Professional Documents
Culture Documents
DNV CG 0134
DNV CG 0134
The content of this service document is the subject of intellectual property rights reserved by DNV AS (“DNV”). The user
accepts that it is prohibited by anyone else but DNV and/or its licensees to offer and/or perform classification, certification
and/or verification services, including the issuance of certificates and/or declarations of conformity, wholly or partly, on the
basis of and/or pursuant to this document whether free of charge or chargeable, without DNV’s prior written consent. DNV
is not responsible for the consequences arising from any use of this document by others.
The PDF electronic version of this document available at the DNV website dnv.com is the official version. If there
are any inconsistencies between the PDF version and any other available version, the PDF version shall prevail.
DNV AS
FOREWORD
DNV class guidelines contain methods, technical requirements, principles and acceptance criteria
related to classed objects as referred to from the rules.
This service document has been prepared based on available knowledge, technology and/or information at the time of issuance of this
document. The use of this document by other parties than DNV is at the user's sole risk. Unless otherwise stated in an applicable contract,
or following from mandatory law, the liability of DNV AS, its parent companies and subsidiaries as well as their officers, directors and
employees (“DNV”) for proved loss or damage arising from or in connection with any act or omission of DNV, whether in contract or in tort
(including negligence), shall be limited to direct losses and under any circumstance be limited to 300,000 USD.
CHANGES – CURRENT
Changes - current
This document supersedes the October 2018 edition of DNVGL-CG-0134.
The numbering and/or title of items containing changes is highlighted in red.
Buckling criteria App.D Alignment of buckling criteria in the appendix with the
requirements of the class guideline without impact on existing
Moss tank designs.
Rebranding to DNV All This document has been revised due to the rebranding of DNV
GL to DNV. The following have been updated: the company
name, material and certificate designations, and references to
other documents in the DNV portfolio. Some of the documents
referred to may not yet have been rebranded. If so, please see
the relevant DNV GL document.
Editorial corrections
In addition to the above stated changes, editorial corrections may have been made.
DNV AS
CONTENTS
Contents
Changes – current.................................................................................................. 3
Section 1 General.................................................................................................... 7
1 Introduction.........................................................................................7
2 The spherical tank concept..................................................................8
3 References........................................................................................... 8
4 Symbols and definitions...................................................................... 9
5 Calculation tools................................................................................ 12
6 Scope of analyses.............................................................................. 13
7 Assumptions...................................................................................... 16
DNV AS
1 Description.........................................................................................71
Contents
2 Loads for pump tower design............................................................ 71
3 Sloshing loads....................................................................................71
4 Inertia and gravity loads................................................................... 82
5 Thermal loads.................................................................................... 83
6 Combination of loads......................................................................... 83
7 Modelling of tower structure............................................................. 84
8 Capacity assessment..........................................................................84
DNV AS
2 Analysis procedure.......................................................................... 121
Contents
Section 11 Bibliography...................................................................................... 124
1 Bibliography list...............................................................................124
DNV AS
SECTION 1 GENERAL
Section 1
1 Introduction
1.1 Objective
This class guideline (CG) provides guidance to DNV-RU-SHIP Pt.3 and DNV-RU-SHIP Pt.5 Ch.7 Sec.21 for
design of liquefied gas carriers with spherical independent tanks of type-B.
1.2 Scope
The class guidline covers the design and assessment procedures for hull structures, cargo tanks and
supporting structures of liquefied gas carriers with independent spherical tanks type-B, including guidance on
fatigue strength assessment and fracture mechanics analyses.
1.3 Application
Structural analysis carried out in accordance with the procedures/methods described in this CG is accepted as
basis for plan approval.
In cases where there are any contradiction between this CG and rules, the rules shall prevail.
Additional guidelines for direct structural analysis of liquefied gas carriers with spherical tanks are given in
DNV-CG-0127 Finite element analysis, DNV-CG-0128 Buckling, and DNV-CG-0129 Fatigue assessment of
ship structures, and DNV-CG-0130 Wave loads.
DNV AS
Section 1
2 The spherical tank concept
3 References
Table 1 lists DNV references used in this document.
DNV-CG-0128 Buckling
DNV AS
Section 1
4 Symbols and definitions
4.1 Symbols
For symbols not defined in this document, see DNV-RU-SHIP Pt.3 Ch.1 Sec.4 and DNV-RU-SHIP Pt.5 Ch.7
Sec.1.
4.2 Abbreviations
The abbreviations described in Table 2 are used in this document.
Table 2 Abbreviations
Abbreviation Description
AC acceptance criteria
CL center line
FE finite element
GM metacentric hight
LC loading condition
DNV AS
Abbreviation Description
Section 1
OSA as defined in DNV-RU-SHIP Pt.3 Ch.4 Sec.2 [1.1]
4.3 Nomenclature
Figure 1 to Figure 4 show the nomenclature for a typical LNG carrier with spherical tanks.
DNV AS
Section 1
Figure 2 Nomenclature for a typical girder
DNV AS
Section 1
Void Cross decks
Void
Bulkhead
Passage tunnel
WB tank
Ring girder
Double side w/
stringers/tween Skirt
decks
Stool with or
without WB Stool frames and
Foundation bulkheads
WB tank deck
Hopper frame
structure
Double bottom
Pipe
passage
Figure 4 Typical arrangement of the cargo hold, the tank left out for clarity
5 Calculation tools
DNV AS
independent tank (cargo containment system), a direct wave load analysis is required to obtain hull girder
Section 1
loads and design accelerations for the cargo tanks, see DNV-RU-SHIP Pt.5 Ch.7 Sec.21 [1.3.3].
3) WASIM is a linear/non-linear frequency domain/time domain computational tool for sea keeping and load
analysis of ships. The complete 3D interaction between waves and hull at forward speed is included. The
computer program is not limited to small waves but can simulate also extreme wave conditions.
4) NVSPHERE is a computer program offered by DNV suitable for scantling assessment for the cargo tank.
The offered calculation cannot completely replace a finite element analysis of the cargo tank system, as
only the membrane stresses are considered in NVSPHERE. In areas where bending stresses are present
FEM models are needed i.e. for equator profile and tower supports. The program requires input of directly
calculated interaction forces.
5) NVSKIRT is a computer program offered by DNV suitable for buckling assessment of the cargo tank skirt
in accordance with the required buckling code, see App.D. The offered calculation is not mandatory but will
be accepted as basis for approval of the buckling capacity of the skirt. The program requires input of directly
calculated stresses.
6) TTSlosh is a computer program developed by DNV based on new sloshing tests, see /17/ and /18/.
7) The Nauticus Hull program package includes tools comprising pre-processors, environmental analysis
programs, structural analysis programs and postprocessors for the purpose of finite element analyses.
6 Scope of analyses
Items to be
Task summary Reference
addressed
Damage stability — The ship shall comply with the requirements for ship type 2G.
and separation of — The calculation shall include all relevant loading conditions with partially DNV-RU-SHIP Pt.5 Ch.7
cargo hold spaces Sec.2
filled tanks.
Cargo hold — A three dimensional integrated ship hull cargo hold and cargo tank DNV-RU-SHIP Pt.5 Ch.7
analysis model FE analysis shall be carried out. Sec.21 [1.3.2] and
— A complete global (full ship) model with cargo hold mesh in the cargo
DNV-RU-SHIP Pt.5 Ch.7
area can be used with application of boundary conditions on different
Sec.21 [2.1.2]
location depending on the considered cargo hold.
Hull fatigue — Fatigue strength assessment to be carried out for the hull structure
strength according to general rule scope as detailed in this document. Sec.9 [5]
— Design target life of minimum 25 years based on world-wide operation
DNV-RU-SHIP Pt.3 Ch.9
(scatter diagram).
— Rules defined loads to be applied, unless CSA notation is specified.
DNV AS
Classification requirements – hull structures
Section 1
Items to be
Task summary Reference
addressed
Temperature — If not known from similar designs, a temperature calculation to be DNV-RU-SHIP Pt.5 Ch.7
calculation documented for material selection of hull structures shall be carried out. Sec.21 [4.1.1] and
DNV-RU-SHIP Pt.5 Ch.7
Sec.21 [4.1.2]
— For definition of ambient temperatures see the listed references. DNV-RU-SHIP Pt.5 Ch.7
Sec.4 [5.1.1]
Table 4 Overview – Cargo tanks and supporting structures in way of cargo tanks
Classification requirements
Cargo tanks and supporting structures in way of cargo tanks
Items to be
Task summary Reference
addressed
Design loads -8
— A complete wave load analysis is required. Loads at 10 probability DNV-RU-SHIP Pt.5 Ch.7
level in North Atlantic environment. Sec.4 [2.1.2]
— Before direct wave load analyses results are available, design loads for
DNV-RU-SHIP Pt.5 Ch.7
the tanks can be determined according to the rule equations in DNV-RU-
Sec.4 [6.1.2]
SHIP Pt.5 Ch.7 Sec.4.
Stress analysis — A whole ship global three dimensional FE analysis is required. Hull DNV-RU-SHIP Pt.5 Ch.7
of cargo tanks model with cargo tanks and covers to be applied. Sec.21 [2.1.2]
and supporting
structures in way — The calculations including modelling, loading conditions, strength
of cargo tanks assessment and allowable stress checks shall be carried out according Sec.3
to the procedure described in this class guideline.
— Mechanical properties for the material of cargo tanks to be documented. DNV-RU-SHIP Pt.5 Ch.7
Sec.6
— For buckling analysis fabrication tolerances shall be considered. DNV-RU-SHIP Pt.5 Ch.7
Sec.21 [3.2.3]
Fatigue analysis — Fatigue analysis shall be carried out for the details listed in Sec.8 [7]. Sec.8 [7]
DNV AS
Classification requirements
Section 1
Cargo tanks and supporting structures in way of cargo tanks
Items to be
Task summary Reference
addressed
— The calculated fatigue life time in North Atlantic environmental DNV-RU-SHIP Pt.5 Ch.7
8 Sec.4 [2.1.2],
conditions during 10 wave encounters shall not have a fatigue damage
factor larger than CW=0.5. In any case the fatigue life shall not be less DNV-RU-SHIP Pt.5 Ch.7
than 25 years (the minimum design life for ships according to DNV Sec.4 [4.3.3] 7-9
rules).
DNV-RU-SHIP Pt.3 Ch.9
— S-N curves shall be relevant for the actual design detail. DNV-RU-SHIP Pt.5 Ch.7
Sec.4 [4.3.3] item 4.1
Crack — A fatigue crack propagation analysis shall be carried out for areas with
DNV-RU-SHIP Pt.5 Ch.7
propagation high dynamic stresses. The analysis shall consider propagation rates in
Sec.4 [4.3.3] item 6
analysis parent material, weld metal and heat-affected zone.
— Detailed requirements for the analysis are given in DNV-RU-SHIP Pt.5 DNV-RU-SHIP Pt.5 Ch.7
Ch.7. Sec.4 [4.3.3] 7-9
Leakage rate — Leak analysis to determine potential leakage rates shall be carried
DNV-RU-SHIP Pt.5 Ch.7
determination out as basis for design and dimensioning of the small leak protection
Sec.4 [2.4.2]
system.
Sloshing loads — Sloshing loads shall be based on model testing and/or numerical DNV-RU-SHIP Pt.5 Ch.7
sloshing analyses. Sec.4 [3.4.4],
DNV-RU-SHIP Pt.5 Ch.7
Sec.21 [5.1]
See also App.C.
Vibration analysis The potentially damaging effects of vibration on the cargo containment
system shall be considered.
— Determination of natural frequencies to be carried out. DNV-RU-SHIP Pt.5 Ch.7
— Forced vibration analysis may be carried out to show that no harmful Sec.4 [3.3.5]
vibrations will be excited by the propulsion system or other machinery.
Added mass of LNG to be considered as relevant.
Thermal analysis If not available from similar designs, steady state thermal analysis of the
cargo hold area and the cargo tanks shall be performed to
— determine steel temperature distribution in the tank system; skirt, DNV-RU-SHIP Pt.5 Ch.7
sphere and equator area as. Sec.21 [4.1.1]
— basis for structural integrity analyses of the cargo tank with respect to
yield and buckling in partial and full load conditions.
Transient thermally induced loads during cooling down periods shall be DNV-RU-SHIP Pt.5 Ch.7
considered for tanks intended for cargo temperatures below -55 °C. Sec.21 [4.1.2]
DNV AS
Classification requirements
Section 1
Cargo tanks and supporting structures in way of cargo tanks
Items to be
Task summary Reference
addressed
Partial secondary — Leak rates shall be determined for the purpose of sizing the partial DNV-RU-SHIP Pt.5 Ch.7
barrier and secondary barrier in order to keep the temperature of the hull structure Sec.4 [2.2.1]
primary barrier at a safe level. and DNV-RU-SHIP Pt.5
small leak Ch.7 Sec.4 [2.4.2] item
protection system 1
— Insulation system for cargo containment system to be documented with DNV-RU-SHIP Pt.5 Ch.7
respect to material and design. Sec.4 [2.8]
— Verification that any leaks are contained by the drip tray for at least DNV-RU-SHIP Pt.5 Ch.7
15 days and that the leaked gas can be disposed of in a safe way. Due Sec.4 [2.2.6],
account shall be taken of liquid evaporation, rate of leakage, pumping DNV-RU-SHIP Pt.5 Ch.7
capacity etc. Sec.4 [2.5.2] and
DNV-RU-SHIP Pt.5 Ch.7
Sec.4 [4.3.3] item 6.3
Material — Tank materials, tank shell and secondary drip tray. DNV-RU-SHIP Pt.5 Ch.7
Sec.4 [5.1.2]
7 Assumptions
DNV AS
SECTION 2 HULL STRUCTURAL ANALYSIS
Section 2
1 Introduction
1, 2, 3 4
1, 2, 4 3
3 1 tank empty, other tanks full ULS design.
1, 3, 4 2
2, 3, 4 1
DNV AS
No. Loading Condition Full Empty Part full Comment
Section 2
1, 3 2, 4
2, 4 1, 3
4 2 tanks full, 2 empty ULS design.
2, 3 1, 4
1, 4 2, 3
1 2, 3, 4
2 1, 3, 4
5 One tank full, others empty ULS design.
3 1, 2, 4
4 1, 2, 3
LC3, LC4 and LC5 consist of several combinations of full and empty tanks designed to give relatively high
bending moments and shear forces on the ship for ULS design. In addition to LC1 and LC2, normally LC3 is
specified together with either LC4 or LC5. These conditions reflect the degree of operational flexibility which
usually is requested to build into the ship. Any limitation to the loading conditions or load combinations will
be noted in the loading manual. For LC7 applied filling levels and corresponding time fractions will depend on
operational profile.
2.1 General
Procedures of cargo hold FE analysis for generic ship types are given in DNV-CG-0127 Sec.3. In the
following, additional description is given with particular consideration on gas carriers with spherical tanks.
2.2 Application
An FE model is made to represent the cargo holds to such an extent that the yield and buckling strength of
the double hull structure could be confirmed to a satisfactory level for the complete cargo area. This requires
as minimum FE analysis of midship holds and no.1 hold (the foremost hold).
Whether or not the aftmost cargo hold analysis should be carried out is dependent on the geometry of the
double hull and the support of the tank in this hold. This should be assessed and decided on a case-by-case
basis to the satisfaction of the Society.
DNV AS
Section 2
Figure 1 Model extent for individual cargo hold analysis
Figure 2 Example of 3 cargo hold model within midship region of a LNG carrier (shows only port
side of the full breadth model)
DNV AS
a) The curvature of the hopper ring girder shall be modelled to give a realistic presentation of the forces
Section 2
in the flange as well as the normal and shear stresses in the web, see Figure 3. The flanges can be
modelled with 2-node beam elements.
b) The stool frames, if any, should be modelled in the same manner as the hopper frames.
c) The interface between the inner side and the foundation deck shall be modelled with supporting structure
(brackets, girders) included. This is to give a realistic presentation of the vertical stresses in the inner
side for buckling assessment.
DNV AS
Section 2
Figure 4 Typical mesh arrangements for the cargo tank for use in cargo hold analysis
DNV AS
— The web of the ring girder may be modelled with 4-node shell elements, typically 2 or 3 elements over the
Section 2
height of the girder depending on the height, and the flange may be modelled with truss elements or 2-
node beam elements.
Table 2 Design load combinations for cargo hold analysis in midship area
Application
% of % of
Draught
Dynamic load
No. Loading pattern perm. perm.
cases/comments
SWBM SWSF
Hull
Mw=0
and 100%
LC1
tank
TSC (hog.)
≤100% Tank load (S)
sup. Sea Pressure (S)
DNV AS
Section 2
Application
% of % of
Draught
Dynamic load
No. Loading pattern perm. perm.
cases/comments
SWBM SWSF
Mw=0
100%
LC2 Hull TSC1) (hog.)
≤100% Tank load (S)
Sea Pressure (S)
Hull
Mw=0
and 100%
LC3
tank
TA2)3) (sag.)
≤100% Tank load (S)
sup. Sea Pressure (S)
Mw=0
100%
LC4 Hull TBAL (hog.)
≤100% Tank load (S)
Sea Pressure (S)
DNV AS
Section 2
Application
% of % of
Draught
Dynamic load
No. Loading pattern perm. perm.
cases/ comments
SWBM SWSF
HSM-2, FSM-2
100% HSA-2, BSR-1P, BSR-2P,
LC6 Hull TSC1) (hog.)
≤100%
BSP-1P, BSP-2P, OST-1P,
4)
OST-2P, OSA-1P, OSA-2P
DNV AS
Section 2
Application
% of % of
Draught
Dynamic load
No. Loading pattern perm. perm.
cases/ comments
SWBM SWSF
Mw=0
Inclination of 30° with tank
Cargo
pressure corresponding
tank
5) 100% to g and a transverse
LC9 and TSC (hog.)
≤100%
acceleration equal to
tank
sup.
g·sin30°=0.5g (S)
Inclined static sea pressure
(S)
Mw=0
Inclination of 30° with tank
Cargo pressure corresponding
tank
5) 100% to g and a transverse
LC10 and TA2)3) (sag.)
≤100%
acceleration equal to
tank
sup. g·sin30°=0.5g (S)
Inclined static sea pressure
(S)
DNV AS
Section 2
Application
% of % of
Draught
Dynamic load
No. Loading pattern perm. perm.
cases/ comments
SWBM SWSF
Notes:
1) Maximum draft with one cargo tank empty may be used instead of scantling draft TSC, if this is stated as operational
information in the loading manual.
2) Actual minimum draft with one cargo tank full from the loading manual.
3) Draught not to be taken greater than minimum of 2+0.02L and the minimum ballast draught.
4) Applicable for ships symmetrical to the centre line. For ship asymmetrical to the centre line, the beam sea and
oblique sea EDWs for both port and starboard shall be applied on the model. This means that BSP-1S, BSP-2S,
BSR-1S, BSR-2S, OSA-1S, OSA-2S, OST-1S and OST-2S EDWs are required in addition to the EDWs given in the
table.
5) This load case is to evaluate cargo tank and supporting structures in way of cargo tank. Therefore it is relevant only
when this case is not covered by other mean such as global analysis for example.
6) If it is confirmed that one tank full and others empty is not relevant condition, this load case can be waived or
modified.
DNV AS
Section 2
2.5 Analysis criteria
2.5.1 General
Acceptance criteria are given in DNV-RU-SHIP Pt.5 Ch.7 Sec.21 [3.1], for ULS of the hull structures. See
DNV-CG-0127 for acceptance criteria.
For cargo tank and it's supporting skirt structure, acceptance criteria are given in DNV-RU-SHIP Pt.5 Ch.7
Sec.21 [3.2] for ULS and in DNV-RU-SHIP Pt.5 Ch.7 Sec.21 [3.4] for ALS.
DNV AS
SECTION 3 GLOBAL ANALYSIS
Section 3
1 General
2 Structural model
Girders One element over the height. Beam elements may be used for stiffness representation
Hopper plate One to two elements over the height depending on plate size
All areas not mentioned above should have equal element sizes.
DNV AS
However global model can be made with finer mesh for entire cargo area or for selected area. Finer mesh
Section 3
size is normally in accordance with the mesh size of cargo hold model described in Sec.2. But it should be
decided based on the purpose of the calculation given in each section from Sec.4 to Sec.11.
3 Analysis steps
In a direct global structural analysis, state-of-the-art analysis procedures should be used. The following steps
of analysis will normally be included:
a) selection of design environment in form of a wave scatter diagram
b) hydrostatic (still water) load analyses based on selected loading conditions from the loading manual
c) hydrodynamic motion and load analysis with a recognised hydrodynamic analysis programme
d) development of an integrated hull, tank structure and tank cover global structural FE model for the whole
ship
e) automatic load transfer (pressures) to the global structural FE model
f) global structural response analyses
g) FE model refinement as necessary for ULS (yield and buckling) analysis
h) carry out yield capacity checks, buckling checks and hull girder capacity check as required
i) carry out fatigue screening analysis based on global model with cargo hold mesh
j) zoom in at fatigue critical areas and carry out component stochastic and/or full spectral (stochastic)
fatigue analyses to determine fatigue lives.
These analyses are described in DNV-RU-SHIP Pt.6 Ch.1 Sec.7 and DNV-CG-0130 and will therefore not be
repeated here.
Some examples of direct analysis of spherical type LNG carriers are published in /12/ and /13/.
Figure 1 Global FE model with pressure transferred from the wave load analysis, see /13/
DNV AS
SECTION 4 STRENGTH ANALYSIS OF CARGO HOLD COVERS AND
Section 4
SUPPORTING STRUCTURE
1 Introduction
1.1 General
1.1.1 The following treats hemi-spherical cargo hold covers with respect to:
1) buckling strength
2) support at main deck.
Fatigue is covered in Sec.9.
1.1.2 Hemi-spherical cargo hold covers are normally not stiffened and hence very sensitive to compressive
shell forces. The contributing load is the ship’s global bending moment combined with torsion moment and in
addition possible internal overpressure or under pressure in the cargo hold.
1.1.3 The cargo hold covers provides torsional stiffness to the ship hull girder. Without covers the hull girder
may be subjected to significantly increased deformations.
1.1.4 The deformation of the strength deck due to hull girder bending generates vertical compression and
tension stresses in the covers. This results in local bending deformations and high local stresses in the
connection area between the cover and the ship deck and shall be checked with respect to fatigue, see Sec.9
[7.5].
1.1.5 The covers, apart from providing additional torsional stiffness of the hull girder, do not contribute much
to the global bending strength of the ship, but will, due to continuity between the cover and the weather
deck, force considerable local loads onto the supporting structure. The magnitude of these loads is dependent
of the magnitude of the longitudinal bending moment and the stiffness of the cover compared to the hull
girder.
The stiffness of the cover makes a minor contribution to hull girder bending. Hence, the hull girder section
modulus should be determined based on hull girder without any effect from the covers.
DNV AS
Section 4
2.3 Load cases and boundary conditions
The boundary conditions should be the same as specified for global analysis in Sec.3.
Following load cases at the considered hold should be taken into consideration as a minimum for analysis of
the covers:
1) maximum vertical bending moment
2) maximum horizontal bending moment
3) maximum torsion moment
4) maximum internal/external pressure.
Figure 1 Bending of cover. The vertical sum of forces acting on the cover ΣF = 0
DNV AS
Section 4
Figure 2 Principal stresses in tank cover for a sagging bending moment condition
DNV AS
Section 4
Figure 3 Principal stresses in tank cover when subjected to torsional loads
For conical or cylindrical parts, the pressure will contribute with hoop and axial stresses equal to:
For a spherical cover, the pressure induces equal axial and hoop stresses in both directions equal to:
where:
DNV AS
Section 4
P
Figure 4 Axial stresses and hoop stresses in spherical tank cover when subjected to overpressure
3.1 General
In the following acceptable buckling codes for cylindrical and spherical parts of cargo hold covers are given.
For other configurations, the applicable buckling code should be specially considered.
The following should be applied when considering the cover with respect to AC-II based on the method given
in App.D:
1) all load safety factors (i.e. gamma) shall be taken as 1.0
2) the material safety factor shall be taken as 1.0.
DNV AS
Section 4
, where the stress designations are defined in the following.
where:
2
σ1 = largest compressive principal membrane stress, positive in compression, in N/mm
σ2 = principal membrane stress normal to σ1 (compressive or tensile), in N/mm2
2
σE = elastic buckling stress, in N/mm .
where:
where:
In lieu of more detailed information, the combined imperfection amplitude should be taken as:
As an alternative, when the actual tolerance area known, δ should be taken as:
DNV AS
Section 4
i.e. as given in App.D.
Allowable usage factor shall be applied as given above regardless of the applied imperfection amplitude.
4 Supporting structure
Due to the enforced bending of the cargo covers from the longitudinal still water and wave bending moments,
a considerable load will be transferred from the covers to the support in way of the strength deck. The covers
should have a rigid/stiff connection to the hull structure, and the brackets and connection structure should be
modelled to a sufficient detail necessary for obtaining reliable results.
For analysis acceptance criteria concerning brackets and supporting structure, see [2.4.3].
DNV AS
SECTION 5 STRENGTH ANALYSIS OF THE SPHERICAL CARGO TANK
Section 5
1 Introduction
In the following, acceptable procedures for strength analysis of the cargo tank are outlined. Procedures for
strength analysis of pipe tower and skirt constructions are given in Sec.6 and Sec.7 respectively. To perform
an effective analysis of the complete cargo tank system, it is advised that Sec.5, Sec.6 and Sec.7 are all
considered coherently.
DNV AS
NVSKIRT can be used for check of the buckling strength based on stresses from the FE analysis. Analysis
Section 5
procedures for the pipe tower are given in Sec.6.
4) Fatigue and crack propagation analysis can be carried out according to Sec.8.
3 Prototype testing
Prototype testing shall be carried out as required in DNV-RU-SHIP Pt.5 Ch.7 Sec.21 [6.1.1] to verify that the
allowable stresses set by the Society's rules are adhered to, and the buckling strength of the spherical tank
system is in order. Two physical full scale tests are required to be carried out.
3.2 Test no. 2: Verification of the static strength of the spherical tank
Each tank shall be tested with a combination of fresh water filling and air pressure determined as follows:
The tank shall be filled with water with no additional air pressure on top until the tank wall approaches the
design buckling limit just above the liquid surface. Then gas pressure shall be applied on top to a level that
gives an equivalent membrane stress in the tank wall not larger than the following, see DNV-RU-SHIP Pt.5
Ch.7 Sec.21 [6.1.1]:
where:
2
σm = allowable von Mises primary membrane stress, in N/mm , see DNV-RU-SHIP Pt.5 Ch.7 Sec.21
[3.2.2]
2
ReH = minimum specified yield stress at room (test) temperature, in N/mm .
Figure 1 Typical pressure diagrams for tank testing and emergency discharge conditions
DNV AS
Section 5
4 Design loads for allowable stress and buckling analysis
*Both the hull girder bending moment contribution and local double bottom bending shouldbe included.
Internal overpressure, not less than Po = 0.025 MPa (0.25 barg) which
1.Fully loaded tank
equals the maximum vapour pressure or the maximum allowable pressure
condition:
setting of the pressure relief valve (MARVS).
DNV AS
Loading condition Design pressure Applicability
Section 5
* This is not a DNV rule requirement. The emergency discharge condition has however evolved to become a standard
loading condition. (Mandatory if only one cargo pump is fitted in each tank).
The interaction forces from the still water bending moment should be calculated as described in [4.5] taking
into account the actual still water and wave bending moment, draught and acceleration for the given loading
condition and tank position.
DNV AS
However simplified procedure to determine interaction forces may be acceptable by the Society case by case.
Section 5
When skirt structure is designed close to boundary of cargo hold so that double bottom bending effect is
considered ignorable, it may be acceptable to exempt it for the interaction force calculation.
If unit loads are applied in global analysis, the results can then be scaled to the design values (still water and
wave) by the actual bending moments, design draughts and accelerations in the tank centres as follows.
where:
DNV AS
An example of shell force interaction results from hull girder bending are shown in Figure 3.
Section 5
Figure 3 Typical distribution of the axial and circumferential interaction force in the tank shell
1
The system was only given a sinusoidal, horizontal motion and represents sway motions only. Roll
motion was not included.
DNV AS
Section 5
S1 S2 — Standing wave (S1, S2) (same as for prismatic
tanks)
— Breaking wave (BR): the velocities of liquid
particles at the surface layer of the wave crest are
DS
higher than the propagation velocity of the wave.
— Swirling (SW): awave crest rotating along
latitudes of the sphere.
— (DS):it has been observed during testing that a
BR SW standing wave can be combined with both lateral
and vertical liquid movements resulting in a wave
top at the middle of the tank.
At low frequencies:
For a given frequency of excitation well below liquid resonance, the height of the waves in the tank increases
until a stable maximum is reached after a relatively long time.
DNV AS
4.6.4 The circumferential critical shell force
Section 5
Figure 5 The transverse force on the tank and the resulting acceleration of the liquid
For the purpose of buckling assessment of the tank, the compressive forces in the tank shell will normally
only be critical when the surface of the cargo is below the equator profile. The weight of the cargo will
then generate a significant meridional tension as the southern hemisphere is ‘hanging’ in the equator.
The circumferential shell force will however, as there is virtually noexternal lateral load on the shell, have
approximately the same magnitude as the meridional shell force but with opposite sign i.e. compression. To
take this into account a minimum external design pressure of 0.005 MPa shallbe applied as a safety measure,
see Table 2. The surface of the liquid may also be below the equator for filling ratios above 50 % if the liquid
surface has a skew angle to the equator plane. Filling ratios above 50 % with a certain skew angle of the
liquid plane will then give higher tension in the critical area due to the increased weight of the cargo and
hence higher circumferential compression which can be seen from the shell equation below.
where:
FE analysis has been performed to establish a procedure to calculate the compressive stresses in the shell,
2
see/3/. Four critical calculation points were decided on; 50º, 60º, 70º and 81º on the southern hemisphere,
where non-dimensional maximum compressive membrane forces were calculated for different skew angles β
(angle between the liquid surface and the equator plane), see Figure 7 to Figure 11.
2
The positions of the chosen calculations points were entirely based on the locations of the weld seams
between changing plate thicknesses on the original spherical LNG carriers built at the Moss Rosenberg.
DNV AS
Section 5
Fx at 10-8probability level. The force
Figure 6 Long term distribution of horizontal sloshing force
is non-dimensional and may be applied to any ship length, tank diameter and cargo density (d =
sphere diameter)
DNV AS
Section 5
Figure 7 Maximum compressive circumferential membrane force at φ = 81°, t1/t2 ≥ 1.3
Note that two graphs exist for the location 81º. At the reference tank this location had a step in plate
thickness which also has been the case for subsequent vessels.
Figure 8 Maximum compressive circumferential membrane force at φ = 81°, t1/t2 < 1.3
DNV AS
Section 5
Figure 9 Maximum compressive circumferential membrane force at φ = 70°
DNV AS
Section 5
Figure 11 Maximum compressive circumferential membrane force at φ = 50°
For stretched spherical tanks, or tanks with other significant geometrical differences from the test tank,
similar analysis as carried out in/3/ may have to be carried out for at least one position of the subject tank to
establish the critical membrane forces.
Alternatively, tests and/or analyses according to App.C should be carried out to determine sloshing forces
together with direct structural FE analyses for determination of the shell forces.
4.6.5 Determination of the critical heel angle and apparent transverse acceleration
In the following a procedure for calculating the apparent transverse acceleration is presented, as first
outlined in /1/. The procedure uses results from a reference ship but takes into account the changes in
sloshing forces for different ship lengths and speeds, tank positions and loading conditions.
1) The original ship used in the sloshing tests, see/1/, had certain geometrical properties which shall be
maintained in order to establish a reference value for the transverse acceleration, ayR= ay of the actual
ship. As the sloshing tests summarized in Figure 4 were carried out by a horizontal regular motion
with no roll effect included, the weight component due to rolling (heel angle) shall be omitted in the
calculation. A modified ship speed corresponding to the Froude number of the original ship shall be
DNV AS
applied as given below and the values GM/B, z/B and x/L0 must be taken equal to those of the original
Section 5
ship. The ship length should be taken equal to the length of the actual ship L0.
where:
κ =
The roll motion amplitude may generally be calculated according to DNV-RU-SHIP Pt.3 Ch.4 Sec.3
[2.1.1].
3) The tangential transverse weight component due to rolling may be calculated as:
Fθ =γ·V·θ
4) Calculate the transverse acceleration,ay, according to the same equations as given in step 1, but with
the ship speed and the geometrical relations for the actual ship.
3
5) Find the reference value for the transverse force on the tank, FxR = Fx/γ·d , for the actual ship length
according to Figure 4 for h/d = 0.29, 0.50 and 0.65.
6) Calculate the amplification factor for the dynamic effect due to sloshing. The amplification factor is
assumed to be constant for each filling level.
7) Calculate the sloshing force for the actual ship according to:
by scaling the sloshing force with the accelerations determined by the rule equations.
DNV AS
8) Transverse sloshing force and apparent transverse acceleration are calculated as:
Section 5
The above procedure is entirely based on empirical equations and is suitable for an initial design evaluation.
However, final verification of the actual ship tank design should be based on direct hydrodynamic
calculations. Hence, the roll angle in steps 3 and 4 above shouldbe based on direct analyses. Please note that
the horizontal accelerations used for scaling of the sloshing force Fx should be based on the equation above
and be calculated without the weight component (ψ = 0) as this is included in the Fθ term in step 4.
DNV AS
Section 5
DNV AS
4.6.6 Determination of the critical circumferential force
Section 5
The objective of the following is to present a calculation procedure to obtain the maximum compressive
3
force in the tank shell just above the liquid surface plane in a heeled condition. The procedure is best
suitable for computer programming as it involves an iteration routine. The procedure is incorporated in the
NVSPHERE programme, see /4/, and is illustrated in Figure 13 below:
3
Often denoted as the minimum circumferential membrane force in reference literature.
DNV AS
Section 5
4.7 Thermal loads
For the yield and buckling analysis of the cargo tank system, for any loading condition where LNG is present
in the tank, thermal stresses should be calculated based on a stationary temperature distribution in the tank
and skirt. Warm weather conditions should be used for the estimation of the temperature distribution which
should be based on a heat flow analysis at full load condition, see Figure 14. The ambient temperatures
are normally to be taken as given below, but may be taken lower or higher with due consideration of the
intended trade, see DNV-RU-SHIP Pt.5 Ch.7 Sec.7 [2.1.1]:
— air temperature: 45º C
— sea temperature: 32º C.
The cargo temperature shall be taken as -163º C.
DNV AS
Section 5
5 Finite element analysis of the spherical tank
The following describes one acceptable procedure for how to perform a finite element analysis of a spherical
cargo tank for verification of the allowable stress level and buckling strength. Alternative methods may be
applied subject to special consideration by the Society.
The analysis shouldconfirm that the stress levels are acceptable when the tank is loaded in accordance with
described design conditions.
Any recognised calculation method or computer program may be applied, provided the combined effects
of axial, bending, and shear deformations are adequately considered. Strength analysis carried out in
accordance with the following will be accepted as a basis for class approval in general.
5.1 General
In general the element model must provide results suitable for evaluation of:
a) The general membrane stress level of the sphere and the skirt (see Sec.7 where modelling and stress
assessment of the skirt is further elaborated) with respect to the allowable design stresses.
b) The local stresses in way of the:
— equator profile, Figure 15 A
— upper tower/dome connection, Figure 15 B
— lower tower connection, Figure 15 C
— the STJ connection, see Sec.6 for allowable stress and fatigue of the skirt.
c) The membrane stresses of the sphere and the skirt with respect to buckling.
One of the two following different approaches is typically applied;
1) Full geometry shell element model with sub models for fine mesh solid element idealization of the stress
concentration areas.
2) Axisymmetric strip model with solid elements including fine mesh idealization.
DNV AS
The tower connection area, the dome connection area, the equator profile and the STJ connection in way of
Section 5
scallops or other areas where stress concentrations may be expected, may be modelled by very fine meshed
solid element models. A global shell element model may be used and refined with sub-models as necessary
in detailed areas.
Where radii of welds and equator groove are modelled, the mesh density may normally be 5 elements along
the radius (i.e. 20 elements around a full circle).
For determination of hot-spot stresses with fine mesh sub-models, see DNV-CG-0127.
Axisymmetric tank models may be used if they allow for the application of non-axisymmetric loads.
The arrows in Figure 15 identify typical areas of stress concentration. Some are subject to yield stress
assessment and all are subject the fatigue limit assessment.
DNV AS
Section 5
Figure 16 2-D plane shell element models of the dome and tower connections illustrating typical
mesh density
DNV AS
Section 5
Figure 17 Groove detail of equator profile with cross section illustrating typical mesh density for
solid model
DNV AS
Section 5
5.5 Design load cases for ULS stress assessment
5.5.1 General
For ULS design stress check, the model should be analysed for the following load cases:
Note: Tank test condition to be checked with respect to yield and buckling according to [3] and Table 7 (with internal
pressure).
DNV AS
1) as a skewed acceleration field,aR, directly to the FE model, or
Section 5
2) alternatively, if the FE software does not have this facility, apply component loads as outlined below:
in alternative 1 above the skewed acceleration field aR is combining the LC 1S b (static) and f-g-h (dynamic)
components and will produce the same resulting pressure distribution as will result from alternative 2 below.
For alternative 2 load components are applied with reference to an upright coordinate system.
To be able to combine the static and dynamic stresses as described in [5.5.3], static and dynamic load
components (LC1S a, b, c, and f, g, h, in Table 3) may be applied as shown below as separate loads (full
tank). For definition of symbols, see App.B:
1) Static pressure components due to weight of cargo, see LC 1 b above:
Pstatic =γR(1+cosφ), (=2γR at the bottom for a full (α=π) perfect spherical tank)
2) Dynamic pressure components due to longitudinal acceleration of cargo, refer to load case 1g. above:
3) Dynamic pressure components due to transverse acceleration of cargo, refer to load case 1f. above:
Figure 18 The pressure components resulting in the skew symmetric load condition
— at CL
— at 45º to CL
— 90º outboard.
DNV AS
For an ocean-going traditional spherical LNG carrier the following may be applied:
Section 5
— With respect to the seagoing fully loaded condition, for the sake of the overall membrane check of the
sphere, only the outboard 90º position needs to be documented. This is due to the dominating transverse
acceleration.
— For the sake of peak stress investigation in way of the tower connections, the equator profile and the STJ
joint, two positions should in principle be assessed and documentedat CL and at 90º.
For new and novel designs where the vessel motions may be expected to deviate from that of the traditional
hull design or where the stress picture in the tank may be expected to deviate from the traditional tank
design, all three positions should be documented.
Note that the presented method for stress summation assumes that the most severe directly calculated
accelerations are adopted in the analysis. Total static and dynamic stresses in the meridional and
circumferential directions are then combined with square root of the sum of the squares (SRSS)summation
assuming the stress components are statistically independent, see DNV-RU-SHIP Pt.5 Ch.7 Sec.4 [4.2.3]:
For thin shell elements Additional for thick shells and solid elements
For the emergency discharge conditions, all the applicable load cases are static and may be combined by
directly linear summation in the FE analysis.
For the full load condition, the stresses given by the static and dynamic load cases shouldbe split in separate
components and combined according to the above equations as follows:
where:
sw = self weight
sp = static pressure
sm = static mass of cargo
sf = static interaction
st = stationary temperature
df = dynamic interaction.
Thin shells:
DNV AS
Section 5
Thick shells and solids:
All equivalent stress alternatives shall be verified against the ULS design stress criterion (allowable code
stresses).
For the membrane check atθ = 90º, shear stresses exist only for the longitudinal acceleration (ax) condition.
The shear stress to be included in the von Mises criterion (see Table 5) will hence only be:
Table 5 Stress combination for the membrane design stress capacity assessment at 90º outboard
σφ σθ τ σvm
1 => σvm‐1
2 => σvm‐2
3 => σvm‐3
4 => σvm‐4
The required shell thickness may finally be calculated according to the following:
If the loads are applied according to alternative 1 above, [5.5.2], the accelerations are combined with the
acceleration ellipse and considered to act simultaneously in the direction of the resulting acceleration vector
aR. The following linear stress combination can then be applied.
Please note that even though aR contain the staticg component and the dynamic acceleration components,
the term aRσi sm is considered to be a dynamic term as it will move like a pendulum with the movement of
the ship.
The stresses associated with the resulting acceleration aR should be determined by applying the static and
dynamic mass loads in the direction of aR in the FE analysis.
DNV AS
5.5.4 Design stresses for ULS stress assessment
Section 5
The allowable stresses for the different parts of the cargo tank shall be taken according to DNV-RU-SHIP Pt.5
Ch.7 Sec.21 [3.2.2].
Table 6 Load cases for ULS buckling assessment of the spherical tank
DNV AS
Load cases Loads Notes
Section 5
a) Tank system self-weight (tank material, insulation
and piping etc.).
b) External pressure 0.005 MPa.
c) Static interaction force due to still water bending
moment and external static pressure [4.5.2].
d) Partial filling of cargo: static and dynamic part
LC 3B:
combined based on resulting skewed acceleration
Sea going with part filling
aR from the acceleration ellipse, see Figure 11.
e) Dynamic interaction force due to wave bending
moment and external wave pressure [4.5.2].
Characteristic forces N1C and N2C are found by setting the load factors equal to one (γfi= 1.0). For definition
of symbols see App.D.
The largest compressive force should be taken as the largest compressive primary force component
regardless of whether it is a meridional or a circumferential force. All realistic combinations of signs on
acceleration (aZ) and force components (N1df andN2df) should be checked in order to determine the largest
compressive value.
DNV AS
The shell forces N1 and N2maybe determined from FE analyses with the tank subjected to the combined
Section 5
skewed acceleration field at the actual liquid filling level.
Systematic analyses of a sphere subjected to a range of filling levels and skew angles have been carried out
in the past, see /3/. The key results are shown in Figure 7 to Figure 11 and are incorporated in the analysis
procedure in the design programme NVSPHERE, see /4/.
For a pure spherical shell:
N1 +N2 + PovR = 0
N2 = – N1 – PovR
where:
Pov is the external overpressure usually taken as 0.005 MPa.
The partial safety factors/load factors are given as follows, see App.D:
γf1 = 1.2 for forces caused by static weights and inertia forces
γf2 = 1.4 for forces caused by deflection of the foundation deck
γf3 = 1.5 for forces caused by external overpressure.
5.6.3 Load application of the static and dynamic stresses
If the FE software allows for it and a half (180°) or a full (360°) circumference model of the tank is made,
the skewed pressure load distribution can be obtained by applying the loads directly in the direction of
the acceleration vector aR. Then only one contribution that includes both static and dynamic liquid cargo
pressures will be needed, see LC 3B d above.
The same pressure distribution can be obtained by applying pressure components in an upright coordinate
system as shown below:
The equations for component based load application have the same form as for the allowable design stress
model, however somewhat difference due to the partial filling.
1) Static pressure components due to weight of cargo, refer to load case 3b. above:
2) Note that ay here is the instantaneous value of the apparent transverse acceleration. The maximum
value should be determined as shown in Figure 12 and the instantaneous value is found by the
procedure in Figure 13:
DNV AS
1) The buckling code given in App.D is written in a partial safety format taking into account loads, load
Section 5
factors, shell geometry and thickness, imperfections and safety factors such that:
f (loads, load factors, shell geometry and thickness, imperfections, safety factors) ≤ 1.0
This means that the whole code shall be seen in its own context and it is in principle not acceptable to
mix this code with other codes with other load factors, imperfections, safety factors or other boundary
conditions. The equations for h,
is a direct extract of the Koiter theory while the post buckling factor b,
is derived from the same theory and adjusted according to laboratory test results, see /11/. The b-factor
is dependent on several parameters such as geometrical proportions, boundary conditions and load
application. Minimum thickness requirement, t >R/750 for aluminium, is absolute and applies for the
spherical shell only.
2) The tolerances may be set by the designer subject to the building manufacturer’s production capabilities.
The required tolerance may be decided by the Society taking into account the manufacturer’s experience
from previous production, shell thickness and type of material. See App.D.
3) All four combinations of σ10 and σ20 should be checked to identify the most severe situation. Note
however that the buckling control of the sphere under bi-axial loading is only relevant when one principal
stress is in compression. Compression is defined as positive stress in App.D:
σ10 – largest positive principal stress (always positive, i.e. compression) in N/mm2, design stress value
σ20 – principal stress normal to σ10, positive(compression) or negative(tension) in N/mm2, design stress
value.
For the southern hemisphere, partial fillings with skew angle of the liquid plane are the governing loads
and the area with the highest compressive stresses is then found in way of the liquid surface. In this
case σ10 = σθ.
For the northern hemisphere, the empty seagoing condition will be the governing load. In this case σ10 =
σθ.
6 Summary
The following is a short summary of applicable loading conditions and acceptance criteria for the allowable
design stress assessment and buckling strength assessment of the spherical cargo tank.
DNV AS
Table 7 Applicable loading conditions and acceptance criteria for allowable design stress and
Section 5
buckling strength assessment
pressure,Minimum0.025 MPa
Acceptance criteria
External overpressure,
Self weight of system
Dynamic interaction
Minimum0.005 MPa
Acceleration of LNG
Full
X X X X X X X X
load
EM I X X X (X) X
stress assessment
Allowable design
Acc. to [5.5.4]
EM II X X X (X) X
Tank
X (X) X
test
Empty
Buckling strength
X X X X X
tank
Acc. to App.D
assessment
Partly
X X X X X X X X X
filled
7.1 Introduction
In order to increase the cargo capacity and to reduce the void space in the traditional spherical tank LNG
carriers other configurations of independent B-tanks have been developed.
— One is the stretched spherical cargo tank with a cylindrical ‘belt’ right above the equator profile, see
Figure 17.
— Another is a cylindrical tank with to ro-spherical top and/or bottom.
DNV AS
— Combinations of spherical, cylindrical, ellipsoidal and to ro-spherical forms.
Section 5
Common for tank systems deviating from the ideal spherical shape is that the tank shell needs to be thicker
to compensate for the extra membrane and bending stresses in the tank shell. Cylindrical shells will in
general need to be twice as thick as spherical shells.
Figure 19 Principal arrangement of a stretched cargo tank with a cylindrical belt above equator
Applicable procedures for yield and buckling strength assessment of for different types of B-type cargo tanks
will necessarily depend on the geometrical properties of the tank and the support. The basis will always
be DNV-RU-SHIP Pt.5 Ch.7 stating the requirements for independent tank of type B both for constructions
primarily constructed of bodies of revolution, DNV-RU-SHIP Pt.5 Ch.7 Sec.21, and constructions primarily
constructed of plane surfaces, DNV-RU-SHIP Pt.5 Ch.7 Sec.20. In principle the procedure for allowable design
stress and buckling strength analysis will be similar for these two, but the critical load cases and buckling
criteria may vary. In the following, the two named tank constructions constructed of bodies of revolution will
be discussed briefly.
DNV AS
7.3.2 Determination of the apparent transverse acceleration
Section 5
The procedure as described in [4.6.5] is developed for conventional spherical cargo tanks. It may however be
assumed that, for a limited extension of the stretched part, the sloshing pattern will be somewhat similar to
that of the conventional design, and the described procedure may be applied with some adjustments:
1) For estimating (reading) the horizontal sloshing force from Figure 4 it is recommended that an equivalent
tank diameter equal to the total tank height be applied.
2) For the buckling analysis the true diameter (radius) of the hemispherical parts shouldbe used and the
density of the cargo be increased to reflect the higher pressure head (height of tank).
For tank constructions deviating more significantly from the true spherical shape new sloshing investigations
will normally be required, see App.C.
DNV AS
Section 5
where:
Local bending of the plate is considered small and is hence neglected. For simplicity is has further been
assumed that no shear stress is present i.e. pure sway motion and circumferential position at θ = 90º is
considered. For other positions where shear may be present, see /5/.
The elastic buckling resistance of an unstiffened circular cylindrical shell may be taken as:
where:
ψ ζ ρ
DNV AS
where:
Section 5
r,t and ℓ denotes the radius, the thickness and the length of the cylinder respectively, in mm.
The usage factor is given by:
where:
DNV AS
SECTION 6 LOADS AND STRENGTH OF THE PUMP TOWER
Section 6
1 Description
Early designs of spherical tanks incorporated a pump tower consisting of a 3D tubular truss structure with a
triangular overall cross-section, the three vertical members being cargo flow pipes. The tower was tapered
down to a pinned connection to the tank at each end. The earliest sloshing experiments carried out in the
1970’s, see /1/, were conducted with a model of such a tower mounted inside the tank.
No vessels with spherical tanks were actually built with this truss design of pump tower. Instead a cylindrical
tower was used, enclosing the pipework and protecting it from the fluid motion in the tank. A series of
sloshing experiments was also carried out in the 1970’s, see /2/ with this type of tower. In some early
designs the tower was supported on pads at its base and restrained horizontally by additional pads at the
base and near the top. In more recent tower designs the cylindrical tube is flared out to a larger diameter
at the upper end and being terminated as a tank dome as shown in Sec.5 Figure 15 B and C . The tower
is vertically supported at the top end. The base is welded directly to a seating consisting of a horizontal
diaphragm plate that is supported on a larger diameter cylinder that is in turn welded to the tank shell. The
horizontal diaphragm plate is relatively thin, thereby providing flexibility in the vertical direction that reduces
the vertical forces and moments that are transferred from the lower end of the pipe tower to the cargo tank
shell.
3 Sloshing loads
3.1 Introduction
Sloshing loads on the pump tower structure occur due to motion of the liquid inside the cargo tanks. The
sloshing loads should be considered for various filling levels according to the ship’s filling restrictions and
operating profile. Normally the foremost tank or the adjacent tank is considered to be critical for sloshing,
depending on their relative dimensions and motions. If the tanks are of equal size, then the foremost tank
will normally be the more critical one. If the foremost tank is smaller, then the adjacent tank may be critical.
Methods for estimation of the sloshing loads in partly filled tanks may be based on followings, see App.C:
— model tests for a particular ship
— computational fluid dynamics (CFD) analyses, to be approved by the Society
DNV AS
— semi-empirical/simplified method described below.
Section 6
3.2 Ship loading conditions
The ship loading conditions that are considered to be critical for sloshing motion in the tanks shall be
determined. Loading conditions that may need to be considered are:
— full load condition
— ballast condition
— part load conditions.
where γ is the specific weight of the LNG. The results shown were based on tests in which the model was
subjected to a horizontal excitation motion with fixed amplitude equal to 5% of the tank diameter d. This
represents the combined effect of the horizontal motion of the ship (sway and/or surge) and the horizontal
motion of the tank due to rotation of the ship (roll and/or pitch), but does not include the rotational motion of
the tank about its centre.
Figure 1 indicates the following:
— The maximum reaction forces at the tower bottom occur for a tank filling level h/d = 0.3.
— The reaction force at the tower top increases with tank filling level; the maximum tank filling level
investigated was h/d = 0.8 and it was concluded that the force was unlikely to exceed this value
significantly at higher h/d. Normal practice has been to assume that this is the worst case.
— The total horizontal force on the tower is greatest for h/d = 0.3 but is approximately constant for 0.3 <
h/d < 0.7.
Based on the same test series, it has been found that the largest bending effects and tower stresses occur
for a tank filling level h/d = 0.5.
On the basis of the above observations, it has been normal practice to consider the three filling levels h/d =
0.3, 0.5 and either 0.7 or 0.8.
DNV AS
Section 6
Figure 1 Lateral forces on tower due to horizontal motion, from /2/
DNV AS
short-term sea states are described by a wave spectrum. In general the Pierson-Moskowitz spectrum should
Section 6
be used in combination with a cosine squared wave spreading function.
For the relevant loading conditions, ship motion analyses should be carried out. The analyses are to provide
linear ship motion transfer functions, which can be used to generate ship motions for specified sea states.
The tank motions are then calculated by accounting for the distance from the ship’s rotational centre to the
centre of the tank.
For the sea states assumed to be critical with respect to sloshing motion, irregular time series of motion
should be calculated using the ship motion transfer functions. The sloshing loads on the pump tower
structure due to the irregular tank motions are then determined, using experiments or analyses.
The ship motion and dynamic sea/tank pressure loads can be considered as linear functions of the sea
state variables, and can be obtained using standard spectral analysis methods. The sloshing load, however,
is a non-linear function of the sea state variables and requires a different treatment. In principle the
determination of the long term distribution requires consideration of the full duration of all sea state
combinations described by the scatter diagram combined with a description of the relation between heading
and sea state. The procedure thus consists of the following steps:
— Step 1: an operational profile for the lifetime of the ship is determined.
— Step 2: ship motion analyses are performed for each relevant loading condition and heading.
— Step 3: short term sloshing load statistics are determined for the given filling levels and headings and the
relevant sea states, specified in terms of mean wave period and significant wave height, obtained from
the scatter diagram. The short term sloshing load statistics may be established either by model testing or
by numerical analyses using CFD, or by a combination of both.
— Step 4: the short term statistics are combined to give a long term statistical distribution of the tower
sloshing loads for each individual filling level accounting for the operational profile determined in step 1.
If a CFD approach is used in step 3, the liquid velocities at the tower may be calculated and the forces acting
on the tower may be determined using Morison’s equation. A CFD approach is most suitable for predicting
loads at medium and high filling ratios. At low fillings (h/d < 0.4) the sloshing motion is characterised by
steeper and breaking waves that may cause more abrupt loads on the pump tower, hence the CFD procedure
should be verified by the Society for its intended use and specified filling condition.
DNV AS
Section 6
Figure 2 Example of a coarse discretisation of a scatter diagram
DNV AS
3.5.2 Establishing the long-term distribution
Section 6
In order to establish the long-term distribution of sloshing loads, the cumulative distribution may be
estimated by a weighted sum over the sea states used for estimating the sloshing load. The long-term stress
range distribution is then calculated from;
where:
pij = the probability of occurrence of a given sea state i combined with a heading j
rij = νij/ν0
the ratio between the response crossing rates in a given sea state and the average crossing rate
For the case of unrestricted filling, special consideration of the long term value of the load should be made.
The FLS assessment should be carried out using the same long-term load distribution as applied for the ULS
assessment.
where Q(σ) is the probability of exceedance of the stress σ, h is the shape parameter, and q is the scale
parameter, defined as
where:
DNV AS
Section 6
3.6 Sloshing experiments and analysis
3.6.1 Available approaches
The sloshing loads on the pump tower resulting from the applied tank motion are a function of the liquid
velocities and accelerations. The velocities and, if possible, accelerations should be determined at several
vertical locations along the axis (height) of the tower.
The fluid forces on the pump tower shell structure may be assumed to be drag dominated, so that the fluid
acceleration is of secondary importance relative to the fluid velocity. The velocities and accelerations may be
determined by one of the following methods:
— experimental tests on scaled-down models
— analysis with (CFD)
— a semi-empirical approach based partly on earlier test results.
The forces acting on the full-scale pump tower can then be calculated using Morison’s equation, [3.6.5].
As an alternative to the above methods, experiments may be carried out to establish more directly the forces
on the tower as described in [3.6.7].
All components should in principle be established for the same ship motion conditions, of which the dominant
ones are those which induce resonant sloshing. However, a conservative estimate may be obtained by basing
the rotational motion component on the extreme motion of the ship, as indicated in [3.6.5].
DNV AS
The experiments should facilitate velocity estimates at about ten points along the vertical axis.
Section 6
This method assumes that the presence of the tower does not disturb the fluid flow significantly in the full-
scale application.
DNV AS
For conditions at or close to resonance the velocity at the surface, VSurface, may be estimated from Figure 4,
Section 6
U = VSurface.
Figure 4 Distribution of maximum fluid velocities at tower over liquid depth for horizontal
excitation amplitude equal to 5% of tank diameter d, from /2/
The velocities in Figure 4 refer to a horizontal excitation of the tank of 5% of the tank diameter at the
resonance sloshing period. For other magnitudes of the tank excitations at resonance the velocities in Figure
4 may be scaled according to
The magnitude of ηres may be determined from calculation of the long term distribution of horizontal tank/
ship motions, by only taking into account contributions from wave components with periods within ±10% of
the tank resonance period for the considered filling height ('filtered response spectra').
An estimate of the velocity caused by rotational motion due to roll may be obtained as:
DNV AS
Section 6
where θ is the roll amplitude and TR the period of roll motion VR should in principle be established for the
same ship motion conditions as VH, namely those which induce resonant sloshing. However, θ and TR may
be taken in accordance with the rules if not found by a direct ship motion analysis. Note that the estimate
may be unduly conservative if based on the extreme value of roll amplitude.
The post-processing should be carried out on each of these time series, in order to determine design bending
moment and reaction forces. From these, the critical time instants for the top and bottom supports and the
main structure may be determined. The velocity fields at each of these time instants should be selected for
the structural analysis, and the sloshing loads acting on the tower at these instants should be applied to the
structural model.
The loads on the pump tower segment can then be calculated using Morison’s equation, as described in /8/.
The force per unit length q(z) acting on the tower is
where ρL is the liquid density, Cm is an added mass coefficient, dt is the tower diameter, a is the particle
acceleration normal to the member axis, CD is the tower drag coefficient, and v is the liquid particle velocity
normal to the tower axis.
Values of CD and Cm should be determined for each structural member according to recommendations given
in /8/. For a cylindrical tube Cm is normally to be taken as 1.0.
where:
dt = diameter of tower in m
T = sloshing resonance period in second
UM = maximum sloshing velocity in m/s.
CD may be taken from Figure 5, see also /8/. For pump towers of aluminium or other
The drag coefficient
smooth material the lowest curve (k/dt < 10 ) may be assumed.
-4
The following alternative equation may be used to estimate the combined effect of the drag and acceleration
terms in the Morison equation:
If this equation is used for CD, the acceleration term in the Morison equation should be omitted. The equation
is also illustrated in Figure 5.
DNV AS
When the load on each segment is determined, a beam model may be used to determine top and bottom
Section 6
support forces and bending moments. The boundary conditions assumed in the model should reflect the
actual boundary conditions of the pump tower. Alternatively the loads may be applied to a finite element
model of the tower and tank.
Figure 5 Drag coefficient CD as function of KC for cylinders in waves, Re > 5.0*105. Note that dt in
this figure refers to the tower diameter
DNV AS
Section 6
Figure 6 Example of test setup for pump tower loads
The inertia effect may be found by measuring the acceleration by an accelerometer. The gravity and
buoyancy should also be accounted for. Then the fluid force on the segment is obtained.
4.1 General
Inertia loads on the pump tower are due to the accelerations of the vessel. Translational tank accelerations
due to the pitch and roll accelerations of the ship should be accounted for. The weight of liquid inside the
tower should be included in the calculation of inertia loads where appropriate.
Gravity loads are due to the self-weight of the pump tower, taking account of the roll and pitch angles for the
ship. When the gravity loads are calculated, the buoyancy of the tower should be deducted if appropriate.
For inertia and gravity loads, the pump tower in the foremost tank is usually the most critical, due to its
distance from the ship centre of motion. However, both the foremost tank and the adjacent tank shall
be considered, in order to determine which one is most critical with respect to the combined effect of
sloshing loads and inertia loads. The effects of sloshing and inertia load should be summed to provide a total
combined load.
For both inertia load and gravity load calculation, the weight of additional elements (structural members
and equipment) that are not included in the finite element model used for the response analysis should be
included as lumped or distributed masses.
DNV AS
Section 6
4.3 Design load
The long-term distribution of inertia and gravity loads should be determined as a weighted sum over all sea
states given in the scatter diagram and all headings, as explained in DNV-CG-0129. The loads to be used
for the ULS assessment should be taken as the load occurring once during the lifetime of the ship. The FLS
assessment should be carried out using the actual long-term load distribution.
For the relevant loading conditions, ship motion analyses should be carried out. The analyses are to provide
linear ship motion transfer functions, which can be used to generate ship motions for specified sea states.
The inertia and gravity loads may be calculated for a vertical axis located at the tower axis. The loads will
vary in the vertical direction, but for each vertical position the accelerations may be taken as equal for all
points on the tower circumference and for all internal components.
5 Thermal loads
5.1 General
Thermal loads are due to thermal shrinkage of the pump tower material in the low temperature condition,
relative to the room temperature condition. However, thermal loads on the pump tower are generally much
smaller for spherical tank systems than for membrane tank systems as the tower and tank materials are
generally the same or very similar.
The temperature effect, if present, is most important for the upper and lower supports of the pump tower. If
thermal loads should be considered, the temperature distribution over the height of the pump tower should
be determined for each filling level. For filling levels h/d > 0.7, the temperature can be taken as constant
and equal to -163ºC. For filling levels h/d < 0.2, the temperature of the submerged part of the pump tower
can be taken as -163ºC, while the temperature of the non-submerged part can be assumed to vary linearly
from -163ºC at the liquid surface to -30ºC at the top of the pump tower.
6 Combination of loads
DNV AS
If for some reason more dynamic force components should be included these can be added using a square
Section 6
root sum of squares summation.
The static loads due to gravity and thermal gradients should be added:
The total load is found by adding the static loads to the total dynamic load:
It should be noted that both positive and negative values of the dynamic loads shall be considered when the
total load is calculated, i.e. ±Finertia. For the allowable stress check, the maximum stress is of interest, while
for the buckling check the maximum compression stress is of interest.
8 Capacity assessment
DNV AS
The following buckling requirement should be satisfied:
Section 6
The actual usage factor η is given by:
where σb denotes the max bending stress in the tower on the compression side and σcr is the corresponding
characteristic buckling resistance.
where:
The elastic buckling resistance of an unstiffened circular cylindrical shell may be taken as:
where:
ζ = 0.702 Z
γm = 1.15
DNV AS
κ = 1.3 if λ > 1.0
Section 6
where:
ℓ = cylinder length between radial stiff supports such as rings, platform decks, in mm
r = cylinder radius, in mm
t = shell thickness, in mm
2
σb = maximum bending stress in the tower on the compression side, in N/mm
2
σE = elastic buckling resistance, in N/mm .
DNV AS
SECTION 7 STRENGTH ANALYSIS OF THE SKIRT
Section 7
1 Introduction
In the following, acceptable procedures for strength analysis of the skirt are outlined. Procedures for strength
analysis of the tank constructions and the hull structure are given in Sec.5 and Sec.2 respectively. To perform
an effective analysis of the complete cargo tank system, it is advised that Sec.2, Sec.5 and this section are
considered coherently.
2 General
Spherical cargo tanks are supported by cylindrical skirts transferring the loads for the cargo tank, via the
equator profile, and down to the foundation deck. The skirt also protects the hull structure against the low
cargo temperature by an upper part constructed of aluminium and stainless steel where the latter acts as a
thermal brake, see Sec.5 Figure 12.
The skirt shall be assessed for allowable rule stresses and buckling.
— The allowable stresses of the skirt are to fulfil DNV-RU-SHIP Pt.5 Ch.7 Sec.21 [3.2.2].
— The buckling strength of the skirt is to fulfil the criteria given in App.D.
For class approval, the Society will require the skirt to be analysed by the means of a linear finite element
shell analysis to derive the stresses in the skirt.
3 Main geometry
DNV AS
Section 7
Figure 1 The main geometry of the tank skirt
DNV AS
Section 7
4 Design loads and stress summation
4.1 General
In the following, it will be assumed that the skirt is evenly supported along the intersection to the foundation
deck. In reality the loads in the skirt will concentrate in way of web frames and stool frames (hard points)
and this shall be assessed after the arrangement and scantlings of the hull structure have been decided, see
Sec.2.
1) Static weights
The forces in the skirt from the static weight of the cargo and the cargo tank system are evenly distributed
around the skirt. The static cargo weight causes vertical membrane stresses in the skirt. Hoop stresses are
very limited and locally distributed in way of radial restraints and bending stiffness changes around equator,
STJ joint, ring stiffeners and foundation deck areas.
2) Steady state temperature difference
The static temperature difference between the cargo tank and the atmosphere shrinks the tank forcing the
skirt to follow. This gives alternating membrane hoop tension and hoop compressive stresses in limited
Static loads
parts of the skirt. The compressive hoop membrane stresses should be considered in the buckling strength
assessment. These stresses should be averaged over the regions where they act. Axial bending stresses due
to temperature differences may be neglected.
3) Static deflection of foundation deck due to still water hull girder bending and double bottom bending
The typical LNG carrier is a hogging ship and the still water bending of the hull girder will induce vertical
tension stresses fore and aft in way of CL of the skirt. For the skirt to be in equilibrium, the sum of forces
along the connection to the foundation deck should be zero. The vertical stresses in the skirt at the outboard
positions are therefore in compression, with absolute value equal to the load in way of CL.
In addition the interaction forces from double bottom bending will depend on the ship bottom and tank
foundation design but will normally be of a magnitude equal to the interaction forces from hull girder
bending.
1) Acceleration of masses
Acceleration of masses are divided in three natural components:
a) Longitudinal acceleration ax.
Causes shear stresses at the outboard sides of the skirt and vertical stresses in way of the CL of the
Dynamic loads
skirt.
b) Transverse acceleration ay.
Causes shear stresses in way of CL and vertical stresses outboard.
c) Vertical acceleration az.
Causes vertical stresses in the skirt.
2) Dynamic deflection of foundation deck due to hull girder bending and double bottom bending
This load will have the same characteristics as the static no. 3 case above, however with the difference that
the load will vary between compression and tension at both positions, outboard and inboard (i.w.o. CL).
DNV AS
Section 7
4.3 Stress summation for allowable stress assessment
4.3.1 General
The design stresses for the skirt should be taken in accordance with one of the methods described below.
Only membrane stresses are considered.
4.3.2 Method 1
It is assumed that dynamic stresses from wave responses:
i) vertical acceleration
ii) transverse acceleration
iii) longitudinal acceleration
iv) interaction hull/tank
act independently and may be caused by different design wave conditions. The dynamic stress components
may then be considered as statistically independent and can be combined with a square root sum of squares
summation (SRSS)
sm = static mass
st = stationary temperature
sf = static interaction force
da = dynamic acceleration
df = dynamic interaction force.
Axial stress:
Hoop stress:
Shear stress:
This approach requires one FE analysis for each dynamic load i.-iv. above.
4.3.3 Method 2
If simultaneously acting accelerations are applied, e.g. by using the acceleration ellipse, the dynamic stresses
can be added linearly.
Axial Stress:
Hoop stress:
DNV AS
Section 7
Shear:
where:
aR = is the apparent resulting acceleration from the acceleration ellipse which also includes the sloshing
effect, see Sec.5 Figure 10.
Please note that (aR-1)σφsm is the dynamic stress (the inertia term of the static mass). The stresses
associated with the resulting acceleration aR should be determined by applying the static and dynamic mass
loads in the direction of aR in the FE analysis.
4.3.4 Method 3
If a stochastic (spectral) analysis approach is used, as for the CSA class notation, the stresses can be read
directly from the FE results on the skirt. This is on the condition that design waves maximising the axial
forces in the skirt at the actual locations have been used. Note that in this case the inertia effect from the
sloshing acting inside the spherical tanks shall be specifically considered, ref the apparent acceleration
approach in Sec.5 [4.6] and Sec.5 Figure 10.
4.5.2 Method 1
Axial stress:
Hoop stress:
Shear stress:
4.5.3 Method 2
Axial stress:
Hoop stress:
DNV AS
Section 7
Shear:
where aR = is the resulting acceleration from the acceleration ellipse, see Sec.5 Figure 11.
The term (aR-1)σφm is the dynamic stress (the inertia term of the static mass). The stresses associated with
the resulting acceleration aR should be determined by applying the static and dynamic mass loads in the
direction of aR in the FE analysis.
The partial safety factors/load factors are given in App.D:
4.5.4 Method 3
The stresses can easily be split in static and dynamic parts, but not into components load effects as for
method 1 and 2. The partial load factors above can therefore not be directly applied. As an initial estimate
γS = 1.15 and γd = 1.32 can be used for the static and dynamic parts respectively. However, the Society will
decide the applicable load factors on a case-by-case basis.
5.1 General
The analysis should confirm that the stress levels are acceptable when the structure is loaded in accordance
with described design conditions.
Any recognised calculation method or computer program may be applied, provided the combined effects
of axial, bending and shear deformations are adequately considered. Strength analysis carried out in
accordance with the following will be accepted as a basis for class approval in general.
DNV AS
Section 7
5.3 Mesh size
The mesh size of the skirt may typically follow the distance between the vertical stiffeners. The stresses will
however peak in way of the STJ both in way of CL and outboard due to local bending phenomena. This can
only be assessed with fine mesh in these areas. The stressed area may be modelled with fine mesh shell or
solid elements. The size of the elements may be approximately 2t × 2t (t × t for 4 node elements) where t is
the shell thickness.
Figure 3 Typical distribution of equivalent membrane stress in the skirt. The equivalent stress is
higher in way of CL due to the shear stress level
DNV AS
Table 1 Load case for finite element analysis of the skirt
Section 7
Loading condition Full load condition Notes
c) Thermal loads.
d) Transverse acceleration.
e) Longitudinal acceleration.
f) Vertical acceleration.
Dynamic
6.1 General
The skirt, being a stiffened cylindrical shell construction, shall be dimensioned against several buckling failure
modes, Table 1. To exclude local buckling of plating, longitudinal stiffeners and rings, explicit requirements
are given in App.D.
DNV AS
— A1: shell buckling between rings, unstiffened cylinders between rings. m is the number of vertical half
Section 7
waves between the rings or between a ring and one end (or between the two ends). n is the number of
full waves around the cylinder. Relevant for unstiffened and ring stiffened cylinders.
— A2: shell buckling between rings and vertical stiffeners; cylindrical panel. m is the number of vertical
half waves between rings. k is the number of circumferential half waves between the vertical stiffeners.
Relevant for ring and vertically stiffened cylinders, orthogonally stiffened.
A1 A2 A2
m = 1,2,... m = 1,2,... m = 1,2,...
Shell buckling n = 0,1,2,... k = 1,2,... k = 1,2,...
m = half waves between ℓ;
n = full waves around
cylinder;
k = half waves between
stiffeners
A3 A3
m = 1,2,... m = 1,2,...
n = 0,1,2,... n = 0,1,2,...
Panel stiffener buckling
m = half waves between ℓ;
NA
n = full waves around
cylinder
DNV AS
Type of skirt cylinder geometry
Section 7
Buckling mode type
Ring stiffened Vertically stiffened Orthogonally stiffened
A4
m = 1,2,...
n = 0,1,2,...
Panel ring buckling
m = half waves between L;
NA NA
n = full waves around
cylinder
A5
m = 1,2,...
n = 0,1,2,...
General buckling
m = half waves between L;
NA NA
n = full waves around
cylinder
Note: A1 to A5 refers to the buckling modes as defined in the computer program NVSKIRT.
DNV AS
SECTION 8 FATIGUE AND CRACK PROPAGATION ANALYSIS OF
Section 8
CARGO TANKS
1 Introduction
1.1 General
In accordance with DNV-RU-SHIP Pt.5 Ch.7 Sec.21 [1.1.2] and DNV-RU-SHIP Pt.5 Ch.7 Sec.4 [4.3.3], fatigue
analyses and a crack propagation analyses shall be carried out at areas where high dynamic stresses or large
stress concentrations may be expected.
Fatigue cracks may start in originally un-cracked bodies due to internal in-homogeneities in the material or in
grain boundaries in the material. The main concern is however macroscopic pre-cracks:
— along the weld toe due to undercut, poor weld profile or root defect
— as lamellar tearing, lack of fusion or lack of penetration
— due to hydrogen cracking or solidification cracking.
Figure 1 Various types of weld defects which typically initiate fatigue cracks
1.2 Fatigue
The objective of the fatigue control is to ensure that all parts of the cargo tank system subject to dynamic
loading have an adequate fatigue life. Fatigue life can be described by three phases:
1) initiation, Ni
2) crack growth, Ns
3) failure.
The fatigue life is defined by N = Ni + Ns. For both welded and unwelded details the major part of the
fatigue life lies in the initiation phase. For welded constructions the crack starts to grow from an initial defect
of the type discussed above to the end of the initiation stage that can be defined by for instance a detectable
crack size or a certain change in stress (strain) close to the crack toe.
DNV AS
Section 8
1.3 Crack propagation
Structural failure may develop in one of the two following ways:
-4 -2
1) Most of the crack growth (fatigue damage) takes place under pulsating loads in the 10 to 10
probability range. Here the stress ranges are smaller and the number of cycle higher than in the extreme
-8
10 load range that just experiences some few cycles.
2) A rapid brittle or plastic crack growth takes place when the crack has developed to a size (length and
depth) that can be defined as critical subject to a sudden high load. In the present context the high
8
load is taken as the largest expected load during 10 wave encounters in the North Atlantic (static plus
dynamic).
2 Loading conditions
2.1 General
The ship loading conditions to be used in the fatigue and fracture mechanics analyses are given in Table
1 below for a normal trading carrier (trading with full load on entire laden voyage and in ballast for return
transit). For vessels intended for trades with part load conditions, the exposure fractions should be specially
considered. The design loading conditions and exposure times giving basis for the fatigue calculations will be
stated explicitly in the ships papers.
* If the tank is sailing with heel (e.g. up to 10% filling height) at ballast voyage, this should be taken into account for the
duration of the ballast voyage.
A simplified approach for tank and skirt may be to calculate the fatigue damage from full load condition 50%
of life time, assuming this will incorporate the damage effect from ballast operation. To be agreed with class.
3.1 General
The nominal FLS stresses in the tank at a probability level of Q = 10-2 should be found by combining all the
following load effects:
DNV AS
a) inertia stresses from accelerations of static weights (cargo and self-weights of the tank system)
Section 8
b) dynamic stresses caused by interaction between hull and tank due to hull girder bending and double
bottom bending causing deformation of the tank foundation deck
c) dynamic stresses due to liquid sloshing loads at reduced tank fillings
d) stresses caused by temperature differences.
The governing loads for fatigue of the tank shell are due to full tank loads (load type a) and interaction forces
(load type b). The tower, and in particular the tower supports will be governed by liquid motion loads at
reduced tank fillings (load type c).
4.1 General
Dynamic (cyclic) stresses are driving the fatigue crack growth, whereas the rupture at a given crack size is
governed by the maximum encountered stress amplitude (static plus dynamic).
The primary parameter for final rupture given a certain crack size is the most probable largest one time
8
stress amplitude (static plus dynamic) during 10 cycles in the North Atlantic. This value equals the extreme
ultimate limit state (ULS) load used for allowable stress and buckling capacity checks. Since fatigue crack
growth and final rupture are parts of the same fracture mechanics analysis and normally checked by the
same software, the total fatigue load stress spectrum should be taken as the most probable largest load
8
spectrum the ship will experience during 10 wave encounters in the North Atlantic.
DNV AS
Section 8
4.2 Fatigue and fracture mechanics analysis
In order to evaluate residual fracture of fatigue cracks over the lifetime of the vessel fracture mechanics
analysis shall be carried out based on the extreme ULS long term stress distribution instead of the Q = 10
-2
probability level to be used for SN-curve fatigue analysis. The stress components should be combined as
follows;
DNV AS
a) By adding up damage contributions
Section 8
— Fatigue damage contributions (Miner sums) calculated for each loading condition can be added
according to the operational profile of the vessel to give the total fatigue damage contribution over
the design lifetime of the vessel.
— For the part load condition the fatigue damage contributions from the considered filling levels can
be added according to the operating time at each filling level. A minimum of three part filling levels
should be used.
b) By establishing a resulting long term Weibull stress distribution. This can be done by combining the long
term stress distribution for all the load cases as a weighted sum according to the operational profile for
the vessel. This reduces the number of stress cycles in each operating mode with the corresponding
fractions. Fatigue analysis on this basis should be compared to the total design lifetime.
DNV AS
Unstable cracks is characterised by either spontaneous crack growth with no additional input of driving
Section 8
strain energy (brittle fracture) or as plastic tearing needing only marginal input of strain energy for the
crack to propagate. If bending stresses are dominating the stress picture in the shell, the crack will grow
predominantly in the longitudinal direction and less through the thickness.
The crack growth may be calculated by stepwise integration of the Paris’ equation:
where:
The additional bending stress from eccentricities can be calculated directly from the FE- analyses or
from suitable equations as given in DNV-CG-0129. If the welds are ground flush or almost flush the local
stress concentration factor at e.g. weld toes, Km, can be set equal to 1.0. If the shell welds as well as the
connection welds to the equator profile are not ground, the Km definition in the CrackWise program can be
applied.
DNV AS
-8 8
level 10 is set equal to 1. The long term wave-induced stress range spectrum (10 cycles) is given in Figure
Section 8
2 is assuming a Weibull shape parameter of ξ =1.0. This is the normal assumption if no better estimate is
available.
In the direct analysis procedure which is required for independent tanks of type B the Weibull shape
parameter can in principle be determined. Hence, the shape of the stress distributions in Figure 2 and Figure
3 can be modified as follows:
where:
= reference stress value, in N/mm2, at the local detail exceeded once in n0 cycles (shown as
normalized to 1.0)
n0 = total number of cycles associated with the stress range level Δσ0, at probability level 10-8.
DNV AS
Section 8
Figure 3 Short term (15 days of storm) wave-induced stress range spectrum
The driving stress range can based on the testing reported in /14/ be set to:
where:
2
= dynamic membrane stress range, in N/mm
2
= dynamic bending stress range, in N/mm
m = dynamic bending reduction factor.
DNV AS
The value of the dynamic bending factor 'm' can be taken as 3. The strain energy associated with the
Section 8
compressive side of the bending stress distribution will not contribute to open the crack, but the tension side
might. As a conservative measure 1/3 of the surface bending tension stress range has been included. This
corresponds to the centre of gravity of the triangular tension part of the bending stress distribution.
m = 3.66
-13
C = 1.923E
The mean plus two standard deviation data (μ + 2σ) above are given in /10/, and the fatigue crack growth
threshold can be calculated according to BS7910, see /19/, as follows:
Other parameters than above may be used if they are documented and accepted by the Society.
9 Leakage calculation
DNV AS
and bending. As a minimum the horizontal welds at the top and the bottom of the equator profile shall
Section 8
be examined together with the tank side of the equator groove.
b) Through-thickness penetrating cracks in the lower hemisphere, the tank bottom and the tower
foundation supports, which is assumed to be below the liquid surface
DNV AS
3) The corresponding crack length at the penetration side can then be estimated as:
Section 8
where:
(1)
(2)
where:
2
σsm = static membrane stress, in N/mm
2
σdm = dynamic membrane stress amplitude, in N/mm
2
σdb = dynamic bending stress, in N/mm
m = dynamic bending reduction factor.
For the sake of conservatism a (1/m) fraction of the dynamic bending stress has been included with a value
of 'm' equal to 3, see [8.6].
For the purpose of average crack opening for leakage calculations the average effective tension stress will
contribute to open the crack.
a) If the membrane tension stress is larger than the dynamic membrane stress amplitude the net effect of
the dynamic stress will be zero, eq. (1).
b) On the other hand, if the static membrane tension stress is less than the dynamic membrane stress
amplitude the average effect of the difference should be included, eq. (2).
DNV AS
Section 8
9.6 Effective crack opening area
For a case of a pure tension stress σ it may be assumed that the shape of the opening in stage III is an
ellipse whose major axis length is the crack length 2ap and whose minor axis length is equal to the maximum
crack opening displacement 2δ. The area of the opening is then given by
where:
Thus:
For cases with combined bending and tension, the crack opening and also the crack length will vary through
the thickness.
where:
Here A is the crack opening area on the penetration side as outlined in [9.6] above.
The orifice coefficient has been derived from the tests reported in /16/. A value of 0.1 has been found to give
a reasonable comparison with the test data.
DNV AS
SECTION 9 FATIGUE ASSESSMENT OF THE HULL STRUCTURE
Section 9
1 General
The aim of fatigue strength assessment is to ensure that the structure has an adequate fatigue life. Results
from fatigue life analysis can also be used for setting up inspection programmes for the fabrication and
operation phases of the carrier.
2 Definitions
SCF = stress concentration factor
Kg = geometric stress concentration factor
Kt = stress concentration factor due to thickness effect
Mwv‐hog = wave induced vertical hogging moment (positive), in kNm
Mwv‐sag = wave induced vertical sagging moment (negative), in kNm
2
Δσ = stress range, in N/mm
ξ = weibull factor
fm = reduction factor for mean stress effect.
4 Loading conditions
Loading conditions representative for the average trading pattern of the vessel should be selected from the
loading manual covering full load, ballast and part load conditions as relevant. See also DNV-RU-SHIP Pt.5
Ch.7 Sec.21 [2.2.1] conditions a), b) and c), and Sec.2 Table 1 and Sec.8 Table 1.
DNV AS
b) the connections between:
Section 9
— the tank skirt and the foundation deck
— the foundation deck and the inner side
— the double bottom and the stool bulkhead
c) knuckles in the inner sides.
Some examples of fatigue sensitive areas are illustrated in Figure 1.
6.1 General
The upper deck and the connection to the tank covers are fatigue sensitive areas due to high longitudinal
stress levels as:
— the deck area is very limited and hence fully utilised with respect to ultimate stress level. High strength
2
steel is typically used, e.g. ReH = 355 N/mm
— the ships are generally in hogging with respect to still water bending moment.
DNV AS
The deck may be penetrated by pipes and fitted with bollards, fairleads, platform decks, deck houses, guard
Section 9
rails, lamp posts etc. causing stress concentrations.
It is recommended that the upper deck main scantlings are adequately designed with respect to fatigue to
allow for a minimum stress concentration and thereby to gain confidence in the estimated fatigue capacity for
a selection of structural details. A suitable selection of details should be analysed so as to provide confidence
that most of the remaining local details meet the fatigue requirements.
The builder should address the fatigue life in the preliminary design phase and choose between:
1) Design the midship section according to normal rule strength requirements, i.e. local yield and buckling
criteria and hull girder capacity. This may imply extensive weld treatment to achieve acceptable fatigue
life for the deck attachments.
2) Design the midship section with increased scantlings, i.e. deck section modulus, to reduce the
stress level and hence improve fatigue life of the deck area to provide a reserve margin for stress
concentrations at deck attachments.
where teff is an effective thickness defined in DNV-CG-0129, which in general is the thickness through which
the potential fatigue crack will grow, and tref is the reference thickness, which is 25 mm. Values for the
thickness exponent n are given in DNV-CG-0129.
The thickness effect can be included directly in the stress calculation as a stress correction factor according to
DNV-CG-0129 Sec.2 [5].
The thickness effect is only present when the governing principal stress range is within ±45º to the weld toe.
DNV AS
Section 9
Figure 2 High stress area in weather deck due to hull girder bending
The below requirements are not mandatory but are recommended to be applied in the early design phase
in order to give an indication of the required hull girder section modulus for compliance with the mandatory
fatigue requirements specified in DNV-RU-SHIP Pt.3 Ch.9 and DNV-CG-0129.
The minimum section modulus for the hull girder may as an initial estimate can be calculated as:
where:
Mwv–hog = hogging vertical wave bending moment for fatigue, in kNm, as defined in DNV-RU-SHIP Pt.3
Ch.4 Sec.4 [3.1.1].
Mwv–sag = sagging vertical wave bending moment for fatigue, in kNm, as defined in DNV-RU-SHIP Pt.3 Ch.4
Sec.4 [3.1.1].
L = ship rule length, in m.
DNV AS
fe = environmental stress range factor for reduction of north atlantic stress response range to actual
Section 9
trading area, i.e. fe = 1.0 for north atlantic wave environment. For world-wide environment the
factor may be taken as 0.8 as given in DNV-CG-0129.
Kg = geometric stress concentration factor in deck as given in DNV-RU-SHIP Pt.3 Ch.9 and DNV-
CG-0129. The criterion is calibrated to a SCF = 1.6 in world-wide environment, fe = 0.8. This
corresponds to, fe =1.0, and SCF = 1.28 in north atlantic environment.
c = calibration factor for spherical LNG carriers ( = 0.9).
Reference in DNV-
Connection type
CG-0129 App.A to SCFs
DNV AS
Reference in DNV-
Section 9
Connection type
CG-0129 App.A to SCFs
It is recommended that a screening analysis is carried out to map the critical stress concentrations. Given the
actual stress concentrations of the different connections, one may verify whether weld treatment or redesign
is necessary to satisfy the given fatigue life criteria.
7.1 General
The extent of fatigue analyses is given by the applicable (specified) class notations. However, in spherical
tank carriers the areas/details listed in the following should be considered for fatigue assessment of cargo
hold details. Other stress raisers may be present, and the application of additional fatigue calculations should
be considered case by case based on screening analysis and/or past experience with similar designs.
Screening analysis should, if carried out, be based on a complete ship coarse finite element model with cargo
hold mesh in the area subject to screening. Course fatigue life estimates can then be obtained based on
assumptions regarding
— scatter diagram
— SN-curve
DNV AS
— type of stress (membrane)
Section 9
— mean stress effect
— stress concentration factor (Kg, Kt).
A screening analysis cannot be applied to document fatigue life due to the coarseness of the FE model, but
will give indications on which areas should be subject to further fatigue analysis.
Fatigue analysis of the listed items should preferable be of the spectral (stochastic) type. Spectral analysis
includes the dynamic global axial force which may be important for some of the details listed below.
Simplified analysis should be based on the loads listed in the Table 2.
bulkhead
DNV AS
Section 9
Figure 3 Typical hotspots for a longitudinal web frame penetration
The fatigue life of the connections (b) and (c) in Figure 3 is included in the scope for the PLUS notation, DNV-
RU-SHIP Pt.6 Ch.1 Sec.6, and depends on the following factors:
a) The external dynamic pressure.
b) The internal dynamic pressure (of water ballast if any) i.e. the vertical, horizontal and longitudinal
acceleration level.
c) The static pressures.
d) The shear stress in the web frame (dependant on the above named factors).
e) The design of the detail, see Figure 3:
— lug type, e.g. open, semi closed or fully closed
— longitudinal geometry, span and spacing
— stiffener in top type and span.
f) Corrosiveness of environment.
The pressure loads in bullets a to c, varies along the length of the ship. For ships subject to extended class
notations (CSA, PLUS), the fatigue life should be documented at the following positions:
1) Parallel midship body.
2) Foremost cargo hold and the adjacent cargo hold.
3) Aft part of aftmost cargo hold.
DNV-CG-0129 gives guidance on fatigue evaluation of the subject connections.
DNV AS
For both alternatives geometrical eccentricity in the knuckle should be avoided or kept to a minimum. The
Section 9
weld connection between the frame and the inner bottom/hopper plate should be of full penetration locally in
way of the knuckle.
For the straight knuckle, the fatigue life is governed by transverse stresses. Thus, to have proper support of
the straight knuckle, brackets should be fitted in ballast tanks in line with the inner bottom. In addition, one
of the following structural solutions for knuckles with angles between inner bottom and hopper plate between
30° and 75°, should normally be adequate:
1) Bracket inside cargo hold. The bracket should extend approximately to the first longitudinals and the
bracket toe should have a soft nose design.
2) Insert plate of 2.0 times the thickness normally required. Insert plates should be provided in inner
bottom, hopper plate, and web frame. The insert plates should extend approximately 400 mm along
inner bottom and hopper plate, approximately 800 mm in longitudinal direction, and 400 mm in the
depth of the web.
The weld connection between the supporting brackets and the frame plating should be full penetration locally
in way of the knuckle adjacent to frames.
The bent knuckle is also subject to longitudinal bending and local support of the inner skin may be found
necessary. This may be provided by fitting of a longitudinal carvel along the knuckle. Transverse carvels close
to and on each side of the floor may be an alternative.
DNV AS
Section 9
Figure 5 Connection between the inner bottom and the transverse stool, strengthened with soft
toe brackets
Figure 6 Local bending in way of cover connection due to global bending moment
DNV AS
Section 9
7.6 Connection of the tank skirt to the foundation deck
The connection transfers the entire load from the cargo tank to the ship structure. Typically, the skirt is
welded directly to the foundation deck with supporting brackets below deck in way of frames and stool
girders. This gives several stress concentration areas to be investigated, see Figure 7 and Figure 8.
Figure 7 Typical location where stress concentrations occur, to be investigated. The arrows show
directions for extrapolation of stresses at two separate locations
Typical fatigue strengthening of this connection may be soft brackets on each side of the skirt. Normally
only the connections at the four mid hold frames need strengthening and as this normally gives satisfactory
fatigue life, possible further documentation of the fatigue life of the detail is subject to special consideration
based on the applied strengthening.
DNV AS
Section 9
Figure 8 Typical fatigue strengthening of the skirt
7.8 Knuckles in the longitudinal bulkheads in forward and aft cargo hold
area
Knuckles in the inner sides in the aft and forward parts of the cargo hold area should be checked for fatigue.
Special attention should be paid to knuckles in way of the passage deck where local strengthening may be
necessary.
DNV AS
SECTION 10 VIBRATION ANALYSIS
Section 10
1 General
1.1 General
Potential damaging effect of vibration on the containment system shall be examined, according to DNV-RU-
SHIP Pt.5 Ch.7 Sec.4 [3.3.5].
2 Analysis procedure
Diesel-electric propulsion with resiliently mounted diesel — Blade passing frequency from main propeller(s).
engines with speed above 500 RPM:
DNV AS
Excitation sources Frequencies
Section 10
Geared propulsion units with resiliently mounted diesel
engines with speed above 500 RPM:
DNV AS
10) For twin screw vessels the forced response should be calculated for two excitation modes:
Section 10
— propellers/main engines acting in phase
— propellers/ main engines acting in opposite phase.
11) The highest response of the two excitation modesshould be considered.
12) The damping applied may be proportional to frequency, but not exceeding 2% of critical damping.
13) The maximum calculated vibration level in a frequency range corresponding to 100% MCR + 15% and
90% MCR – 15% should be considered to be excited at full speed.
14) The calculated vibration levels for each of the applied excitation sources may be evaluated separately.
DNV AS
SECTION 11 BIBLIOGRAPHY
Section 11
1 Bibliography list
3
Hansen, R and Rimeid, B. E.: “Moss Rosenberg Verft A/S, 87.600 m LNG Carrier, prediction of loads
/1/ due to sloshing in tanks. For Moss Rosenberg Verft A/S.”, Det Norske Veritas, Report no. 72-45-C,
May 1972.
Olsen, H. A. and Hysing, T.: “A Study of Dynamic Loads Caused by Liquid Sloshing in LNG Tanks”,
/2/ Det Norske Veritas, Report no. 74-276-C, December 1974 (Also as US Maritime Administration
Report no. MA-RD-920-75040)
Hansen, R. E.: “Moss-Rosenberg LNG-carrier. Buckling analysis of partly filled spherical tank.”, Det
/3/
Norske Veritas, Report no. 73-99-C, July 1973.
Valsgård, S.: “User’s Manual – NVSPHERE, Design program for spherical gas tanks of Moss design”,
/4/
Det Norske Veritas, Report no. 2003-0514, April 2003.
Det Norske Veritas: “Buckling strength of shells”, Recommended Practice DNV-RP-C202, October
/5/
2002.
/6/ IIW Fatigue Recommendations XIII-2151-07/XV-1254-07, May 2007
Valsgård, S.: “Computational methods for analysis of ship hulls”. Paper Series No. 2000-P007
/7/
(2000).
Det Norske Veritas: “Environmental Conditions and Environmental Loads”, Recommended practice,
/8/
DNV-RP-C205, April 2007.
/9/ “Crackwise 4”, The Welding Institute, Cambridge, UK
Fredheim, S.: “9% Ni-steel and aluminium alloys AA5086 used in LNG tanks in ships. A probability
/10/
of failure analysis.”, DNV Technical Report 33-0041 (1983)
Odland, J.: “Theoretical and Experimental Buckling Loads of Imperfect Spherical Shells”, Journal of
/11/
Ship Research, Vol. 25, No. 3, September 1981, pp. 201-216
Kamsvåg, F., Steen, E. and Valsgård, S.: ”A Computational Method for Analysis of LNG Vessels with
Spherical Tanks”, The Seventh International Symposium on Practical Design of Ships and Mobile
/12/
offshore Units (PRADS’98), The Hague, The Netherlands, September 1998 (DNV Paper Series no.
99-P002)
Lindemark, T., Kamsvåg, F. and Valsgård, S.: ”Fatigue Analysis of Gas Carriers”, Design and
/13/
Operation of Gas Carriers, RINA, London, September 2004
Tenge, P. and Sollie, O “ Fracture mechanics in the design of large spherical tanks for ship transport
/14/
of LNG”, Norwegian Maritime Research, Vol.1, no. 2, pp.1-18, (1973)
3
Sollie, O. and Tenge, P.: “Aluminium Alloy 5083 for use in 125 000 m LNG Carrier of the Moss
/15/ Rosenberg Design, Part 1: Fracture toughness, geometry and propagation rate of fatigue cracks
initiated from surface defects, Det Norske Veritas Report no. 845031/1”, 21 July 1973
3
Sollie, O. and Tenge, P.: “Aluminium Alloy 5083 for use in 125 000 m LNG Carrier of the Moss
/16/ Rosenberg Design, Part2: Determination of gas leak rates through penetrating cracks”, Det Norske
Veritas Report no. 845031/2”, 21 July 1973.
Hysing, T., Ruth, E. and Zang, Y.: “Experimental Study of Sloshing Loads in Spherical LNG Tanks”,
/17/
DNV Report 2009-2034, rev.02 2010
Nyseth, H. and Renaud, F.: “Prediction of Sloshing Loads in Spherical LNG Tanks”, DNV Report no.
/18/
2015-0107
/19/ BS 7910: “Guide on the Methods for Assessing the Acceptability of Flaws in Metallic Structures”
DNV AS
APPENDIX A FORCES IN SPHERICAL TANKS
Appendix A
1 Force calculation
Note:
φ is the angle determining the force calculation point.
α is the angle determining the liquid level.
FR is the filling ratio given by:
DNV AS
Table 1 Shell forces in general may be found in see /23/
Liquefied gas carriers with spherical tanks of type B
Class guideline — DNV-CG-0134. Edition October 2021
Nφ Nθ Nφθ
Weight of shell
insulation
LOWER HEMISPHERE
Uniform pressure
Horizontal
SPHERE
DNV AS
acceleration*
Weight of shell
insulation
UPPER HEMISPHERE
Appendix A
APPENDIX B DERIVATION OF THE SKEW-SYMMETRIC EQUATION
Appendix B
1 Skew-symmetric ULS loads
The following derives equations for application of the skew-symmetric ULS loads (allowable stress and
buckling) in the cargo tank finite element model in an upright coordinate system.
Alternatively, if the FE software allows for it and a half (180°) or a full (360°) circumference model of the
tank is made, the same pressure load distribution can be obtained by applying the loads directly in the
direction of the skewed acceleration vector aR, see Figure 1.
Figure 1 Spherical tank pressure distribution – definition of meridional positions and acceleration
vectors
DNV AS
Appendix B
Definitions:
The resulting pressure head along the θ0 median, as shown in Figure 2, is given by:
If PR < 0 then PR = 0
Otherwise the pressure PR is given by the expressions below:
: quasi-static pressure.
Square root sum of squares (SRSS) summation according to the rules for assessment of allowable stress of
the tank system and buckling strength of the skirt, requires that the static and dynamic parts of the loads are
separated. This can be done as follows:
The static pressure is found by setting aH = aV = 0:
DNV AS
Appendix B
The dynamic pressure is found as aH ≠ 0 and av ≠ 0
DNV AS
APPENDIX C SLOSHING DESIGN OF SPHERICAL LNG TANKS
Appendix C
1 General
The acceptable sloshing design approaches listed below are applicable to LNG ships intended to operate in
part load conditions.
3 Background
DNV AS
Appendix C
4 Sloshing assessment procedures
The simplified approach above using the 1 d.o.f. tests considers only the main dominating sloshing excitation
in the horizontal plane. Influences of other motion components like roll and vertical motions as well as
excitations with periods outside the sloshing resonance range are disregarded.
The approach is as follows:
a) Calculation of resonant sloshing excitation (max amplitude) of spherical tank in the horizontal plane
(sway, surge) based on irregular ship motions (filtering of ship motions at sloshing resonance periods).
b) Use of regular motion model test results for horizontal tank excitations to predict the maximum sloshing
loads in irregular seas.
c) Calculation of horizontal tank loads and loads on tank tower and establish simplified long term
distributions.
If the model tests are supplemented with CFD calculations this will enable calculation of liquid motions and
loads in the tanks due to irregular ship motions in 6 d.o.f. The approach entails the following steps:
a) Correlation/scaling against regular motion model test results, see /10/, i.e. variation of filling height and
excitation amplitude, for verification of the CFD approach.
b) The correlation will cover both liquid velocities in the tank, total forces on the tank and forces (drag
coefficients) for the tank tower.
c) Calculation of liquid motions, tank forces and tower forces for a selected number of critical sea states
(e.g. 12 – 16 sea states) to establish the long term distribution of the sloshing forces for the various
filling levels.
The CFD calculations accounts for all the motion components and all the excitation periods in a realistic
manner. However, comparisons carried out indicate that CFD calculated liquid velocities shall be compared
to those of regular model tests. In addition, drag coefficients for estimation of the tower sloshing forces also
shall be assessed.
DNV AS
APPENDIX D BUCKLING CRITERIA OF LNG SPHERICAL CARGO TANK
Appendix D
CONTAINMENT SYSTEMS – SKIRT AND SPHERE
1 General
This appendix contains acceptable buckling design criteria for the skirt and sphere for LNG spherical cargo
tank containment systems. The criteria are relevant for independent type B spherical tanks.
2 Design principles
The design stresses should be based on direct calculations normally using general purpose linear FEM
program systems or equivalent software. The FEM models may include the complete ship or parts of the ship
only.
The most critical design stress combination (σ10, σ20, τ0) with respect to buckling and its location in the
structure should be identified.
The buckling criteria are written as limit state functions consistent with the format used in modern reliability
analysis. Due to the serious consequences of failure the acceptance level of the limit states corresponds to a
higher safety level than is normally implemented in conventional ship rules.
The dynamic part of the design stresses to be used in the limit state functions should be based on a 25 years
return period under North-Atlantic environmental conditions.
The buckling criteria should be compatible with the maximum tolerance level with respect to deviations from
perfect form due to fabrication. However, the criteria herein are of a form that allows the strength to be
assessed also for exceedance of the tolerance limits. In such cases the Society will consider the relevant
imperfection amplitudes to be used in the design procedure.
The present methods may be substituted by more refined methods using direct calculations. Particularly
non-linear FEM programs may be used for assessing buckling strength with due considerations of effects
from non-linear material behaviour, realistic boundary conditions etc. Special emphasis should be put on
geometrical imperfections and how these are implemented in the computer model. The Society will consider
such analyses and their relevance on a case-to-case basis.
3 Skirt buckling
3.1 General
This Appendix gives acceptable methods for calculation of the buckling strength of the skirt supporting the
spherical LNG tank, see Figure 1. The skirt may be stiffened by rings or vertical stiffeners and they may be
arranged on the inside or outside of the shell, see Figure 2.
The skirt shall be checked against several possible types of buckling modes, see Table 1:
a) Shell buckling: buckling of shell between rings/stiffeners. a1) Shell buckling between rings- unstiffened
cylinders between rings, a2) Shell buckling between rings/vertical stiffener- cylindrical panel (NVSKIRT
notation; A1, A2).
b) Panel stiffener buckling: overall buckling of vertical stiffeners including shell. Rings frames are nodal
lines. Coupled to torsional buckling of vertical stiffeners (NVSKIRT notation; A3).
c) Panel ring buckling: overall buckling of rings including shell. Vertical stiffeners (if any) are nodal lines
(NVSKIRT notation; A4).
d) General buckling: overall buckling of rings and vertical stiffeners including shell (NVSKIRT notation; A5).
e) Local buckling of vertical stiffeners/rings. Explicit requirements for excluding these local failure modes
are given.
The buckling criteria given herein require a minimisation procedure with respect to the number of waves
in axial and circumferential direction. This procedure is a simplified eigenvalue analysis accounting for
DNV AS
non-linear pre-buckling effects and the elastic knock-down factor due to deviations from perfect forms
Appendix D
are accounted for. The present criteria are best suited for programming on a PC-computer or equivalent
hardware.
The buckling criteria herein may also be used when the tolerance limits in [3.5]are exceeded. This needs
special considerations from the Society.
The methods and procedures given here may be substituted by more refined analyses using direct
calculations. General purpose non-linear FEM programs or more specialised programs such as BOSOR4,
BOSOR5 for rotationally symmetric shells are relevant analysis tools.
The design criteria are written in a limit state (partial safety) format and are suited for reliability calculations.
(1)
where:
(2)
(3)
(4)
(5)
(6)
σx0 , σθ0 ,τ0 are the design stresses including load factors as given in [3.3].
The partial safety factorsκ and γm are given in [3.4].
ΛCR is related to the critical buckling stresses as given in [9].
FE may be taken as:
DNV AS
Appendix D
(7)
Values for ΛE are given under [3.2.3]and will depend on the buckling mode considered.
3.2.3 Elastic buckling stresses, eigenvalue calculations
The elastic buckling parameter ΛE for a cylindrical shell under combined axial compression, shear stress and
circumferential compression will vary depending on the buckling mode considered. The general expression for
ΛE valid for modes a), b), c) and d) defined in [3.1] is
(8)
For mode a1), b), c) and d) the parameters are defined as (for mode a2, see below)
(9)
(10)
(11)
(12)
(13)
F0 in the form given in eq.(13) is valid for typical LNG skirt designs, see eq.(19) to (23) for unstiffened parts
of the skirt.
C is a buckling coefficient defined as:
(14)
The parameters m and are unspecified parameters i.e. the ΛE shall be minimised with respect to m (m =
1,2,...) and n (n = 0,1,2,...) for each type of buckling mode. m, n and are buckling wave parameters (see
notation list in [3.1]).
For mode a2) (local shell buckling between vertical stiffeners), the parameters take the following values:
(15)
(16)
(17)
DNV AS
Ω is a load parameter defined as:
Appendix D
(18)
(19)
(20)
C0 is defined as:
(21)
(22)
ΛE is minimised with respect to m and α. m is the number of half-waves in the vertical direction and it takes
discrete values m =1,2,3,..,20. α is a continuous parameter describing the inclination angle in radians of the
buckling pattern relative to the horizontal plane. α will take values in the interval 0.0 – 1.0, depending on
the ratio between the acting biaxial/shear loads. In a minimisation procedure an increment in α of 0.02 ( ~ 1
deg) will give sufficiently accurate results.
The minimum positive value for ΛE shall be found. This means that the buckling condition to be found is in
the same load quadrant as the corresponding design load point. The axial stress σx0 is always positive (i.e.
compressive), τ0 is always positive (shear stress independent of direction) while σθ0 may be both positive or
negative. If b > 0, the negative root sign in eq.(8) applies while if b < 0, the positive root sign applies.
The parameters Z, K, aR, γR, as, γs and δ will depend on the type of buckling mode considered and are given
under [3.2.4] to [3.2.8] respectively.
3.2.4 Shell buckling between rings, unstiffened cylinders between rings (a1)
The elastic buckling parameter ΛE is calculated from [3.2.3] with the following parameters substituted:
(23)
(24)
(25)
DNV AS
Appendix D
(26)
(27)
(28)
ℓ is the length between rings. For the upper part, ℓ is the distance between equator profile and upper ring.
γs = as = γR = aR = 0
δ is the production tolerance limit given in [3.5.2], i.e.
(29)
(30)
(31)
ℓ is the length between stiff radial supports. For the upper skirt part ℓ is the distance between the equator
profile and upper ring. R is the shell radius, t is the shell thickness.
γs , as as defined in notation list. γR=aR = 0
δ is the production tolerance limit given in [3.5.3], i.e.
(32)
DNV AS
ΛE is minimised with respect to m = 1 and n = 0,1,2,...,50;
Appendix D
3.2.7 Panel ring buckling (c)
The elastic buckling parameter ΛE is calculated from [3.2.3] with the following parameters substituted:
(33)
(34)
(35)
(36)
(37)
DNV AS
The combined design stresses are
Appendix D
(39)
(40)
(41)
In practise the problem will be to identify the load combinations that are most severe for the skirt with
respect to buckling, i.e. to identify the design stresses σx0 , σθ0 , τ0 and their positions that produce the
most critical buckling condition. All combinations of (σx0 , σθ0 , τ0) should be checked to identify the most
severe situation with respect to the buckling criterion as given in [3.2.2]. It is sufficient to calculate the
stresses along the meridians 0° and 90° measured from the centre line (θ = 90° is normally most critical).
The load factors γfi are given in [3.4].
γf3= 1.4 for forces caused by deflection of the foundation deck, static
γf5= 1.4 for forces caused by deflection of the foundation deck, dynamic.
(42)
Slenderness factor κ:
(43)
(44)
(45)
(46)
DNV AS
Appendix D
3.5 Tolerances
3.5.1 General
The deviations from nominal perfect shape of shell, axial stiffeners and rings should normally not exceed the
tolerance limits given herein.
Under special circumstances the Society could accept tolerance exceedance, and the strength parameters
calculated under [3.2] should then be modified to account for the measured imperfections.
(47)
(48)
where ℓ is the unsupported length of the vertical stiffener/ring spacing. The out-of-straightness should be
measured along the stiffener/shell intersection line and also sideways at the free end of the stiffener.
(49)
where:
(50)
(51)
DNV AS
Appendix D
3.6 Local requirements for vertical stiffeners and rings
3.6.1 General
In order to prevent torsional buckling of stiffeners and rings which is not prevented by intermediate brackets,
the following requirements for stiffener properties shall be satisfied.
(52)
Flat-bar stiffeners:
(53)
(54)
3.6.3 Rings
Stiffener web slenderness:
Rings with flange:
(55)
Flat-bar rings:
(56)
(57)
(58)
DNV AS
where:
Appendix D
(59)
(60)
and
(61)
(62)
(63)
If eq. (58) is not satisfied the torsional buckling strength can be assessed using more advanced methods,
and the characteristic torsional buckling strength can be used as reduced yield stress in the stiffener and
ring buckling assessments in [3.2.6] and [3.2.8]. An accepted method is to calculate the torsional buckling
strength of vertical stiffeners as
(64)
where:
(65)
σT is substituted for ReH in eq.(2) and (3). If this procedure gives strength within required limits, eq.(58)
need not to be satisfied.
DNV AS
3.7.2 Semi-direct method
Appendix D
In the present context semi-empirical methods mean to assess the buckling stresses under the assumption
of linear elastic material behaviour, i.e. to assess the parameter ΛE which enters the design procedure under
[3.2]. It is necessary to include realistic levels of geometrical imperfections as given under [3.5] and to
identify the lowest buckling strength and associated mode.
The elastic buckling parameter ΛE is defined as follows:
(66)
(67)
(68)
i.e. it is the value of the proportional loading factor Λ that corresponds to elastic buckling of the cylinder
under the action of generally axial stress σx, hoop stress σθ and shear stress τ.
The value found forΛE is substituted into the procedure in [3.2]which includes the effect from material
yielding and appropriate safety factors. From eq.(66-68) ΛE is found as:
(69)
For one single load component only, the buckling capacity is not dependent on the value of ΛE as long as the
related value for the design stress σi0 is calculated from the definition:
(70)
(71)
where:
(72)
and
(73)
σxCR, σθCR, τCR are the values of the scaled nominal stress components at the point of ultimate buckling
capacity, i.e. they represent the ultimate stress capacity for combined loading. L0 is the radius vector of the
design stresses defined as:
DNV AS
(74)
Appendix D
For a single acting stress component (e.g.: pure axial compression), the limit state is simplified to:
(75)
Safety factors:
(76)
γm = 1.15 and κ from eq.(43-45). Definition of κ also necessitates an assessment of the elastic buckling
capacity including the knock-down effect from geometrical imperfections in order to calculate the slenderness
parameter λE defined in eq.(3). ΛE is defined in eq.(69).
When using direct analysis tools the safety factors shall be discussed with the Society on a case by case basis
and deviations from eq.(72) may be considered.
4 Sphere buckling
4.1 General
The buckling strength of thin spherical shells is known to be extremely imperfection sensitive, and classical
linear buckling analysis is not sufficient for determination of the buckling strength. The level of imperfection
sensitivity depends on the type of loading with uniform external pressure as the most severe. This note gives
acceptable buckling criteria based on Koiter’s post buckling theory.
The criteria are developed for use in the design of spherical LNG tanks.
The buckling calculation should be carried out for the separate zones with constant thickness, and within
each zone the most severe combination of loads with respect to buckling should be identified.
Direct buckling strength calculations using recognised software may be accepted by the Society, see [4.7].
The design criteria are written in a partial safety format and are suited for reliability calculations.
DNV AS
(77)
Appendix D
where:
(78)
(79)
(80)
(81)
(82)
σ10 and σ20 are the design stresses including load factors as given in [4.3].
The partial safety factors κ and γm are given in [4.4].
FE should be taken as:
(83)
ρ is found from:
(84)
(85)
(86)
(87)
(88)
DNV AS
Appendix D
(89)
The classical buckling strength parameter ΛCL can alternatively be assessed using direct methods as
indicated in [4.7].
(90)
(91)
In practise the problem will be to identify the load combinations that is most severe for the sphere with
respect to buckling, i.e. to identify the design stresses σ10 and σ20 and their positions that produce the most
critical buckling condition. All four combinations of σ10 andσ20 should be checked to identify the most severe
situation with respect to the buckling criterion in [4.2.1].
The load factors γfi are given in [4.4].
Normally it is sufficient to calculate the stresses along the meridian 90° off the centre line.
γf1 = 1.2 for forces caused by static weights and inertia forces
(92)
(93)
where:
(94)
(95)
DNV AS
Slenderness factor κ:
Appendix D
(96)
(97)
(98)
(99)
(100)
4.5 Tolerances
The imperfection amplitude or deviation δi is the light opening between the spherical shell and a segmental
template with the nominal radius of the shell. Two different imperfection amplitudes are defined:
δ1 is measured by means of a template of length
The value of δ1 to be used in the calculations will be decided by the Society taking into consideration the
manufacturers experience from previous production, shell thickness and type of material. An acceptable
value for δ1 is:
(101)
where:
(102)
or
(103)
(104)
DNV AS
- aluminium tanks:
Appendix D
(105)
(106)
- aluminium:
(107)
(108)
σ1,CL and σ2,CL are the principal stresses at elastic buckling found from a linear eigenvalue calculation using a
recognised FEM program package or equivalent. The load parameter ΛCL calculated by this procedure is then
used next in the definition of FE in eq.(83), and the design procedure in [4.2.1] is otherwise unchanged.
DNV AS
Appendix D
5 Definitions – skirt
σx = axial stress, vertical direction in skirt
σθ = hoop stress, circumferential direction
τ = shear stress
σx0 = design stress in vertical direction including load factors,
σθ0 = design stress in circumferential direction including load factors
τ0 = design shear stress
σe0 = equivalent design stress including load factors
Note:
Sign convention: Positive stresses in compression
---e-n-d---o-f---n-o-t-e---
af =
as =
aR =
γs =
γR =
DNV AS
As = cross-sectional area of vertical stiffener excluding shell flange
Appendix D
AR = cross-sectional area of ring excluding shell flange
Is = moment of inertia of vertical stiffener including full stiffeners
IR = moment of inertia of ring including full ring spacing ℓ
E = Young’s modulus
ν = Poisson’s ratio
L = total cylinder length between stiff radial supports
ℓ = length between rings, length between stiff radial supports
ΛE = elastic buckling parameter including knock down effect from geometrical imperfections
λE = slenderness parameter.
Subscripts:
6 Definitions – sphere
σ1 = largest principal stress (always the largest compressive stress, normally in the circumferential
direction for LNG tanks) > 0 always due to present sign convention
σ2 = minimum principal stress (compression or tension, normally meridional direction)
σ10 = design stress in circumferential direction including load factors
σ10 > 0 always; i.e. compressive
σ20 = design stress in meridional direction including load factors
σe0 = equivalent design stress including load factors
DNV AS
Note:
Appendix D
Sign convention: Positive stresses in compression
If both principal stresses are in tension the present buckling criteria are not relevant.
---e-n-d---o-f---n-o-t-e---
7 Tables
Table 1 Relevant buckling modes to be checked for LNG skirts
shell
buckling γs = γR = as = aR = 0
cylinder length =ℓ
γs = γR = as = aR = 0 γs = γR = as = aR = 0
cylinder length =ℓ cylinder length =ℓ
ΛE minimised with
m = 1,2..; ΛE minimised with ΛE minimised with
n = 0,1,2…; m = 1,2..; m = 1,2..;
α = 0.0,..,1.0 α = 0.0,..,1.0
panel
stiffener Not relevant γR = aR = 0 γR = aR = 0
buckling cylinder length =ℓ cylinder length =ℓ
DNV AS
Type of skirt cylinder geometry
Appendix D
Buckling Ring stiffened
Vertically stiffened Orthogonally stiffened
mode type (Unstiff. between rings)
panel
γs = as = 0
ring
cylinder length =L cylinder length =L
buckling
ΛE minimised with ΛE minimised with
m = 1,2..; m = 1,2..;
n = 0,1,2…; n = 0,1,2…;
DNV AS
Appendix D
8 Figures
UPP. DK.
TANK SKIRT
DNV AS
Appendix D
DNV AS
Appendix D
Figure 3 Cruciform joints
DNV AS
Appendix D
9.2 Elastic buckling strength
The principal membrane stresses at the point of elastic buckling is calculated as:
Where:
Λreq = γsum
γsum = κγm , κ and γm are given in [3.4].
The buckling design criterion in [3.2.2] (g > 0) is rewritten as:
ΛCR ≥ Λreq = γsum
ρ is found from [4.2.1] eqs.(84) and (85) and ΛCL from [4.2.1] eq.(88).
ρ = 1 corresponds to ideal elastic buckling (classical) condition.
DNV AS
Appendix D
Where:
Λreq = γsum
γsum is given in [4.2.1], eq.(82).
The buckling design criterion in [4.2.1] (g > 0) is rewritten as:
ΛCR ≥ Λreq = γsum
DNV AS
CHANGES – HISTORIC
Changes – historic
October 2018 edition
This is a new document.
DNV AS
About DNV
DNV is the independent expert in risk management and assurance, operating in more than 100
countries. Through its broad experience and deep expertise DNV advances safety and sustainable
performance, sets industry benchmarks, and inspires and invents solutions.
Whether assessing a new ship design, optimizing the performance of a wind farm, analyzing sensor
data from a gas pipeline or certifying a food company’s supply chain, DNV enables its customers and
their stakeholders to make critical decisions with confidence.
Driven by its purpose, to safeguard life, property, and the environment, DNV helps tackle the
challenges and global transformations facing its customers and the world today and is a trusted
voice for many of the world’s most successful and forward-thinking companies.