TECHNISCHE UNIVERSITEIT Laboratorfum voor Scheepshydromechana Archlef Mekefweg 2,2828 CD Deift Tel.: 015- 788813 - Fac 015-781838
IMPROVED DESIGN METHODS FOR SPANNING OF PIPELINES
LB. Bryndum
H.]. Vested
K.G. Nielsen
'f-9 89 Eie an Seminar
IMPROVED DESIGN METHODS'iFOR SPANNING OF PIPELINES
3 :c
-Authors
M.B.Bryndum
H.J. Vested
K.G. Nielsen
H. Gravesen *)
Danish Hydraulic Institute
Agern Allé 5
DK-2970 HØrsholm
*) Dansk Geoteknik A/S
Granskoven. 6
DK-2600Glostrup
ABSTRACT
The paper describes a set of rational calculation procedures for
determination
o.the fatigue damage and niaxirnum stress o
free
spanning pipelines, exposed to wave and current induced
hydrody-nainic loads. The procedures are described in a Project Guideline
or. Pipeline Span
valuation Manual and is supported by aPC-bascomputer programre. The programme handles the extensive and to
some degree tedious
computations required for transformation of
the hydrodynamnic input in terms of waves and current into
hydro-dynamic response of the pipeline span.. Both the response due to
wave action and the hydroelastic vibrations caused by vortex
shedding are included in the assessment of the pipeline
behavi-our.
The structural model deals with a single Span and is based on a
simplified beam-column model associated with elastic foundation
supports. The mddel incorporates the flexural induced normal
for-ces as well as any residual normal force from the lay. profor-cess o
LIST OF CONTENTS
ABS TRACT
INTRODUCTION
1.1 Backgroun4 fQr the Project
1.2 Improvements to Free Span Desi..
PIPELINE SPAN MODEL TESTS 2.1 General oxrtments
2.2 MOdel Tests
DEVELOPMENT OF. COMPUTATIONAL MODEL FOR FATIGUE INVESTIGATIONS 3.1 General Comments 3..2 ydrographic Data 3.3 Soil Interaction 3.4 Structural 4ddel 3.5 Data Base 3.6 Damping 3.7 Real Spans
3..8 Damage Calculation Procedures
4 MODEL FOR EXTREME LOAD CONDITION
4.1 General Comments
4.2 Funcion1 Load
4.3 Extreme Vertical Load
4.4 Extrémé Horizont1 Load
4.5 Voti MisesS tress Check
5. ACKNOWLEDGEMENTS
REFERENCES
APPENDIX A: EXAMPLE OF APPLICATION OFPROGRAMME ...DHISPAN
1. INTRODUCTION
1.1 Background for the. P.roiect.
.Frèe spanning pipelines may develop
for a number of reasons.
Pi-peline spans may be constructed intentionally or. may be the
re-suit of certain activities or natural processes
taking place
af-ter pipelines installation. On purpose built
free spanning
pipe-line Sections can be found within platform areas, at expansion
offsets, at pipeline crossings etc. Natural irregularities
of the
seabed may prevent a continuous contact and thus give
rise to
free
spans.
Furthermore, erosion of the sea bed may lead to
under-scouring of the pipeline and hence to free spans. Finally,
the trenching operation often leads to a
nuirber of free spans
especially if hard bottom or boulder clay are encountered.
The inspection fOr detection of spans and the Subsequent
verifi-cation of critical spans have proven to be very costly.
Therefo-re, a strong incentive exists .to
establish realistic and
practi-cal methodologies for evaluating all risks associated
with
pipe-line spans. The assessment of the risk associated
with flow
indu-ced vibrations have, to great extent been performed
using methods
which have been established for free pipes. Experimental
investi-gations have shown that the rnethçds
o some extent were
inadequa-te when dealing w.it.h the free spans near the sea bed.
..A number of studies have therefore been
initiated with the
purpo-se of improving or developing methods for free span. aspurpo-sessment.
A free spanning pipeline is. a flexible structure
susceptible to
large defornaio,ns..It has in general anui±er.o.well:defined
natural modes and eigenfrequencies. When exposed to
cyclic loads
very large responses may develop when
the frequenc.y of the
At. present such hydroelastic response cannot be. accurately pre-dicted using numerical Or analytia1 methods only,, but have to be determinethusingprocedures' which include the resuits'of model or
full' scale tests. .
In.l984/85 an extensive model test programme was carried out at
the DanishHydrau1ic nstitute. to determine he hydroelastic
re-sponse of a free spaninq pipeline exposed to a wide range of
environmental conditions. The reslts of the programme in terms
of different amplitude response parameters represent a
comprehen-sive database which not only can be used directly to assess the
hydroelastic response of a free Span, but can also be used to calculate the fatigue damage induced by the environmental loads
However the use of the data. base for practical cases generally
4.nyolves a large number of calculations and a considerably amount
of work in order to establish the fatigue damage induced over a
longer period of time. A project was therefore formulated with
the purpose pf. developing a PC-based computer programme which,
based on the comprehensive model test results, would perfor the
extensive and somewhat tedious calculations required to establish the fatigue damage. A number of. methods and procedures had to be developed in order the generalize te test results, i.e.
structu-ral model, soil interaction description, damping, fatigue
cal-culation procedure It was. found that he most rational procedure
would be to develop a guideline or a manual for the assessment Of
free spans and to supplement this manual with the computer
pro-gramme.
The project was proposed to and.spported by the industry and the
present. paper gives an , QVerVieW of the manual and the computer
programme and further .illutrates. the use of' the programme by an
1.2 Improvements, to Free Span Design
A uniber. of methods for the design of free pipeline spans have been. presented in project guidelines or in the oen literature,
Ref. 3, 4, 5 and 7. Some of these are based on experimental
in-vestigations in steady current and as. such. operates with distinct fiow.-.regiifléS' in which either in-line or cross flow vibrations
exists. Others operate with application of more extensive data
material and refined numerical modelling.
The present Pipeline Span Evaluation Manual in several ways
re-presents significant improvements to the assessment of free
spans, and as such the Manual is one major step towards the gene--ral' Design Guideline applicable for all span configurations ant environmental conditions. The most important improvements to the
assessment of free pipeline spans 'introduced in the Manual are
listed belOw:
The fatigue a.lcülations are baSed on a database containing experimental results from a large number of model tests
cove-ring a wide range of environmental conditions. The fatigue
damage is thus established through interpolation between ex-perimental results.
Detailed and extensive environmental iiput caxt be handled
efficiently eliminating the need for simplification of the
problem for corputational reasons.
The structural model includes important non-linear effects
such as. the additional axial force induced by the lateral
deflection.
d) The calculationS can be carried out:using a number of
diffe-.rent'SN-curves according tothe users requirements.'
--The-. calculationS are performed,: including the effect of
structural, soil and fluid damping. The vibration amplitudes
of lông flexible spans are found using an energy budget
2. PIPELi SP MODEL TESTS
2.1 GeneraL çprnments
The key element in the Pipeline Span Evaluation Manual is the
data base containing theresults of a large nuther of model tests perorrned in the laboratory. The database contains both response
- amplitude paametes and reference fatigue damage 'parameers. The
laboratory investigations have been perfOrrned using an idealized
model Of the véy complex prototype phenomenae. To fully appraise the procedures and caculations required when, transforming the
test results to protOtype fatigue damage, the model test investi-gations are briefly described.
2.2 Model Tests
The hydroelastic vibrations of a free spanning pipeline exposed
to wave and current action have been investigated in scaled-down
model tests. The objective was to etablish a database containing
high quality data, descri.bin.g response amplitude parameters for free pipeline spans.
The Model
2-i
The model consisted of a spring thounted rigid pipe segment, which
ws 0.74. in long and O.1Oit indiameter. The seabed was modelled
by a flat plate. The pipe axis was vertical and in. this way the spring and daztipthg system could be placed above the water. The
spring system consisted of four flexible steel rods arranged in a
squae configuration.
The "structural damping"of ,te system
could be varied by apply4ng ar-i external force,., opposing the pipe
movements in phase with the vibration velocity. The principle of
-iO TO R
MOVING CARRIAGE
t.UME BOTTOM
TOWER WITH SUSPENSION
SYSTEM oq MOEI
HORIZONTAL PLATE1
TEST PIPE
PLATE SIMULATING THE SEA BED
Fig. 2.1 Model Set-Up in Flume and Elastic Suspension System Instrumentation
The model was equipped with a
number of force transducers andmotion gauges which measured the following parameters:
- Pipe motions and accelerations in two perpendicular directions. - Reaction forces (total forces on pipe segment) in two
perpendicular directions.
- Damping forces in two perpendicular directions. Scope Of Work
Tests have been performed for the following environmental
condi-tions:
-
Steady Current - Regular Waves- Combined steady current and regular waves
The four most important non-dimensional parameters in relation to the model tests were identified to be:
The gap ratio
e/D0
The reduced velocity based on the wave induced: flow velocity
Uw,
Vr - f D
no
The reduced velocity based on the steady current
U
r
fD0
The Keulegan-Carpenter number
UT
KC =
D 0
where
S : the distance between the sea bed and the pipe
D0 : the pipe diameter
U : maximum wave velocity
w
T : wave period
natural frequency of the oscillating system steady current velocity
Following ranges of the parameters have been investigated:
- Keu].egan-Carpenter number, KC, from 5 to 120. - Reduced velocity, Vr from 1 to 12
- Reduced velocity, U, four values 1.0, 2.5 and 4.0 - Gap ratio, e/D0, three values 0.0, 0.5 and 1.0.
2-3
TEST MATRIX (GAP/DIAMETER RATIOS, olD 0.0.5.1.0)
O REGULAR WAVES
o REGULAR WAVES STEAOY CURRENT
Fig. 2.2 The test matrix
All tests ifi the matrix have been performed with one damping (K5
= 0.4). A number of tests have been repeated with other values
for the damping in order to quantify the effect of varying this
parameter.
Data Aialysis
The immediate result of each test was a number of time series
describing displacements and forces. Such time series are. imprac-tical for use, and therefore these were analysed in order to
pro-duce parameters which were easy to interprets. Fig. 2.3
illu-strates the parameters derived from the displacement time series
X. NORMALi CIPtACEMENT
_____
LTItiitiLiItMY11TTWi
!1i1III'IIRN
In PEIO0 TEST CUATlON 10.0 20.0 .0 TIME 1Fig. 2.3 Basic Results of Time Series Ana1yss. Disp1acement
are Normalised with with the Pipe Diameter.
- REDUCE-P VELOCITY, V, 1 2 3 4 5 6 7 B Q '0
ii
12 IL) z U 5 000 -10 o ooO 00 000 -0 0-15.
0 000..
0000. .
.o':---;1-. -- -_ -- 000 0000 0000 S 000 0-- -- - -0 00000000000 SSI
00 0 0--
- -40.5.14..000ao:0000000eeo 00 oo a 0 0 -U000000
- 00 00S 0_
0 S. S - - -80 0 00 00 0 0 0 0 0 0 0 0 - -00 -o 0.3 MAX IX) 0.2r
2-5
The Results
The parameters found through the analysis are listed below.
1. Mean (x) : Mean value of total record
2.: Max (x) : Maximum value of total recàrd
Mm (x) : Minimum value of total record
Std (x) : Standard deviation of total records
Max(x)
Maximum peak to peak value,(maximum "wave height")
Min(x)
: Minimum peak to peak valueMean (x)
: Mean of peak to peak valuesStd(x)
: Standard deviation of individual peak to peakvalues
Mean (T) : Mean zero up-crossing period
x is the displacement time series (in-line or cross-flow) norma-lized by the diameter of the model pipe.
Fig. 2.4 shows an example on the response amplitudes
Std(x)/D and Std (y)/D for the regular wave tests.
l.2 I.. 1.1 I.. 3.4 3.2 a.. II 12 3 Vr STD (V ) /
.. RAWDTA
- CALCULATED DATA tI e 12 VrFig. 2.4 Std(x)/D and Std(y)/D
against Vr for KC40 and e/D =
0.5. The calculated data are determined
by
interpolation and extrapolation from the measured
values. STD(X)/D
Ref. 9 presents the results of a similar but less extensive
stu-dy. The results do in general agree with the present.
The tests have been analyzed furthe in. order to pro4uce fatigue damage parameters The analysis comprised peak to peak analysis
according to the rainflow
counting
method and transformation ofthese peak to peak values into fatigue damage parameters. In-line and cross-flow fatigue damage parameters
D (x. ./D)
-x N0
-pp,i
a = .2,3,4,8)(Y/D0)m
(In = 2,3,4,8)x and y are all peak to peak athplitudes in One test defi-pp,i pp,i
ned according to the rainflow counting methods. N0 is defined as
fITT :natural frequency times test duration) and in principle is
thenu±nber of vibrations in one tet . is oUter pipe diameter.
The calculatio of fatigue damage parameters is described in
3-1
3 DEVELOPMENT O COMPUTTIONAL MODEL FOR FATIGUE INVESTIGATIONS 3. 1 eneral Cornxnents
The Span Evaluation Manual nd the computational model are
struc-tured in the same manner and the calculations perorIned by the
programme can be carried out by hand following the instructions
of the Manual. The key element is the database
contanig the
results of the experimental iivestigations. The database is in
the Manual presented in the form of a series of contour plots of the Vaious'parameters, see example in Fig. 3.5.
The procedures and methods required for transforrnaton of the
en-vironmental input, the soil-, the pipeline- and span data etc.
iflto relevant vibration data and fatigue data are
described
irthe Manual The same procedures are implemented, in the programme
in a user friendly interOtive menu
driven tructure. Thepro-gramme automatically generates a number of entry. parameters to
the data base, retrieves the appropriate data and calculates the accumulated fatigue damage.
The sequence of ca1cu1tions are illustrated, on Fig. 3.1 next
/SPAN CONF0URATION OAT4 PIPE SYSTEM DATA
SOIL. DATA
STRUCTURAL ENVIRONMENTAL MODEL MODEL
y
RESULTS AND ENTRY
P.RAMETERS FOR DATA BASE:
IcC.Vr.Ur.eic.
DATA BASE FOR OSCILLATIONS AND REFERENCE DAMAGE CALCULATION OF ACCUMULATED REFERENCE DAMAGE DAMAGE
Fig. 3.1 Span Analysis Sequence
3.2 Hydrogaphic Data
The basic hydrographic
data required for
thespan
assessment arethe wave and current condition on the location. These conditions
are required in a forma.t which is commonly used in offshore app-lications.
The waves are given in terms of appropriate long term distributi-ons, e.g. omnidirectional scatter diagram of H, T.
Each seastate can be described by a standard Jonswap or
Pierson-MoskOwitz Spectr1n. The boxes (H5, T) in the scatter diagram can be given a directional distribution. Once the directional
distri-bution has been specified, the surface wave pattern is defined
-for the period ccnsi4ered.
HYOROGRAPPIIC DATA / FATIGUE CRITERIA LEGEND RESPONSE AMPLITUDES INPUT CALCULATIONS PROCEOUE DATA FILE.! RESULT
I
SCALING TOREAL SPANS /DAMPINGPARAMErERS -/
/.
-/
DAMAGE RATIO
The surface waves ae transferred to the sea bed using linear
wave theory.
Thereby
the velocity spectrum at the sea bed is
found. This spectrum is SubSequently decomposed into single waves
assuming a Rayleigh distribution of these The spectral derived
mean period is assignèdl to each wave so that the duration of all waves equals the duratiOn of the original sea States.
The ure
in the. pipe level can bespecified
in two independentcomponents:
A constant component that is always presert
A cOmponent which may vary in fttaghitude. and direction
The current is combined with the indiiidual waves and each wave
and c'rrent combination defines what is called a "single event". The events are characterized by following parameters:
wave velocity at pipe level. wave period
steady Cu ent at pipe. level:
TDIJR : duration of event
Only the velocity cotrponent perpendicular to the pipe axis iS
considered in this context The calculation procedure for
S...( r) SU( t) d 11 T TOUR
T, LJ
TOUR Uc-Uw SURFACE SPECTRUM H51T,, LINEAR TRANSFORMATION BOTTOM VELOCITY SPECTRUM RAYLEIGH DISTRIBUTION SINGLE WAVES CURRENT DISTRIBUTION DISTRIBUTION OF SINGLE EVENTSFig. 3.2 Calculation of Individual Wave and Current situations, "Single Events".
The parameters characterizing the single events are used to
defi-ne a number of non-dimensional parameters which are used as
en-tries to the data base when retrieving amplitude or fatigue data. The parameters are KC, VrF Ur and e/D0.
3.3 Soil Interaction
tnoder. to describe the i
;cioeen. the pipeline and the
surrounding soil, a number of representative, soil parameters has :to be speified. ..
The; strutural model
of the spanning pipeline requires themodulus of subgrade reaction and. the axial coefficient of
friction. Further, to include
the effect of soil damping as
described in the QllQwing section, the transverse coefficient of friction and the soil damping ratio must be specified as well. For cases where sufficiently dèt'ailed,inforrnation on soil
charac-teristics is available, the parameters to be specified can be
deducted through analytical and empirical relationships as
de-scribed in the open literature, Ref. 1 and 12, and referenced in the Manual. Alternatively, for cases where only a rough
descrip-tion of the soil can be obtained, guidance can be found in the
estimated soil parameters of Table 3..l below. The table i.s ainly
based on Ref. 7.
Frjct.ori Coefficient 3-s
Very seit clay 1-10
Soft cLay 3-33 Aial soil resistance
force is iA the fE
ediu clay 9-33 of adhesion
a.rd clay 30-67
A.xial. (1ongitiidial) Tra.nsverzal (lateral)
Concrete Steel,epoxy Concrete Seel.epoxy
miniu 0.4
- 0.4
- 0.2
- 0.2
Table' 3.1 Soil Parameter Ranges for Various Soil Types Sandy clay! moraine clay 13-140 0.4-L 0.4 to 0.5 t.00se sand 5-13 0.3-0.9 0.2-0.5 0.3-0.7 Dense sand 25-48 0.2-0.5 0.5-0.7 Silt 1-11. 0.4-0.5 Rock 550-52000 0.5-1 0.4-1 0.7-2 0.4-2 Rock dith a.rine growth 550-52000 0-1.0 0-1
3.4 Structural Model
The pipeline span çonfiquatiOnmOde11ed in the. programme
repre-sntsthe common
jÜáiOfl wherë the spañ develops through a
scour process. The pipeline is installed in horizontal position on the sea bed . In this situatiQfl the pipeline is pressure
tes-tèd,an
prior to pipeline commissibfliñg and operation the pipe-line can be trenched or rock covered. A scour hole is formed over a certain length in any of these situations and a free span deve-lops. In addition to the axial forces already present in the pipean axial tension component developS in the
span due to
thedeflection and the axial friction along the supported sections.
The situation is illustrated in Fig. 3.3.
SCOUR NOUCED
:..:.:::.::;:::;:;::::::::.;....
-
_..._i-Fig 3.3 Scour Induced Free Span
Static Model
The prototype free span is described through analytical
expressi-ons giving the deflection, the tension and the bending moments.
The analytical expressions are develOped base4 on the idealized
ELASTIc FOUNDATION
3-7
FREE SPAN
Fig. 3.4 Idealized Model of Free Span
The structural model calculates the bending moments,midspan and at the end suppots, as well as the corresponding deflections.
The relationship established, between deflection and. associated stress and strain variation is utilized in the subsequent fatigue dathage and extreme
ipa
lysis. Further the natural frequencyof the pipeline span is calculated.
The interaction between the supporting soil and the pipeline is modelled by application of the Winkler model, assuming that at
any point along the pipeline within the adjacent soil, the
trans-verse reaction forces are prcportional to the beam delection at
that point.
The effect of the axial force on the pipeline .eflection and ben-ding moments
as
well as the influence on the xatural frequencyhas been included. The resulting axial force, is composed by the
linear axial force, which is present before the span is formed, and by the non-linear axial force' induced by the deflection of
the pipeline. The lInear axial force is the result of functional
loadings and the poible residual lay tension.
The weight of the spannig prt of the pipeline is supported by
r'eácton forcis,
a
eachend
f the span. 'The flexüral induced'axial force s gradually t'educed' by axial frictibñ fórcès along
the buried sections of the pipe until the normal forci equals the original value (lInear axial force component).
ELASTIC FOUNDATION
Dynamic Model
The same structural model is used to establish the natural
fre-quency of the pipelIne span for the first sytninetrical vibration
mode together with the second mode. The effect of the axial for-ce, composed of the linear and non-linear components, is included in the formula for eigenfrequencieS.
It has been found that for cases in which the pipespan vibrates
in its symmetrical first mode, the ratio between the maximum dis-placement amplitude at the mnidspan and the corresponding bending
moment respectively at the midspan and at the soil shoulder are
very similar to the ratios found for the case of uniform static
loading acting on the span. The ratio between the dynamic momen
and the deflection at midspan (mode shape factor) is calculated
using the formulas and parameter relationships developed for uni-form static loading.
The theoretical work leading to the various analytical expressi-ons and formulas associated with the established structural model is mainly based on Ref. (8), Hobbs (1987) and Ref. (6), Hetenyi
(1946). Generally the calculations are based on analytical
ex-pressions, but an iterative calculation scheme has been required, because most of the parameters are implicitly given in the
equa-tions.
Fatigue calculation
The accumulated damage and the associated fatigue life, is
calcu-lated using the rule of accumulation
of the partial damages(Paingren-Miners Law) .
The basic assumption according to this
summation method is that the damage to the structure for each
load cycle is constant for a given stress range. The relationship between the stress range and the corresponding allowed number of
stress cycles (the SN-curve) can be specified by the user prior
'C.>-'
3-9
The fatigue calculation .s performed in two steps. The first step
is bad"on a direct application
of the model test results andtheresult-is.-. a..refereflCe fatigue damage.. The test results
re.-flect the" hydrodyn'amiC response of a spring mounted rigid cylin-der for one p,eci.fic damping and therefore a second step., is
re-quired for converting the. reference damage to prototype value.
The. second. step includes an integrated scaling of the in-line and cross flow vibrations and assdciated fatigue aInages in order to
adjust the results to actual values of damping and incorporate the 3-dimensional effect of the flexibility of the actual span
configuration.
.3.5 Data Ease
The results from the model tests have been used to establish a
data base. The hydrographic 'conditions combined with pipe and
structural data determine values for 'KC, Vrl U and e/D. These
parameters define the
conditions
for the performed model testsand constitute the key to the da.taase.
The model tests data did. not cover a complete KC', VR range and
were not equally spaced. Therefore the original results were
smoothed and extrapolated by means of a second-order
two-dimen-sional
interpolation/extrapolation
procedure. The resulting datawere plotted as contour plots and visually checked. An example of such a contour plot can be. seen in Fig'. 3.5. In these plots the points for the original data are marked with a cross.
12' 10.2- 4-2- 4. /P#. / I I
Fig. 3.5 Example on contour plot of the hydroelastic Vibration data ontaine in. the data base
Data Base Key Range Parameter interval or actual values
KC
V
r
Ur
e/D0
AssOciated with each set of the key
parameters are 14 seleted
values of the condensed model test results.
The final data base covers following r3nges of the entry
parame-ters: e/Da = 1.0 Ur 0.0 0-120 0-12 0-4 0-1.0 0, 0, 5 0.5 1.0, 2.5, 0.5, 1.0 4.0 0 0 2 4 6 8 10 12 Vr e/Dó t.OUrO.O STD(X)/Do STD(Y)/Do
Contents of data base: Amplitude resp9nse 3-11 peak-peak, in-line low-pass, filtered high-pass, filtered peak-peak, cross-flow
standard deviation inline
standard deviation cross-flow Fatigue damage parameters (see Chapter 2.2)
D (in = 2, 3, 4, 8) : damage. parameters, in-line
D (in = 2, 3, 4,8): damage parameters, cross-flow
The vibration and fatigue parameters stored in the data base can be used for assessing the fatigue damage. The fatigue damage pa-r-ameters can be transfOrmed to represent the relevant stresses,
using the SN-ctrve arid the mo4e shape factor. The damage deterthi!.
ned directly from the data base corresponds to two-dimensional conditions, and a sealing of the model test results to a real
three dimensional span has to be performed in most cases together with an adjustment to the actual darnping conditions.
3.6 Damping
The vibration of a free spanning pipeline is a resonance
phenome-non and as such the damping has a dominating influence on the vibration amplitudes. Traditionally the damping is described by
the, stability parameter, defined.
.2 meo /pD0 Mean (x) /D0 Mean (x) /D0 LP Me an (x) ID0 HP
Mean (Y)
/D0 Std (x)/Do Std (y)/P0where
me: structural plus hydrodynamic mass per uniElénqth
p : density of water
DO: external pipe diameter logarithmic decrement
The stability parameter is a combined mass ratio and damping
pa-rarneter. The daping is expreSed by the logarithmic decrement,
can be dete*mined from decay tests in air and is the natural
logarithm 9J the ratio between two succeeding vibration
amplitu-des.
The total dampin is basically xtade up of three terms. Structural damping
Soil damping Fluid damping
The structural damping is relatively well known and values based
on measurements áan be found in the open literature.
The soil damping is mOre complex and only limited data are avail-able. The soil damping included in the Manual is therefore partly based on information from foundation theory (Ref, 12) and partly based on calculation of tife soil friction forces when the pip
moves
relativey
to soil. Three physical mechanisms areconsidered.
- Hysteretic a!nd radiation damping
- Soil friction along the pipe due to axial movements induced
by
êf1ectins
,.Soil friction due to transverse movements of the pie at the
Supports.
The fluid damping is closely related to the vibration amplitudes.
Even with very low damping the amplitudes of vortex induced
3-13
due to the fact that the regular vortex shedding will.
Collapse je the amplitudes significantly exceeds this limj d thus
the
fluid damping will reduce the vibrations.
'When exited the .prototpe ...span will ibrá with cOntinously
Va-rying amplitudes along the pipe
and
therefore the fluid dampingand the excitation forces will vary along the pipe. At each sec-tion of the pipe there will, be a netto energi
transfer from the fluid to the pipe or from the pipe to the fluid depending on
whe-ther the excitation forces or the damping are dominating.
This
overall principle is used when the results of the 2-dimensional
mOde.1tst are transformed to the
3-dimensional conditions of a
flexible prototype span
The relationship between the fluid damp!ng and the vibration
amplitude has been determined through a nuxtber of tests with
va-rying daiiping conditions. Tests have also been made
for
situati-ons where the pipe has been foced to oscillate with athplitudes
larger tha.n natural undarnped vibrations This way the energy
dis-sipated by damping or introduéd by excitation forces
has been
determined,
3.7 Real Spans
The model tests have been performed under strict
2-dimensional conditions. The flow was uniform along the pipe and the pipe
it-self was a rigid spring mounted pipe segment. To be
applicable under prototype conditions the test results had to
be scaled to compensate . for the differences between the model set-up and
the
flexible pipe span.
This scaling is based on an energy principle. Damping reduces the amplitudes and dissipates kinetic energy through
heat or
turbulence. The excitation forces maintain the vibrations and as
long as a steady state condition is present the
energy supplied by the excitation forces equals the energy dissipated by damping. This equilibrium may be written
Denominating ELSS to. E a relation between E an4 6, the loga-rithmic decremezt can be established for lightly damped systems
!
-(n)2.E(n-l)
2 Et
where is the maximum kinetic energy for the vibration consi-deréd. Et0t is proportional to the amplitude squared.
The energy equilibrium is estalishe4 for the flexible span ii
the following way, see Fig. 3.6. The span is divided into a
nuxn-ber of segments, each representing two dimensional conditions.
The energy input or dissiptiç for each segment can be
determi-ned assuming that te effect of structural and soil damping is
identical for all elements and that the fluid damping depends on
the actual amplitude only. By integrating along the span for a
given mid span vibration amplitude and geometry, the energy input and disipation can be determined. This. process is repeated for a
nuin.ber
of midspan
amplitudesuntil
the correct energy balance isfound.
CD
\\I
CD
MODELLING OF FLEXIBLE SPAN
NETTO ENERGY TRANSFER
3-15
0
ENERGY TRANSFER,FROM FLUID TO PIPE (DRIVING FORCES) ENERGY TRANSFER,FROM PIPE TO SURROUNDINGS(DAMPING)Fig. 3.6 Definition sketch for the energy balance for a free
span.
The vortex induced vibrations are scaled according to the
proce-dure described above.
The wave induced in-line movements are
scaled, usin.g the correct value for the cross flow vibrations to determine the effect of the drag forces.
The scaling is performed in three steps. First the cross flow vibrations are scaled, secondly the high frequency part of the
in-line motions (vortex induced) and thirdly these two sets of
vibrations are combined to give the total
in-line
motion3. 8
Damage Calculation Procedures
The basis for a.l the
fatigue calculation in the Manual is the
?alzngree-MiI1erS Rule
p = Z D
= 1ALL,i
D is the total 4amage, N
is the number of cy1es with a strain
variation
d NALLi is the allowable number of cycles with
a strain range o,f
The failtiré criterion is D1..
The allowab.e nmer of cycles,
Lis found from SN-curves c
the general form
NL =
C()m
where C is a cthistant and the exponent in normally
have a value
between 2 and 8.
The fatigue damge for a certain strain range
Amay thus be
written as
D. N.
'1
iC 'ti
The sréss range
to be used in the fatigue
damage c'aiculation
are found the, following way'..
The environmental conditions
cotnbned with the soil and pipeline
span data defize a sere.s.of "single events".
Each event
is
associated with
a
duration,
TDUR,jl
and
the
fatigue
damage
parameters D
and D
scaled to reflect prototype conditions
The midspan
defiectiofl
and
the
strain
in
the most critical
section is relatéd through the
mode shape factor,
K1,
and the
modulus of elasicitY.
c.
= K. E(Y
./D
Similar expression is valid for the in-line direction.
The fatigue damage induced by each "single event" can then be
calculated: (EK1)m
DAM,
= C f (EK4)m DAM . = y,i where E: Youngs modulusK1: the mode shape factor
The total damage is
found
adding the effect Of all the singleeVents.
DAM = Z DAM
-. ,xii
DAM = Z DAM
y,1
DUR,i f D7
4. :MODEL FOR EXTREME LOAD CONDITION
4 1 General Comments
The purpose of the extreme load analysis is to calculate the
ma-mum stzess inuàed by the normal operational loads and the
an-ticipated extreme envirQn±netal load. to unamnbigious efinition of the extreme load is possible and the final acceptance criteri-on for the calculãtéd stresS has to be evaluated accordingly. The
maximum stress fo each load case considered is fOund using von
Mises stress criterion.
Three different lcad cases are considered:
1. Fu±ictional loads.
Analysis of pare functional loads
This analysis defines the properties ofthe span (natural
frequency, an4 axial force) in the static situation.
Extreme vertial load,
Functional anã. environmental loads in the vertical plane. ctreme horizântal load.
Functional and environmental loads in the horizontal plane..
The cothbined
load,
cases 2 and 3 are both investigated assumingthat the naturai frequency is equal to the frequency found in
load condition 11 The tension includes both 'linear"
contribu-tions and the fi ural induced contribution orreSpbnding to
4.2 !uncti9na1oad
The functional lOads .iSa stati load case: and the normal force
and the moments are found using the procedures outlined in
Chap-ter 3. These forceS are used to calculate the stresses in the
pipe wall1, and the maximum conthined stress check can be
perfor-med. '
-in
addition to the moment and nonal force, the natural frequencyis found.
4.3 Extreme Vertical Load
The analysis is carried out usiig the natural frequency deter-mined under operational conditions. The frequency is
usd to
de-terrnine the actual value of Vr and Ur corresponding to the
selec-ted extreme load case. Correspondingly a KC.-nunther. is defined.
These parameters and the gap ratio are used as etitries' to the
response amplitude database
and
a maximum peak to peak response Ypb/DQis fOund.'
The response found may be significantly smaller than the posg:ible ma,drnum. If a larger response, can be found forvalues of Vr and Ur which are smaller than those found for the
extreme load case it is recommended to select the largest respon-se for the calculations.. It meanS that resonance and synchroniza-tion with vortex shedding takes place for Smaller. fiow velocities
than those. induced by the 'extreme load case. "
The bending moments are found assuming that the maximum amplitude midspan is :.'
4-2
4.4 Extreme Horizonta]. Load
The.-.ana'lysis- is-carried .out::-Similarly to the analysis- in the
vertical plane. . ---'.. . -:.
The actual values of V,tJ,: KC 4nd. e/D0 are used
the response -data base -and
the maximum peak:
response x/D0 is found.. Again, should a larger for -lower .v values than the- extreme,
r
..
-these. The bending moments are amplitude mid spn is
x =
4.5 Von Mises Stress Check
The stresses foufld in the various load cases are combined using
Von Mises Stress, Check. .
Von --Mises stres is usually defined by the stresses in the main
irect--ions of t1e pipewall (resp. hoop and combined longitudinal
stresses)
as entries to
to-. peak ...otal
response occur it is ecothmended to use
found assuming that the maximum
/2
+ 2- aH(L ±.
The stress condition in the pipe wall can be considered
2-dimen-sional and the fbliowing components are included: Hoop stress : Effect of pressure difference
Normal stress Effect of axial force
Bendiig stress ab Effect of bending thoment fQm.functional and extreme loads
5. ACKNOWLEDGEMENTs
The wok is part of
a Joint Industry Project "Spaiting ofPipelines1' sponsored b the fo1löwin companies: EP thternãtjbnal Ltd. (UK)
CoPoco Inc. (USA)
Department of Energy (UK)
banish Oil & Gas Production A/S (DK) Exxon ProdUction Research Company (USA) Norsk Hydro A/S (N.)
Statojl A/S (N)
The authors iish to thank the Companies and
the members of the
Steering comittee fl1owing
the Poject for their encoiraementand permission to publish the material.
It is noted that
the
ideas and opinions expressed in this
paper are those of the
authors and have not been approved or endorsed
by any o
thesponsoring companies
6.
REFEREtiCS/1/
Audiert & Nyrian (1977), "Soil Restraint against
Horizontal Motion of Pipes", ASCE, VOL. 103, GT1O.
/2/
Blevins, R.D
(1979), "Formulas for Natural Frequency and
Mode Shae", VanNostrand Reinho]4 Company, New York.
/3/
Bruschi,
R.,
Montesi,
K.,
Ragaglia,
R..and Tura,
F.
(1987), "A New Boundary Element for Free Spafl Aflalysis'!
OMAE 1987.
/4/
Bruschi, R. and Vitali, L.
(1988), "Large Amplitude
Os-cillations of. Geometrically Non-Linear Beams Subjected to
Hy4odynamic Excitation",OMAE 88, Additional Paper.
/5/
DetNorske Veritas (1981), "Rules for Submarine Pipeline.
systems.
/6/
Hetény.j. (1946), "Beams on Elastic Foundation", University
of Michigan.
/7/
.Hinstrup, P0, Coiquhoun,
R.S. and Gravesen,
H.(1985),
"Danish Submarine Pipeline Guidelines", 1st draft,
prepa-red by
anish Hydraulic Institute and Raxnbl1
&Hanne
mann.
/8/
Hobbs,
i.E.
(1986), "The Effective Length of a Pipeline
Free
Spain", 5th OMAE, Tokyo.
/9/
JacbbSen, V.,
Bryndum, M.B., Nielsen,
R. and.Fines,
S"Vibrations of Offshore Pipelines exposed to Current and
Wave Acion., 3rd International Symposium on Offshore
6-2
/10/ Mousselli, A.H. (1977), "Pipe Stresses at the Sea Bed
during Installation and Trenching Operations", OTC 2965, Offshore Tech. Conf. Houston.
/11/ Naess, A.A. (1985), "Fatigue Handbook, Offshore Steel
Structures", Tapir, Trondheim.
/12/ Richard, Hall & Woods (1970), "Vibrations of Soils and Foundations", Prentice-Hall International Inc., London.
/13/ Roark, R.J. and Young, W.C. (1976), "Formulas for Stress
and Strain", 5th ed., McGraw-Hill Book Company, Auckland.
/14/ Tsahalis, D.T. (1983), "The Effect of Seabottom Proximity on the Vortex-Induced Vibrations and Fatigue Life of Of f-shore Pipelines", OMAE and ASM.E Journal of Energy Resour-ces Technology.
/15/ Tsahalis, D.T. and Jones, W.T., "Vortex-Induced Vibrati-ons of a Flexible Cylinder near a Plane Boundary in
Stea-dy Flow", 13th Annual Offshore Technology Conference,
Houston, Texas, OTC 2991, 1981.
/16/ Tsahalis, D.T., "Vortex-Induced Vibrations of a Flexible
Cylinder near a Plane Boundary Exposed to
Steady andWave-Induced Currents", ASME Journal of Energy Resources
0tT
na1 (I)iitess J3
aI1 ()ijt
t1ni) ()Extena ixrro,oncotir, )
ciri
ti (/.3)
eigit xating (k/a)
tarna1 orrosiOn tir (kJ3)
Ccnt (k/3) EL4TiCIrf jiu of (PJ ct4Lra1 d31rb]
(1 'i rrt)
SI Rejal (kfl) EffRA11iKflmdC'ciui
ic1lx3, ZJ1 nr iP,)('
E:c.!e r'urn 'i,E*iit iu (Cttl 1) h1D
APPENDIX: ExAMPLE. QF APPLICATIQN OF PQRAMMEDHI$ PAN
The methOdology and procedures for, the Oalculation ,Qf the fatigue
damage and maximuth stress of a free spanning
pipeline as
descri-bed in the preet paper have been impl!mented.,i
the PC-based
computer programme DHISPAN.
The DHISPAN programme is a set Qf menu
riven pogramnxnes and
con-tains procedures
catng structural properties,
generati-on of envirgenerati-onrnenal input arid calculatigenerati-on of fatigue damage and
extreme lo4d stresSeS. Options are available for
displaying
ana-lyzed data on the: monitor or on a prthter/p1ottr.
The application of DHISPAN has been illustrated by running th
program with selected inpat data
epresenting pipeline, span and
fatigue parameters a4 environenta
data., Four span lengths have
been considered.
Input Data for Analysis.
Pipe, Span and Faigue Data
The input data for span length equal to 40 m is presented below
Npe }ieru .1 . lri. 9.3
0.73 0.8l3 0J37(3 ti3.i) 65.0 2100 0.uiCi)3 10.0 7.0 1!L1.l
DIRECTIOH n1a relative to nirth (eg)
L;m Span (a)
A-2
o.o
SOIL
Stiffre ()
: 2.SJ)3DnV 1977 Rules, Appendix C, A stress concentration factor of 2.0
has 'been aliéd.
Environmental Iiiput Data
The input is given as a wave scatter diagram. A directional
dis-tribution is applied fo# each box Of the scatter diagram. For
smaller waves the dominating directions are selected as waves
from SW, W and NW. For increasing wave height the directions W,
NW and. N are dominating.
Axial frictic ccfficiest :
Tr.vse frictic, coefficient :
Ocing ratio th teria r ation)
O.E))
i tO
'.iti ()
0.5I3AOOED Ccfficient C; : 1.1)3
E3co& reti.rn to Edit ru Itrl
The SN-curve has been chosen as the curve X (Tubular Joints) of
DHI( '
11M Scatt K 8.1 - V.s.dcWatDth(i)
ati of Eèa
Stata (:)'
Pea¼Periec)
b!o.,.'hn,M:
4,3.33 14.The current has been specified with directional dStiution and
associated probaiiit-y.
Csrt
eniB.258.33
Sig. Wave Hecgtz (:)
59.50.08
4.375 7.125 12.625 1.258 L8 1233.8 158 0.8 2.75 1G.8 333.8 433.13 13.0 4.259 8.8 8.8 33.0 18.8 5.758 8.9 13.9 .8 .97.8
8.8 13.0 8.9 .18.9ria1
-,Aiticn YretLrfl to Edit 1erU (Ctrl T> h,io
C4zret (a/z)
:0.3)
Cia-rent ircticfl relative to kcth (deg)
:0.33
t iti,Wtion
IE of Ittterva3, Hr i tax Soe*i (/) I 0.23 OM
Curt t.S3tributin beke Diroticial :ttibition Yj1
C&rent J F'robai1ity I '1 I 1 1 I ;i I I i
0.20 1. L13 0. 3.24 J.30 0.E 0.3 0.f.)) 0.3)0
Results of Analysis Strutural-Aia-iysiS
The results of the structural analysis for the span length of 40
rn is
shown
below1U T YSIS iDOITIuAL JTP'JT
LC):
(N/ti): 36.57
Vrtica1 Load qv (N/ti): 57
Hi:oiit31 L'i (N/a): I3.3
.:1 PIPE i)IL D:
3(11 (ti) :
ii (a) : 8.7&'3
tri ()
: O.Q19lEl
(2)
4E8E3 TAn3 (N) I 4E+i'J5-ff (Ks/a) 165 51 Kztt (KS/ti) 2Q ..2
n,1 (N/a. 2 i r'c I)5.
ML FRPJ1ETER PJ1O FETM L8Th:
LiO Caa : 4.U3 LabdaL :
of It tio: 33 6Ii*&ES T,I'N 3.4 Bet 0.151 :TTIC O'hTIi: 'aaxt JLT.4 Xaaxfl_ FRIcTtcN AO tffCtt( LE(TH: L3/t i.8 Lo/L : OThNIC FROPTIES:
Fnat (It) : 0.771 hapefa : 0.i118 Lcq :
Generation Of Environméntl Sing].eEvents
A-4
The hydrographic input is transformed into a large number of
sin-gle events at the pipeline level. The distribution o the wave
induced orbital velocitieS at pipelinerlevel, tJ, and the associ-ated periods together with the current
distribution
is shown ver-sus the relative duration for each event on the figure. below. ThevelOcities are for each event given as' the component perpendicu-lar to the pipeline.
Uw 1.50 1.25-1.00 0.75- 0.50- 0.25-0.00 1.OE CT Tzu 4 UU 12.00 iaoo 800 0.3 0.30 rim 0.20 0.15 0.10 0.05 0.00 ICE Uc IlIllUlIll i; ii
1.OEG6 1.OELO5 1.CELO4 1.0E1-03 1.0EL ICELCI 1.OE+00
t 111111 I I
6?°OE-0T ICEL05 iDE 05 iCE 04 1.CE-03 iCE ICE
07 iCE 06 iCE 05 i.OE-04 I CE-03 ICE iCE
PcI duration
BPeI duration
01 1.OE+00
Pelduration
01 i.OE'+OO
ing!eeents environment file:IBCEX.LC3T
Damage Calculation
A-6
Example of output for the damage calculation for the length of 40
in is shown below
T0T. E
0age Ca1c1atin Prii (days) Z.5.08
Irrtir 1.059 E-3
Cros-fka ø.18 E3 I 0It? CF SL1E airs DE
2 DISFU? CF SDE EUE{1S JLI} FACT
3 DI1A?CFSflG.E E{rS OATh EAi
4ftOT
5 R1T FtOTS
-..-
ATT OtAGMtoesuit
ri
(Cttl 1> ip1I4 0aaeRej1t ni 0.2 Vz. 0.i3
cz Daiage atter Nei 0.2.1 s. 0.W
SCATTER DiA1CF
!n-iincross-floi(N):
y Total DageDratiai of the C3lculation Pioi (da)
Oae is 1tipii with 1.EE3 Tp 4. 7.13 9.27 I2.b
1.5 9.::: O. 8.96 8.
2.75 O.3 e.r O.( 8
4. e.a 9.O 9. 9.072
5.75 O.CO3 0.( '3.121 8.E
O. 8. '3.9 8.733
Hs
rtr,iaticn Ratio: 8. Ertrlation Ratio: 9.14
Example of detailed listing of.
40 0.77
50 0.53
60 0.38
65 b.33
N KC
R XI
Ypo YSC?pp 0 C3I KY 1thoX ItnoY12)112.2 L4 02 0.02 0.01 .3.133 0.01 1.22 1.56 1.42 5 2 .272152 t 0 002001002002 122 155 142 12 15.2 1.5 3.4 0.02 0.01 022 0.01 123 1.56 1.42 1224 15.2 1.6 0.2 022 0.01 022 022 1.20 1.56 1.42 1235 12.2 1.2 0.0 3.2 !).02 823 022 1.22 US 1.42 £225172 12 14002 Out 002302 IJ 158 142 0 1227 172 U 0.3 3.02 0.02 323 13.02 122 1.56 1.42 S
H KO UP UP U W)y Xçç XpoLO Xçã1 Ypp Xstd ?d
122112.2 i 1.4 L -I.62 L 0. 022 UI 022 0.138 12 15.2 1. 2.0 -6223 312 3.07 3.08 022 0.01 322 0.31 1233 15.2 1.5 0.4-62.12 -3I.07 0.16 0. 322 0.01 '322 323 1224 15.2 1.6 3.3 -62.22 -3.) 023 023 1322 0.21 0.02 0.01 1225 17.2 1.2 0.9-6223 .5 0.37 023 0.02 322 022 0.01 i221T2l3 046137 3137 't08 0020.01 302001 123717.2 1.3 3.3 -6L61.-26.72 U? 8.08 U? 1322 022 8.91
Results of al.ysis for Span Lengths: 40, 50, 60 and 65 m
SP.
NATtJBALLH FtThCY
(1.st.
rtde)
Cm) (Ha)NWSPN
STIC
DLECrION (in) damage calculation: 0JT Ui 0.3 4.1E -13.2 -.8intp .. 0.0 l.OE+0002 -11.8 45.1)intp 0.4 L'i302 -11.5 -260 intp 0.3 1.E0l? -11.5 -26.3 irttp 0.2 ?.'1 -11.5 -26.0 intp 0.4 1.gE431 -11.4 263int U 7.13Ei31S11 -11.4 -.0icltpL DAGE
FOR 1 YEAR N VE + CURR.FL.
1 61x103..oio
6.49x102 1.36x101.53
58
,J U S 5 0 2.43x1O5 5.95x103 13 COB ES ThGEATIJE
LIFE CAILIA WI A LflVIIT DAfr RATIOOF 0.1 IN-LINE CPCSS-FLO1 62 .5 1.5 4120 0.74 17 O.Y- KC 1221 I.-0E U.CE+XO 13.2 1.4 12 4.1E4))35 3.0E'3 15.2 1.5 1 4.E- 9.UE' 15.2 1.6 1224 46E-l3 15.2 1.5 iZ 1.-Th 00C4O) 17.2 1.3 i 2eO35 ojcj 17.2 1.3 1227 1E- 0CE+LL)3 17.2 1.2 0,.10 0.21 0.40 0.53