TECHNISCHEWIIVERSITEIT Laboratorium voor
00pshydromeChaflI
Archief MekeiWeg 2, 2628CD Deift
T&.:O15.7868fl015'81
WEGEMT School
Trondheixn, 1991 January 21 - 25
slender structures
basic
equations and some special effects
Marinus van HÖ15t
Professor Mechanical Engineering
Delf t University of Technology
p= -pgZ
where
p pressure
p density of the liquid
g gravitati9nal acceleration Z coordinate of the point P
A small area dA through the point P experienCes a force
di=pndA
where:
d' force vector on dA
n = inward normal vector on dA
It is assumed that the inner surfaçe of the area is not exposed. to the liquid. The magnitude of F is indepefldent of the orièntatiOn of dA.
(2.1)
(2.2)
WEGEMT slender structures jan'91
TR
i
i Summary
This section deals with an introduction to the basic equations that describe the static and dynamic behaviour of submerged slender
prismatic bodies In a more practical sense this means the equations for marine risers and submarine pipelines during thé lay operation.
2. HydrostatiCs
2.. 1 Basics
In principle hydrostatics is very simple and does not offer unsolved problems The hydrostatic pressure in a homogeneous body of
liquid in the earth's gravitational field depends on the vertical distance from the point under consideration to the liquid surface only.
Take an XYZ coordinate system (Fig
2.1)
in space. The XY plane liesin the liquid surface and the Z-axis points upwards. In a submerged point P(O,O,-Z) the presure is
WÊGEMT slender structures jan9i
TR
2With the. help of the integral proposition of the veçtor calculus it can be proved that
ffpffc1i=fffVP
dV
(2.5)2.3 Buoyancy force distribution
The Archimedian law is a proposition on the integral of the
hydrostatic pressure It does not say anything about the distribution of the forces The distribution of the hydrostatic forces have to be obtained front expression
(22)
For example see Fig 2 2 The solid vertical cylinder with a cross
sectional area A, is partly submerged over a length i The weight of the cylinder is smaller than the weight of the displaced water
Vertical equilibrium is obtained by a support at the top of the cylinder. The buoyancy force applies at the bottom of the çylinder
and is not axially distributed as the weight The pressure distribution
iS
shown in Fig 2 2a and the resulting internalaxial force distribution (due to hydrostatics only) is shown in Fig 2 2b Fig 2 2c presents the internal axial force distribution
including the effect of the weight of the cylinder an4 the support reaCtiQfl.
2.2
Archlmediafl lawA body submerged in a liquid buoyancy force B that
equals
ff=
This
relatiot is derived wit1i the BB.
B2ï=ffpffdA
the = experiences weight o o pgV principlean upward force, the of the displacèd liquid,
stated n 2.1.
r:
(2.3)
or:
WEGEMT slender structures jan'91
3 Buckling due tò hydrostatic forces
Assume a long, slender, prismatic, steel rod suspended from the earth's surface in a deep drilled hole filled with drilling mud See Fig 3 1 The hydrostatic force is a compression force at
the lower ènd of the rod. The normal force distribution is linear inÇ:
N)=-p0gA1+p2gAC
(3.1)
where:
p0 density of the liquid in the hole
p5 density of steel
A cross sectional area of the rod
1 submerged length of the rod
ç vertical distance from the lower end of the rod
Ç is solved for N(Ç) O from:
p8gA(p0gA1
p0gAl
poi
p3gA p2Example
p0 20Ó0 kg/rn2, heavy drilling mud
p5 7800 kg/ma, steel 1 1000 m
then
2000 l000
264m7800
-So 264 ni of the rod is under compression. Note that the cross
sectional area has disappeared from the expression The cross
sectional area can be very small, for example 1 cm2 - iO rn2 The hydrostatic force on the lower end of the rod is then F
= p0
g A i =2000 9 81 l0 1000 1962 N, working on a rod with a cross section of 1 cm2, 264 in length under compression This is a situation that
buckling of the rod is expected However, buckling will not occur This is understood by considering the potential energy.
The earth's surface is the zero potential energy level. The potential energy of the ród is:
WEGEMT slènder structures jan'91
TR
4The potential energy of the displaced liquid i
E=O.5gAl2
-The total potential energy is then:
=E+E0=o.5gA1Z(pPo)
For
p5> Po is E > O. In a possible buckled state the combined centreof mass of both the rod and the liquid shall rise with respect
to the straight situation This means the potential energy of of
the system in the buckled state Eh > E But buckling is only possible when
the potential energy in the system decreases Conclusion no buckling will occur.
Buckling for the described system is possible only when P5 < Po' in words when the density of the rod material is
smaller than the density of the surrounding liquid.
(3.5)
TR
WEGEMT slendér structures jan'91 4 Derivati9fl of basic relations
To derive the formulas for a (steel) riser first the equations for a solid bar instead c'ff for a pipe are derived A circular cross section with constant diameter is assumed The coordinate system x,y is a member coordinate system In the undeforifled state the x-axis coincides with the bar axis and is parallel to the global Z-axis of section 2 The analysis will be restricted
to the two dimenSional case.
The forces acting on a small section with length ds of the bar
are shown in Fig 4 1 In the unloaded and undeformed situation the axis of the bar coincides with the x-axiS The independent variable is the coordinate s measured along the (deformed) bar axis.
e angle between member axis and and positive
x-axis
T = axial tensile force V - shear
M
bending momentb hydrostatic force per unit length perpendicular to
the. mémber axis
-q all other forces perpendicular to the member axis per
unit length
w weight of the bar = Ap5g
A cross sectional area of the bär
= density of the mäterial of the bar g acceleration of gravity
radius of curvature f the deformed member axis,with
i
r
dB ds
(4.1)
The magnitude of the hydrostatic force b will be derived in section
5 For the time being it is sufficient to assume
that the direction
of b is as indicated as i Fig 4 1 The magnitude of b is not so obvious as will be shown later.
Constder in Fig 4.1 the equilibtitm o the forces ix. a direction parallel t the vectors b d and q ds:
_cos(4O)+(V+d) cos(-.dD)+
-Tsìn(-dB) -(T+d')
sin(--dB) + -w sinOds-qd$+bdSQ
WEGEMT slender strUctures jan'91 'IR M- (M+dt'Í) + Vrsin(*dO)
+(V+dV)rsirk(--d6) +
Tr(ì-cqS(-dO))-(T+dT) r (1-c6s(-c))=O
worked
Out: =0 now are:dTr(1_cos(4-O)) and dT/rsin(4.c)
small of higher order änd are neglected, further
sin(c)--dO
then is:(4,7)
(4,8)
(4,9)
(4,10)
worked out:dVcos(-.a)
(4, 3)
withcos (4c)
=1 sin(*c) =-.c(4,4)
we obtain:dv- TaO-wdssinO-qds+bdS = D
(4 ¡5)
or:f=2
+wslriO+q-b(4,6)
equilibrium of moments with respect to the point Q (see Fig 4.2) yields:
_dM+2Vr-4d8=O or:
d4=VrdO
and because:zdO=ds
d
WEGEMT slender structures jan'91
TR
7 Equilibriurn of forces in the direction perpendicular to the direction of the vectors b ds and q ds (sée Fig 4.1) yields:-Tcos(--dO)
+(T+dT)äoS(-0)
-+Vsin(-o) +(V+dV)
sin(4) -wcoaßds'O
worked out:
dTcos(-.a)
Now is
dVsin(4.c)
small of higher order and is neglected. Further is
and
this yields:
dT+VdO-wdscÔSO=O
or:
Thê x and y coordinates are related with 9 and s by:
.=cosO and
..=3inO
ds
ds
The. deformation añd the bendig moment are related through:
1_M
r El
according to the Eui.erian beam theory.. Because off:
also
yields:
For small rotàticns s -' x.
The derivatives with respect to s are replaced by derivatives with
respect to x.
d!=_t'. +wcosO
ds
ds
(4,12) (4,13) (4,14) (4,15) (4,16) (4,17) (4,18) (4,20)WEGEMT slender structures jan'91
TR
8And further is:
for the bending moment:
dM
y
dx
and for the
tension:
sinOO=-a11d
---(
d
dc
r
d. dic(4,23)
The equation for the shear then yields:or:
(4,22)
d2y
(4,21)
¿k;
P4,7. f4 = - V --w or:
(4,24)
+ w dicEI
dk
EI
dM- y
dx dxEI
dT
-dx
EI
dx
dBM
d. EI
dMdx
dV_
-b
dx
q
wO + q - bWGEMT. slender structures jan'91
TR
9Wé rewrite all derived formulas agath in such a way that all
differential quotients appear once in the left hand term For
comparison the "nçrmal" Euler beam equations are also presented for a transverse distributed load q - b.
Beam, transverse load q - b, axi tensiofl large rotations: = sinO da
c. M
da EI (4,25).._rTf1+wsin8+_b
ds EIBeam, transverse load q - b,, axial tension, smal rotations:
(4,26)
Beam, transverse load q - b, no axial tension, "normal" e4uations
and the assumed direction in Fig 4.1 appears to be correct.
For small values of the curvature yields:
=Ap0gsiflO
See also [1].10
WECEMT slender structures jan'91
5 The hydtostÁtiC forçe on a beam element
The element in Fig 5..l is bounded by the shell surface (originally prismatic) the upper cross section the lower cross section The shell
surface (the wet surface) is exposed to hydrostatic pressure The surfaces forming the cross sections are isolated from the hydrostatic pressure by the parts of the neighbouring sections of the member The
resulting hydrostatic force is therefore not equal to volume times density of the sea water times acceleration of gravity. (Archimedes) To calculate tite resultant of the hydrostatic pressure on the
wetted surface, a surface integral as in chapter 2 has to be evaluated This raises great practical difficulties en can be avoided The hydrostatic force on the member section ds in Fig
5 1 has been drawn in an arbitrary direction and has the unk±tòwn magnitide
(5,1)
Imagine that the hydrostatic pressuré also acts on the surfaces of the cross section The forces on those surfaces can be
writtèn down immediately
Ap0g(d-*' -4
dgcos8)+PA
(5,2)i-;i =Ap2g(dX- -
dscosO}+P8A
p5 is the atmospheric pressure at the water surface. The vector b ds is completely arbitrary The components are b ds and b1
ds.
The vector VB is the buoyäncy force, working vertical upwards under the condition that also the cross sectional surfaces are also exposed to the hydrostatic pressure;
=Apgd5
The vector equation for the equilibrium of the forces is
Now solve the magnitude and the direction of b from these euations.
The dérivation is presented in the appendix. The result is:
=A
(p0g(siflO- (d-x)
(5,5)
(5,6)
(5,3)
-Z=T ±1W
sinO+q
ds
0EI
Ñow consider the differential equation for T in (4.25) and
substitute (6.2) and (6.3). Then we obtain:
--
(T0-Ap-g(d-X)).
= +(w0+Ap0g) cosOWith
or:
dT
dx
M_1+Ap0g-.--
_V_Eii
+w9côsO4-Ap0gCO36
= cosO
ds
equation (6.5) becomes:dT
M.4
=-V-1
wcosO
(6,4) (6,5) (6,6) (6,7)WEGEMT slender structures
jan'91
TR
11
6 Efféctive teúsion
and
effect:ive weightSubstitute (5 ..) in 4 25) and neglect the infiuence of the atmospheric pressure, this results in:
f=T.
48jo+q_Ap0gsino+Apog(d) --
(6,1)Or rearranged:
We now call:
T0=Tog(d
(6,2)the effective tension and
w0 =
w-Ap0g
(6,3)
the effective weight.
The effective tension in a bar is. the real tensioh (that is the tension that causes axial stress and strain that can be measured by strain gauges) in the bar increased by the weight of a column liquid with density p0, with the same cross section as the bar and a height equal to the vertical distance of the point under consideration to the water surface The effective weight per unit length of the
bar is its mass per unit length multiplied by the acceleration of gravity
(this is the weight "in air") dtirtished by the weight of the
displäced water per unit length.
Using effective tension and effective weight equation (6 1)
WEGEMT slender. structures jan'91
TR
12The set of equations (4.25) now
becoêS:
= sinO. ds
cfM
d.. EI dÑ . T0 .. +w,sinO + q - - V - + w .cos OAnd in a simi),ar way the set (4.26) becomes:
dx
dx
EI dMy
V=Tefi +w,0+q
!
+w dicEI
(6,8)
(6,9)
obviously the efféctiv weight and the effettive tension control the bending behaviour and therefore the stability öf the submerged bar Using these quantities explicitly in the mathematical description
considerably simplifies the equations
Consider again the bar suspended in the deep hole of section 3
Calculating the effective tension in the bar with expression (6 2) it appears that the effective tension vanishes at the lower end and increases when x increases The effective tension is positive in the entire bar and therefore no buckling will occur.
WECEMT slender structures jan'91
TR
137 Pipes
The behaviour of the solid bar submerged in water has little practical application (though the equations (4 25) and (6 8) also
apply to steél wire ançhor cables etc.). The subject of this course is however risers and pipelines The influence of the surroundig water on the outside of a pipe is
of
course completely the same asthe influence on a solid bar The influence of the liquid inside the pipe can be derived in a similar way as for outside and will not be repeated here The convenience of the use of effective weight and
effective tension now becomes more clear The equations (6 8) and (6.9) remain valid, only the expéssions for w0 and T0 have to be adj usted:
= T+A0p0g(d0-X)
-A1pg(d1-X)
(7,1)
We = w-A0 p0g+Apg
with:
A0 - outside cross sectional, area of the pipe
A inside cross sectional area of the pipe
Po density of the liquid outside the pipe
Pi density of the liquid inside the pipe
d0 - X = hydrostatic head of the liquid outside the pipe d - x hydrostatic head
of
the 1iquid insidé the pipeThe idea of the effective weight and effective tetision are explained in a somewhat. different but clear way in [2].
8 The effect of internal, pressure on the bending stiffness
Ïs it likely that the bending stiffness and thus the buckling
behaviour is affècted by the internai pressure in the pipe? This
question in various formulations has often been the subject of discussion between people working on risers or similar slender pipe structuìeS.
Consider equations (6 9) and (6 10) The internal pressure does not explicitly appear in these equations. One of the controling
parameters for the shape of the riser is T (7.1) in which the hydrostatic head inside the pipe appears Now hydrostatic head is nothing else than pressure therefore
one
might conclude that whenoverpressure is applied at the top of the riser this pressure has to be treated as an increase of the hydrostatic head resulting in a
decrease of the effective tension and increasing the danger of
buckling (Although people are apt to think intuitively that internal pressure increases the bending stiffness) This however is not the
case Because to apply overpressure inside the riser, the riser has
to be provided with closed ends. The internal pressure' results in a force on the closed end and increases the real tension in the same amount as the increase of the hydrostatic head and the effective
tension remains unchanged For a riser, where for simplicity the outside hydrostatic head has been omitted, this has been illustrated in Fig 8.1. See [2] for an exhaustive discussion.
WEGEMT slender structúres jan'91
TR
14
9 The eq,iatioflS
for pipelInes
There is
no principal
dffèrence
in theequations
for risers and for pipelines But it is rather natural for risers to
select the pipe bound x-axis vertical in the uxideformed situation, while for pipeline problems it is more convenient to take the pipe bound x-axis
horizontal in the undeformed situation This means that the equations derived in the previous sections are riser equations rather than pipeline equations The pipeline equations are presented
below with reference to Fig 9.]..
.=sinO
ds
a'8M
ds
EI
(9,1)
ds
w osO
!±=_v.!L+w.sinO
10 Riser dynamicsThis course thainly deals with hard piped risers. The
defôrmati0,
displacement and rotation is usually small for this type of risers therefore equations (6.9) w.11 be used rather than (6.8).
The dynamic effect, and this is utLderstoó4 to be the
bending
vibration, is introduced as the inertia force
lateral to the to the pipé
axis.
This force is a compÌlént of theload q in (6.9); d2
q-
m-
dt2
where m includes the mass of
the
teei pipe and everything attached to it, the mass of the liquid in the pipe and theadded mass This results in:
¿ri =
A6)g
+ A11 f
Çwith:
m mass pet unit length
A8 cross sectional area of steel
- cross sectional area defined by the inner diameter A0 cross sectional area defined by the outer diameter
p5 = density of pipe material (steel) p density of fluid
inside the riser
p0 density of sea water
coefficient of adde mass
The differential equation for
the
shear ï(6.9) then becomes: M o
-
+ We +q
-'7T
(10,1)
(10., 3)
(10,2)
op0WEGEMT slender structures jan'91
TR
1511 Special effects due to fluid flow 11.1 IntroductiOn
Up till now no attentton has been paid to fluid motion inside the pipe or in other words it was assumed that the fluid flow
velocity
inside the pipe vas zero Obviously this is usually not true for risers in operation The effect of fluid flow on riser
behaviour
iS
-not very great and can often be neglected n practical cases. Inliterature the effects to be discussed are treated with the use of the momentum theorem This is an excellent tool to obtain
overall effects no matter how complicated the flow pattern is
In our case we are interested in the detailed distribution of the forces and this will be obtained by considering the fluid particles.
11.. 2 Flow through a curved pipe
Fig 11.1 shows an element of a riser. A fluid particle that flows through this pipe element experiences a centripeta1 acceleration
v2/r. Where y is the velocity of the particle (that is assumed to be equal to the average fluid velocity with respect to the
pipe) and
r is the local bendig radius To obtain this centripeta], acceleration the pipe wall excerts a centripeta]' force on the fluid particle and consequently the pipe wall experiences an opposite and equal force, a centrifugal force When it is assumed that all particles in a cross section f011ow the same path thé centrifugal force q per unit length on the pipe wall is
qf
=-A1pj--Compare the sign convention with Fig 4.1.. 11.3 Cor-jolis effect
Consider the straight vertical pipe in Fig 11 2
The pipe oscillates around a hinge at the lower end and has an ins tntaneouS
anEular
velocity dO/dt. The fluid velocity inside the pipe is directed
upwards and has the magnitude y. A body of fluid that covers the internal cross section of the pipe and has a length dx experiences a coriolis acceleration .a - 2 dO/dt y in the positive y direction. The inertia force or the force on the pipe wall is q per unit
length.
= -:2VAPj
(11,2)The cor-jolis force is opposed to the direction of the velocity due to the oscillation around the hinge It has a positive damping effect
When the fluid velocity is reversed, from top to bottom,
the cortolis force has the same direction as the oscillating velocity and has a
negative damping effect As stated before this effect is
usually small.
Literature
1. P. Tendrup Pedersen, Equilibrium of offshore
cables and pipelines diring laying
International Shipbuilding Progress, vol 22, Dec
1975, no 256
2 C P Sparks, The influence of tension, pressure
and weight on pipe and riser deformations and stresses.
Transactions of the ASME, Journal of Energy Resources
TechnolOgY, vol
106, March 1984:
3 B G Burke, Analysis of marine risers
for deep water OTC 1771, 1971.
WEGEMT slender structures jan'9l
a. hydrostatic pressure
on a verticäl
sub-merged cylinder
Fig 2.1: Hydrostatic pressure on
an area dA.
a
*
dF = pidA
ifth p=-pgZ
N=pgaA
Fig. 2.2: Hydrostatic load on a
vertical partially submerged c4inder
dN:__
dZ
TR17
b. resulting
c. resulting
axial internal
axial internal
force N due
force. N due
to hydrostatic
to hydrostatics
L=1000 m
H
guide//
I
F=p*A p=pisg*LE::
earth surface
hole, filled with
drilling mud
/j
steel bar
cross sectiOnal area A=1 O .m=264m
Fig. 3i: Bar suspended ¡n. a
druid hole filled with
drilling mud.
TR18
distribution
of normal
force N
tensiofl pressionFig. 4.1: Forces on o
small, deformed section
òf a submerged bar.
Lever arms of V with
respect to Q
Lever arms of T with respect
o Q
Fig. 4.2: Moment lever arms of Shear and tension.
(auxiliary diagram of Fig. 4.1)
Fig. 5.1:
Hydrostatic forces on .0 small, deformed section
of a submerged. bar.
r
X
A
/1/
'ft
T
=
(2)Te (x) =
cose 1: open riser
= F - w(d - x)
Te'°(X)
=
T°(x) Ap1g(d-
)
cose .2:
closed riser, overpressûre at the top
F + w(d x) + pIA.
T(2)()
-'
A1p1g(d x) - pA1
T
(x) +p A1pg(d1 x) - p1A
T°(x) A1pg(d x) = T1(x)
Fig. 8.1:
Internal overpressure and effective tension.
TR22
T.
does not change by application
V
y
:z:
Fig. 9.1:
Suitable coordinates for pfpe laying.
TR23
d
T
V
Fig. 11.1: Centrifugal load on O curved pipe.
V2/r
dF= dm
2=ftp1 rde
dF= dF
2-= ---Ap rd9-!_
1R24
2 Vr
= A1p1
dF.
ra
qf=
coriolis acceleration: a = 2.
y
dt
dF (on the flùid). = a dm
.= A1p dx o
dF (on the pipewaH) = - Ajpdx a
dF
dO
q=
= - Apa= 2 Apv
Fig. 11.2:
Coriolis. load on oscillating pipe.
Qdl.C.L4&
;-TR25
5trrcL
£(4
1s>çJ
cr
APP- i
Appendix
From section
5:
+ + ds = V
For the components in x-direction:
I cos (8 - 1/2dO) - cos (0 + 1/2d8) + bds =
For the components in y-direction:
sin (8 - 1/2do) - sin (0 + 1/2dO) + bds = O
Elaboration of thé equation in x-direction yields:
Ap0g {d - x + 1/2 ds cos 8} cos (0 - 1/2d9) + pA cos (8 - 1/2dO)
- Ap0g (d - x 1/2 ds cosO) cos (8 + 1/2do) + bds
- p5A cos (8 + 1/2dO) = Ap0g ds
Ap0g [(d -
x)
{cös (8 - 1/2dO) - cos (8 + 1/2d8)} +1/2 ds cos 8{cos (8 - 1/2do) + cos (O + 1/2d8)}] + bds +
p5A sinOd8 = Ap0g ds
Now is:
cos (8 - 1/2do) - cos (8 + 1/2do) = -2 sinO sin(-1/2dO)
= 2 sinO sin 1/2do = sinO dO
and
cos (8 - 1/2dO) + cos (e + 1/2dO) = 2 cc:O cos (-1/2do)
= 2 cosO cos 1/2do = 2 cosO
So:
Ap0g[(d-x)sin8 d8 + 1/2 ds cosO 2 cosO]
+ ds Ap,o .g ds
-Ap0 g[(d-x) sinO dB + ds cos2B] + Ap0g ds sinO dO b ds
bds
= Ap0g[ds-ds cos28 - (d-x) smBde]
-p5.A sinO dObds
= Ap0g[S1fl20 ds
- (d-x)
sinO dO] sinO dOdO dG
=
Ap0g sinG [sinO -
(d-x)ã---] -pA sin8
For the components 1n y-direction:
Ap0g {(d-x) + 1/2ds cosO}
sin
(8-1/2do) +pA sin(8-dO)
-Ap0g{(d-x) 1/2ds cos0} sin (0+1/2do) +
bds
-PA
sin(9d8)=O
Ap0g[(d-x) {sin(8-I/2d8) - sin (8+1/2d8} +1/2ds cosO {sth(8-1/2d8) + sin(0+1/2d8)}] +
bds
= ONow is:
sin(8-1/2d0) sin(8+i/2d0) .2 cosO sin(-1/2d8)
= -
cosO dO
and
siñ(8-1/2d8)
+sin(8+1/2d0)
2 sinO cos(-1/2d8)2 sinO
So:
Ap0g[-(d-x)cos8 d8+ ds cosO.sinel
bds -
PA cos0d8
= O ds = Ap0 g[(d-x)cose d8 - ds ccsO sinO]-
d8
= Ap0 g[(d-x) cose - cose sinO] b y b y t
an
dO .b = apòg cose[(d-x)--- - sine] = Ap0g cose[sine
-cose 1
bx
sin8tanO
APP- 2
d8
bids = -Ap0g cose [ds sin8 - (d-x) d] + pA cbsOd8
= -Ap0 g cose [sinO - (d-x) - J + pA cose
The magnitude ofb is:
II=
b2 + b2)
The direction of is:
= cose
b - sinO
te
X
i.e perpendicular to the tangent of the bended beam
element, as
alreay assumed in section 4.
In addition yields:
sinO
= -ICI
cose
-de . de
Ibi
= A0(p0g sine- a ) - A.(p.g sinO