the economical
transport and storage of gaseous hydrogen at ambient temperature
for use as a motor fuel and for other applications, Ambient
temperature gaseous hydrogen at high pressures can cause
environmental crack growth and can increase the growth rate of
fatigue cracks in ferritic steel alloys. The problem is exacerbated
as the yield and tensile strength of the alloy increases.

   23.5.10.1 Scope of KD-10 Paragraph KD-1000 limits the scope
of coverage of Article KD-10 to non-welded vessels with
hydrogen partial pressure exceeding 6,000 psi (41 MPa) or
welded vessels with hydrogen partial pressures exceeding 2,500
psi (17 MPa). However, the hydrogen partial pressure limits are
reduced to 750 psi (5.2 MPa) for non-welded vessels constructed
of materials with a tensile strength of 137 ksi 945 MPa) or greater
or for welded vessels constructed of materials with a tensile
strength of 90 (620 MPa) ksi or greater. These scope limitations
were based on the experience and judgment of the SG-HPV and
the ASME BPV Project Team on Hydrogen Tanks. Materials
within the scope include high and low alloy steels, as well as the
aluminum alloys 6061-T6 and 6061-T651.

   23.5.10.1 Fracture Mechanics Analysis Paragraph KD-1010                Formatted: Indent: First line: 0.13", Space
requires that the design fatigue life be determined using the             After: 0 pt
fracture mechanics approach in Article KD-4, modified to use the
FAD approach, as well as the stress intensity and reference stress
solutions, in API RP-579. In addition, the fatigue crack growth
rate must be determined in a high pressure hydrogen environment
for each material to be used in construction.               Also, for
determination of the critical crack size, the critical crack tip stress
intensity is defined as the smaller of the KIc or KIH values. KIc is      Formatted: Font: Italic
the critical crack tip stress intensity, determined in the
                                                                          Formatted: Font: Italic, Subscript
conventional way. KIH is the threshold for subcritical crack
growth in the high pressure hydrogen environment. Article KD-             Formatted: Font: Italic
10 contains detailed requirements for determining these values.           Formatted: Font: Italic, Subscript
The data must be obtained by the manufacturer for at least three
heats of steel. The data thus obtained can be used for other              Formatted: Font: Italic
vessels made from the same or similar material, if the yield and          Formatted: Font: Italic
tensile strength of the material does not exceed that of the tested
material be more than 5%.
   As more data become available in the literature, assuming that
the data fall within a relatively narrow band, it is the intention of
the SG-HPV to publish values of crack growth rates and possibly
the threshold for subcritical crack growth, for use in design
23.5.11 Article KD-11 — Design Requirements                            23.5.12 Article KD-12 —                  Experimental
        for Welded Vessels                                                     Design Verification
   When the need for Division 3 was first recognized and the              Although modern analysis tools have reduced the need for
Committee was formed, most of the participants envisioned a            experimental stress analysis, it is still an appropriate technique in
Code that would cover only the forged, nonwelded type of con-          many some cases. The rules in this article are very similar to the
struction that was being used extensively for vessels with design      rules in Appendix 6Annex 5.F of Division 2. One significant
pressures of ~20,000 psi (140 MPa) to ~50,000 psi (350 MPa).           difference is in the determination of the collapse load. Figure 6-
However, some Manufacturers and Users had experience with              153 in Division 2 illustrates the “two angle” approach; however,
welded construction for vessels in the lower part of this pressure     Division 3 simply requires that the strain be limited to 2%. It
range, so it was agreed to include welded construction within the      further states that this strain should be measured in a location that
scope. To a large extent, this explains the significant restrictions   will produce a “primary strain”—in other words, the intent is not
on welded construction that have been incorporated into Division       to limit the strain in local notches or discontinuities. At first
3.                                                                     glance this seems to be inconsistent with the requirements of KD-
                                                                       240(b)(1), which limits the strain at any point to 5%based on the
   23.5.11.1 Types of Joints Permitted Paragraph KD-1110 per-          strain limit damage accumulation. However, in an experimental
mits only Type No. 1 butt joints, with only a few minor                procedure, Division 3 requires that the vessel reach the collapse
exceptions. Type No. 1 butt joints are described in paragraph KF-      pressure without failure due to high local strains, so it is the
221, which is referenced in paragraph KD-1110. The description         responsibility of the designer to ensure that local strains are low
is similar to that in Divisions 1 and 2., However, noteexcept that     enough to avoid failure during the test.
all Type No. 1 butt joints used in Division 3 construction must be
ground or machined flush. However, Code Case 2592 permits un-             23.5.12.1 Paragraph KD-1260 Experimental Determination
ground welds if the weld is not accessible, the design fatigue life    of Allowable Number of Operating Cycles Article KD-1260 is
is calculated using Article KD-4 (fracture mechanics) and the          similar to the corresponding requirements in Division 2. This
finite element analysis includes the “worst case” weld                 experimental approach can be used only when the traditional “S-
misalignment, peaking and weld profile. The original concern of        N approach” in Article KD-3 is permitted by the requirements of
the SG-HPV members with welds that are not ground smooth was           paragraph KD-140 —that is, when a leak-before-burst mode of
the difficulty of predicting the appropriate stress intensification    failure can be shown. It would be possible to develop an experi-
factors for use in the fatigue analysis. However, successful           mental fracture mechanics procedure that could be used as an
experience with the fracture mechanics approach has made it            alternative to the requirements of Article KD-4 if there are
possible to have confidence in the design fatigue life calculations    enough potential applications to justify the effort.
in the presence of weld flaws. It is anticipated that this Code
Case will be incorporated into the Code in the future.
23.6       PART      KF                        —                       HPV agreed to include it. However, forged, non-welded
           FABRICATION                                                 construction is still used for the majority of vessels with design
                                                                       pressures over about 30,000 psi (210 MPa).
           REQUIREMENTS
   In general, the fabrication requirements in Division 3 are            23.6.2.1 Grinding or Machining of Welds To minimize local
similar to those in Division 2. Therefore, the focus in the            notches, paragraph KF-204 requires that all welds be machined or
following discussions is primarily on the differences.                 ground to a blend radius and surface finish consistent with the
                                                                       requirements of the engineering design. This may be a difficult
23.6.1     Article    KF-1    —     General                            requirement to meet for surfaces that are difficult to access, but
           Fabrication Requirements                                    exceptions are currently permitted only if a the requirements of
  Article KF-1 gives general requirements that apply to all            Code Case that describes appropriate alternatives is
vessels constructed to Division 3. Subsequent articles in Part KF      approved2592 are met (see paragraph 23.5.11.1).
apply to specific types of construction.
                                                                          23.6.2.2 Types of Weld Joints Permitted Paragraph KF-220
   23.6.1.1 Examination of Materials Because of the high level         permits only Type No. 1 butt joints, with a few exceptions,
of concern about fatigue, it is important to ensure that flaws that    because of concerns about local stress intensification and
could be sites for the initiation of fatigue cracks are detected and   difficulty of examination. The most significant exception permits
repaired to the maximum extent practicable. For example, para-         a full-penetration groove weld for nozzle attachment, but only if
graph KF-121.1 requires that all edges cut during fabrication be       the detail for a modified set-on nozzle as shown in Fig. KD-1131
examined and repaired if needed. The details are found in Part         in Division 3 is used. Restrictions on the weld procedure
KE.                                                                    qualification for these welds in paragraph KF-222.1 are intended
                                                                       to reduce the probability of flaws. Other exceptions permit the
   23.6.1.2 Cutting of Materials Paragraph KF-121.2 permits            installation of lugs, clips, heating and cooling jackets, etc.
thermal cutting of materials, but requires slag and detrimental
discoloration of material to be removed by mechanical means. In        23.6.3     Article KF-3 — Fabrication Requirements
addition, the effects of thermal cutting on mechanical properties                 for Materials with Protective Linings
must be considered. Although these provisions apply to all                This article applies only to integrally clad or weld overlay-type
materials, they are intended primarily to address a concern with       linings.
the high-strength materials that are typically used for high-
pressure vessel construction. Although degradation of material         23.6.4     Article KF-4 — Heat Treatment of Weldments
properties in the heat-affected zone of a weld will presumably be         The requirements for heat treatment of weldments are similar
detected during the weld procedure qualification test, degradation     to those in Divisions 1 and 2. Note that heat treatment require-
due to thermal cutting may not be detected because no                  ments for quenched and tempered materials are covered in
qualification test is required. It is therefore important either to    Paragraph KF-630.
remove this material or to be sure that the properties are still
acceptable after cutting.
                                                                       23.6.5     Article KF-5 — Additional Fabrication
   Although not specifically addressed in Division 3, it should be
noted that “abusive grinding,” which can remove material
                                                                                  Requirements for Autofrettaged
quickly, can heat a thin layer at the surface of a quenched and                   Vessels
tempered steel above the austenitizing temperature, resulting in a        As discussed in the design section (KD-5), autofrettage has
thin layer of extremely hard material that can crack. Although         been used extensively in high-pressure vessels to introduce com-
these cracks are rarely more than 0.010 in. (0.25 mm) deep, they       pressive mean stresses that significantly improve fatigue life.
can be sites for the initiation of fatigue.                            Since autofrettage involves increasing the pressure in a vessel
                                                                       until significant plastic deformation has occurred, the pressure
   23.6.1.3 Radius Requirements for Inside Edges of Nozzles It         required is typically very close to the plastic collapse pressure. In
is well known that “cross-bore intersections” as they have tradi-      addition, small changes in pressure can result in large changes in
tionally been called within the high-pressure community, are sites     stresses and strains, resulting in significant differences in the level
where fatigue cracks can initiate. A “cross-bore intersection” is      of residual compressive prestress that is achieved. For these
the intersection of an opening in a cylindrical shell with the shell   reasons, it is necessary to control the process carefully.
bore. The negative experience was primarily with forged, non-             It should also be recognized that autofrettage only works in
welded vessels in relatively high cycle service. However, the          those areas of a pressure vessel where steep stress gradients exist
SWGSG-HPV decided to require that the inside corner of all             under pressure loading. This situation exists in areas of local
nozzle-to-shell intersections be radiused to minimize the local        stress intensification, such as at the inside corners of nozzles, in
stress intensification. The basis for this is largely empirical, and   many vessels. In fact, vessels constructed to Divisions 1 and 2
some recent finite element analysis work has indicated that the        frequently undergo some plastic deformation in areas of high
stress intensification effect may be much smaller than previously      local stress during the normal hydrostatic test. The resulting
believed. The SG-HPV has an agenda item to evaluate whether            residual compressive stresses in these areas are favorable not only
the radius requirement is needed.                                      for fatigue life, but also in reducing the probability of brittle
                                                                       fracture if the vessel is pressurized at a low temperature.
23.6.2     Article KF-2 — Supplemental                                    23.6.5.1 Autofrettage Procedures Paragraph KF-520 requires
           Welding Fabrication Requirements                            that a detailed, written procedure be prepared. It is particularly
  Article KF-2 gives general requirements that apply to all weld-      important to develop a procedure for controlling the extent of aut-
ed vessels constructed to Division 3. On a historical note, many of    ofrettage . The most common technique has been to install strain
the individuals who advocated the development of a high-               gages on the most highly stressed external surfaces. The specific
pressure Code in the middle to late 1970s envisioned covering          location of these gages should be determined by the designer. All
only the construction of the forged, non-welded vessels that were      gages should be monitored during the autofrettage process to
being used at that time for hot and cold isostatic pressing, quartz    ensure that the strain is not excessive, but is adequate to achieve
crystal growing, polyethylene manufacture, and a few other             the desired level of compressive prestress.
applications. However, others noted that welded construction was
used for high-pressure vessels in some cases, so the SWGSG-              23.6.5.2 Number of Pressurizations and Examination after
Autofrettage The local strain during autofrettage can be very            in critical areas and to check gage behavior during the initial
high. As with the plastic strain that occurs during forming opera-       stages of pressurization.
tions, it is important to ensure that the material has sufficient duc-      Paragraph KF-530 reinforces the requirement for an
tility to withstand these strains without damage. Typically, the         examination of all surfaces after autofrettage. An alternative
highest strains are on internal surfaces that are exposed to the         examination is required for surfaces that may be inaccessible after
fluid pressure. These areas usually also have low triaxial stresses,     autofrettage.
which increases the apparent ductility, so cracking due to the
single application of the autofrettage pressure is not common.           23.6.6     Article KF-6 — Additional Fabrication
    However, if there are difficulties with the autofrettage process                Requirements for Quenched and
such that several attempts are needed to reach the autofrettage                     Tempered Steels
pressure, low-cycle fatigue cracks can result. For example, if a           This article only applies to those quenched and tempered steels
pump fails or if a seal leaks, it may be necessary to depressure the     that can be used for welded parts. The quenched and tempered
system, fix the problem, and then repressurize. The author is            high-strength, low-alloy steels that are typically used in non-
aware of one case in which a compressor component, which was             welded construction of high-pressure vessels are outside of the
not constructed to Division 3, failed after five attempts to reach       scope of this article.
the design autofrettage pressure. It is therefore very important to
properly inspect sealing elements and carefully assemble all joints
                                                                         23.6.7     Article      KF-7 —       Supplementary
before the initial pressurization. Double block and bleed valves
should be provided to isolate the pump so that it can be repaired
                                                                                    Requirements for Materials with Welding
without depressurizing the vessel. Another problem that has                         Restrictions
resulted in more than one attempt to reach the autofrettage                 This article provides for the repair of defects by welding, with
pressure is strain gage failure. It is prudent to use redundant gages    significant restrictions, for materials that are not permitted for
                                                                         welded construction.
23.6.8    Article    KF-8  —    Specific                            category. Shrink-fit vessels, in which the layers are either forged
          Fabrication Requirements for                              or of non-welded or welded construction, are covered in
          Layered Vessels                                           paragraph KF-810. Concentrically wrapped welded layered
   As discussed in paragraph 23.5.8, there are several very         vessels are covered in the rest of the article.
different types of construction that fall into the layered vessel
   23.6.8.1 Shrink-Fit Vessels Paragraph KF-810 covers require-        those making up the inner layer. This is done so that if a hole or
ments for these vessels. Shrink fitting can be difficult for large     crack develops in the inner layer, the leak can be detected before
vessels because practical limitations on the temperature difference    pressure builds up between layers. If vent holes are not provided,
between the inner and outer cylinders result in a very small gap       a leak through one or more of the inner layers could overpressure
for assembly. If assembly does not go smoothly, the outer              the outer layers, resulting in a rupture of the vessel. Although
                                                                                                        1
cylinder can shrink onto the inner one before they have been           Division 3 requires only two 4 in. (6 mm) diameter holes in each
positioned correctly, which can result in having to scrap the          plate, the designer should consider the need to provide larger
assembly and start over. It is usually impractical to cool the inner   holes on a large vessel to ensure that a pressure buildup cannot
cylinder for assembly because moisture from the air will               occur.
condense on the surface. The upper temperature limit for the outer        The requirements in paragraph KF-826 for controlling gaps
cylinder is typically governed by the tempering temperature of the     between layers should be noted. This is particularly important for
material. However, the possibility of temper embrittlement due to      thick-wall high-pressure vessels, since a gap between two of the
excessive holding time at temperatures below the tempering             innermost layers can lead to high cyclic stresses as the material
temperature should also be considered. The Manufacturer must           deforms to close the gap under pressure or thermal loads.
provide a written assembly procedure and report, which are used           Paragraph KF-825 gives special nondestructive examination
to verify that the design residual stress distribution was achieved    requirements for the welds in concentrically wrapped welded lay-
during the fabrication of the layers and the shrink-fitting            ered vessels. It is particularly important to examine these welds
assembly.                                                              carefully during fabrication because it is difficult to examine
                                                                       welds on internal layers after the completion of the vessel and
  23.6.8.2 Concentrically Wrapped Welded Layered Vessels
                                                                       during service. On the plus side, layered construction usually
Paragraph KF-820 covers requirements for these vessels.
                                                                       gives inherent leak-before-burst performance, because subcritical
Paragraph KF-824 requires vent holes in each plate, except for
                                                                       cracks will typically arrest at the interface between layers.
23.6.9     Article KF-9 — Specific Fabrication                       individual wires will not result in loss of containment, as friction
           Requirements for Wire-Wound                               will act to maintain the prestress in the other wires.
           Vessels and Frames
   Since the wire used for the wrapping on these vessels has a
                                                                     23.6.10 Article KF-10 — Specific
very high strength, it is not practicable to make butt-weld joints           Fabrication Requirements for
between continuous lengths of wire that have the same strength as            Helically Wound Interlocking Strip
the wire. These reduced-strength welds are permitted in the inner            Vessels
wire layers because the reduction in strength of the vessel as a       Article KF-10 has been deleted from the 2004 edition of
whole is negligible. It is necessary to reduce the winding tension   Division 3 as discussed in paragraph 23.5.10.
for up to two turns before and after the weld. However,
experience with these vessels indicates that failure of a few
23.7       PART KR — PRESSURE-
           RELIEF DEVICES
   The overpressure protection requirements in Division 3 are
intended to follow the philosophy of Divisions 1 and 2 to the
maximum extent possiblepracticable. One significant difference is
that Division 3 permits inherent overpressure protection. (See
paragraph KR-125). The SWGSG-HPV recognized that it was
common practice to limit the pressure within vessels in
nonreactive service (e.g., hot and cold iso-static pressing) by
limiting the supply pressure of the compressors or intensifier
pumps rather than providing a device on the vessel itself. The
source of pressure must be external to the vessel. For example, if
a high-pressure gas compressor piston or plunger that supplies
argon to an isostatic press is driven by a low-pressure hydraulic
cylinder, a low-pressure relief valve in the hydraulic system can
effectively limit the discharge pressure of the compressor and
thereby limit the pressure in the vessel.
   Code Case 2211 permits overpressure protection by systems
design for Divisions 1 and 2 service, but this Code Case requires
detailed analysis of potential sources of overpressure. Paragraph
KR-125 in Division 3 simply requires that the pressure be under
such positive control that the pressure in the vessel cannot exceed
the design conditions by more than the accumulation permitted in
paragraph KR-150.
   “Pop action” safety relief valves are not readily available for
pressures above 10,000 psi (70 MPa). Small spring-loaded valves,
with progressive opening, similar to liquid service relief valves,
are available for very high pressures, but the increase in pressure
from initial opening to full-rated capacity is typically well over
the 10% accumulation permitted by Division 3. In addition, even
“pop action” valves with set pressures of about 5,000 psi (35
MPa) constructed to ASME Division 1 and Division 2
requirements may not perform as expected.
   A “pop action” valve is designed so that when the disk lifts a
small amount, a larger area is exposed to the pressure, resulting in
additional lift. The increased area must be large enough so that
the lifting force will overcome the additional force exerted by the
spring as it is compressed from the seated to the full open condi-
tion. However, if the additional area is too large, the pressure will
drop excessively before the valve closes (excessive blowdown).
When a specific valve is tested, the accumulation and blowdown
can be verified experimentally. Scale-up of valves designed for
low pressure to a higher pressure involves careful consideration
of the increase in force on the valve disk due to the higher
pressure and the relative increase in spring force. As the pressure
increases, the deviation from ideal gas behavior becomes more
pronounced. This can result in a smaller relative increase in the
lifting force as the valve set pressure increases. To compound the
problem, there is an incentive to select springs for high-pressure
valves that are relatively stiffer than the springs in low-pressure
valves to minimize spring length and diameter. Since there are no
scale-up rules or guidelines in the Codes, the SWGSG-HPV
decided to require testing at the design pressure to ensure that full
lift and flow capacity will be achieved within the 10%
accumulation permitted.
   For these reasons, Article KR-5 prohibits the practice of
extrapolating the results of flow capacity tests at low pressure to a
higher pressure. The SWGSG-HPV recognized that this would be
difficult to achieve because most facilities for flow capacity
certification do not have the capability to test at high pressures.
However, as a practical matter, almost all high-pressure vessels
use either rupture disks or inherent overpressure protection, so
until an acceptable procedure for scale-up is developed and
incorporated into Division 3, the testing requirement is a prudent
way to ensure that appropriate protection is provided. As of this
writing, the SWGSG-HPV is evaluating alternatives for new
rules.
23.8       PART      KE                         —                      alternative approaches and stress intensity solutions as
           EXAMINATION                                                 appropriate. In particular, the author recommends that the weight
                                                                       function method in API-579-1/ASME FFS-1, 2007 Edition be
           REQUIREMENTS                                                used. This recently published method includes the most recent
   The examination requirements in Division 3 are similar to           work in this area.
those in Divisions 1 and 2, but there is increased emphasis on            This AAppendix D also gives good guidance for determining
surface examination to detect small cracks that may be initiation      critical locations where cracks may start and rules of thumb for
sites for fatigue. It is anticipated that small surface indications    crack aspect ratios.
will be repaired by blend grinding without weld repair if the
resulting cavity is within the limits specified in paragraph KE-          23.11.1.1 Fracture Toughness Correlations Paragraph D600
211.                                                                   gives some equations that can be used to provide an acceptable
                                                                       estimate of fracture toughness from Charpy V-notch impact
23.8.1     Performance      Demonstration                  for         energy. Correlations are provided among the following toughness
           Ultrasonic Examination                                      and impact parameters such that an acceptable estimate of any
  The performance demonstration requirements for ultrasonic            parameter can be obtained if one is known:
examination in Division 3 are similar to those in Code Case 2235,                (1) plane strain fracture toughness, KIc;
which applies to Divisions 1 and 2.                                              (2) J-integral, JIc;
                                                                                 (3) crack tip opening displacement, CTOD; and
23.8.2     Final Examination                                                     (4) Charpy V-notch impact strength, CVN.
  Due to the concern about fatigue, Article KE-4 requires exami-
nation of all surfaces of pressure boundary components by the             There are many fracture toughness correlations available in the
wet magnetic particle method for ferromagnetic materials and by        literature. In particular, some of these correlations are found in
the liquid penetrant method for nonmagnetic materials after all        API-579-1/ASME FFS-1. All are empirical and based on data
fabrication and testing have been completed.                           from a limited number of materials. It should also be recognized
                                                                       that the degree of constraint at a specific location within a high-
                                                                       pressure vessel is rarely plane strain because of the influence of
23.9       PART    KT   —                                              triaxial stresses. When a fracture mechanics specimen is loaded
           TESTING                                                     uniaxially, triaxial stresses are generated by self-constraint, since
                                                                       the specimen geometry is typically designed to inhibit necking in
           REQUIREMENTS
                                                                       the planes transverse to the direction of plastic flow. However, in
  The testing requirements in Division 3 are similar to those in       actual structures, multiaxial loading and local geometry effects
Divisions 1 and 2.                                                     can result in stress states that are significantly more severe than
                                                                       plane strain conditions. Conversely, these factors can also
                                                                       produce stress states that are more favorable than plane stress,
                                                                       even in very thick wall vessels. Therefore, it should be recognized
23.10      PART   KS               —   MARKING,                        that the correlations provided in paragraph D-600 are only
           STAMPING,              REPORTS, AND                         approximations.
           RECORDS
                                                                       23.11.2 Appendix      E      —                   Non-Mandatory
   The requirements for marking, stamping, reports, and records                Construction Details
in Division 3 are similar to those in Divisions 1 and 2.                 This Appendix provides a useful collection of experience
                                                                       based, semi-empirical design approaches that have been used for
                                                                       common details on high-pressure vessels.
23.11      MANDATORY AND
                                                                       23.11.3 Appendix G — Non-Mandatory Design
           NONMANDATORY
                                                                               Rules for Clamp Connections
           APPENDICES                                                     The rules in this Appendix are based to a large extent on the
  A few comments on Non-Mandatory Appendices D, E, G, and              rules for clamp connections in Division 1, Appendix 24.
H are provided below.
                                                                       23.11.4 Appendix H — Non-Mandatory Openings
                                                                               and their Reinforcement
23.11.1 Appendix    D   —     Non-Mandatory                               This Appendix provides for the reinforcement of openings
        Fracture Mechanics Calculations                                using the “area replacement” method of Divisions 1 and 2.
  Appendix D provides guidance on one approach that can be             Although the area replacement method was initially developed for
used for the necessary fracture mechanics calculations when            thin shells, and there is little theoretical basis for its application to
Article KD-4 is used for fatigue analysis. This Appendix is based      the thick shells typically used for high-pressure vessels, the expe-
on linear-elastic fracture mechanics, and is based to a large extent   rience of SWGSG-HPV members is that it provides an
on ASME Section XI, Article 3000. It was made non-mandatory            acceptable, although conservative, design in many cases.
because the SWGSG-HPV recognized that there are many other             Therefore, it was decided to include it as a non-mandatory
approaches to the analysis of crack-like flaws in the literature.      alternative.
New stress intensity solutions are also published frequently.
Therefore, it is intended that the designer be permitted to use

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:5
posted:9/11/2012
language:Latin
pages:42