Вы находитесь на странице: 1из 6

On the application of similitude to installation operations of offshore steel jackets

SUBROTO KUMAR BHATTACHARYYA


Ocean Engineering Centre, Indian Institute of Technology, Madras 600 036, India
The role of similitude in scaled simulation of major installation operations of offshore steel jackets, namely, loadout, launching and upending has been studied. The physical modelling can be looked upon both as an adjunct to numerical modelling using computers and also as an independent tool of investigation. The problems of design of models and of experiments are discussed and the prediction equations based on similitude are given.

INTRODUCTION

The installation process of offshore steel jackets involves three major sequential operations, namely, loadout, launching and upending. Together they represent a problem in the area of rigid body hydrostatics and hydrodynamics. The engineering objective is two-fold. Firstly, the appreciation of the problem in its entirety is desired. Secondly the ability to predict the effects that will result from a change that is imposed on the system must be established with good confidence. In other words, the transfer functions of the system are required with respect to different specific causalities. The system is described completely in terms of its geometry, material properties, initial conditions and boundary conditions. The perturbations effecting the system (say the jacket) are due to interaction with either the associated system (such as the launch barge)or the fields or both. The justification to employ the technique of model analysis based on similitude stems from the very nature of this problem. Firstly, due to the sheer magnitude and scale of operations involved in the problem render it impossible to make direct observations of its behaviour and thus translating them to the ability to predict. Secondly, the numerical modelling techniques based on a mathematical description of the problem have inherent limitation due to the associated assumptions and approximations resulting from the complexity, vastness and nonlinear nature of the problem. Clearly physical modelling offers an excellent 'second opinion' under the circumstances. An excellent account of the scope of model mechanics is given by Hossdorf. 1
THE OPERATIONS AND MAJOR PARAMETERS

Next, the major parameters of interest in each individual operation both from the point of view of the model analyst and the designer will be identified. For the designer, the characteristics of the parameters of interest will form a basis of answering fundamental design questions. In loadout, barge ballasts, winch pull, fender reactions, mooring line forces, rocker arm reactions are of design interest subject to depth of water front, tide range and rate, coefficients of friction between launch truss of the jacket with ground, rocket arm and launch girders. Together, they constitute the problem under study. In launching, rocker arm reactions, rocker arm rotations, barge surge, t~ajectory and depth of penetration of the jacket, effect of initial condition as defined by trim and draft of the barge and effect of jacket entry velocity on these parameters are of design interest. The field parameters of this operation include viscous, intertial and gravitational forces; static and dynamic pressure forces; flexural forces and force due to surface tension. In upending, the crane hook load, hook elevation, jacket ballast, jacket orientation, righting moment and bearing load when the jacket settles down to seabed are important design queries. Static pressure generating buoyancy force and jacket inertial force constitute the field parameters. The water depth at site represents the boundary condition in both launching and upending from the safety point of view. Clearly enough, the whole problem can only be assessed completely in terms of a large number of parameters interrelated by many causality relationships.

PRINCIPLES OF SIMILITUDE

The jacket as a system forms the focal point of study in the installation operations. In load-out, the launch barge and the jacket are coupled together forming the jacket-barge system. In launching, this jacket-barge system functions together until the instant when the jacket leaves the barge. In upending, the derrick barge-jacket system forms the theme of study. An account of these operations is presented by Graft. 2
Accepted February 1984. Discussion closes December 1984.

Through similitude relationships one must formulate proper system equations and prediction equations in order to project the model test results to prototype scale. The general procedure for this can be described as follows.3 Let
7r, = f~lr2, ~3 . . . . .

7r.)

(1)

be the characteristic equation of the phenomenon in which lr~ is the dimensionless group containing the prediction quantity and 7r2, rr3,..., zrn are-dimensionless groups having significant bearing on the phenomenon. In general f is unknown.

0141-1187/841040221-06 $2.00 1984 CML Publications

Applied Ocean Research, 1984, Vol. 6, No. 4

221

Application o f similitude to installation operations o f offshore steel ]ackets: S. K. Bhattacha~.va

Corresponding to (1), the modelled system will have a characteristic equation of the same form which is
7rim = f m ( l r 2 m , *ram . . . . . rgnrn)
(2)

respectively to

F;
W
-

-u

(15)

where the suoscript m has reference to the model. If the prototype and the model phenomena are identical,
f=fm

Fl
--

(3)

= 0

(16)

which then results into the system equations


1r i = trim, i = 2, 3 . . . . . n

(4)

(3) and (4) obviously results in

But experimentally, a more realistic simulation of environment will give a first hand knowledge of the nature of variation of these forces. For. barge ballast and rocker arm hinge reactions, the characteristic equations are
fi

rr,

= 7r,,.

(5)

which can be used to arrive at the prototype prediction values from the model values. This is called the prediction equation of the system. Due to practical limitations resulting from scale effects it may happen that
rrir n ~ rri

w, :z'ir,
a

(17)

where
o~ = f ( X )

(18)

(6)
X=X l'

is chosen. In other words,


7rim = erri

(7)

where
[3 = f ( X )

This is so called distortion modelling which results in rewriting (5) as


7/"I "~" ~/7r I m (8)

(20)

Thus for prediction purposes one needs to establish 3,-e relationship, which can be done in a number of ways. 3
SIMILITUDE IN LOAD-OUT A N D UPENDING

Here c~ and 3 are linear functions of the position of the jacket with respect to the barge, which is defined by the jacket travel X. In upending, one of the simplified characteristics equations can be written as Ht W f \ W ' B' (21)

These two operations involve hydrostatic modelling and therefore are considered together. Geometrical similarity is assumed. In load-out the situation is simple; Fw, Ff, F m depend directly on ta as these forces result from system balance against frictional forces. In simplified similitude symbolism,

Others can be written on similar lines. Main points of interest based on equations (9)-(21 ) are recorded below. (1) Only the rigid body parameters of the systems are of interest. (2) The groups a, fl, h/l indicate geometrical similarity requirement. (3) In load-out only I has to be similar. (4) In upending the group W/B signifies that both W and B must be similar. (5) In load-out barge ballast, F b, need to be similar. (6) In upending, jacket ballasts, Fi, need to be similar. Further it can be concluded that

Fw
-

f(ta)

(9)

Ff = f(/a) W

( 1O)

6 --=
W

f(u)

(i l)

which are the characteristic equations of the system. Also


F f = Y Ffi,
i

i = 1,2 (usually)

(12)

Ft= ~Fti,
i

i = 1 to 4 (usually)

(13)

where i represents the ith fender or the ith mooring line. In writing these equations one should recognise the existences of a more rigorous representation, like say,

(1) Jacket inertia should be similar. (2) Tide, in load-out, should follow geometrical similarity. (3) Water depth and hook elevation should follow geometrical similarity in upending. (4) Jacket orientation in upending should be identical for both the prototype and the model. (5) Disposition of the hook slings in upending should be identical for both the prototype and the model. (6) Sling loads in upending should be similar. (7) Jacket centre of gravity and centre of buoyancy should follow geometrical similarity.
DESIGN CRITERIA OF MODELS

Fw -~ = f (/a, -'~", "-'~)

(14)

and so on. However, simple representations are preferable when causalities are not intractable without mathematical manipulations. In ideal static case (9), (10)and (11)reduce

The dependence of a successful model study on proper design and fabrication of the models cannot be overemphasised. The design criteria must be based on appropriate similitude. From the formulation of similitude in the last section as applicable to the present problem, one can

222

A p p l i e d Ocean Research, 1984, Vol. 6, No. 4

Application of similitude

to installation

operations of of/shore steel jackets: S. K. Bhattacharyya

directly list down the major design criteria. For the barge these are as follows: (1) Geometrical similarity as dictated by Xt must be maintained for the hull. (2) As the model barge must support ballasting (and deballasting) the internal volumes of the tanks must obey X~t = X}Xp (22)

BUt

x~ Combining (30) and (31) ?'d 2om -- i~ = kp~.~ (02p-- @Iv)

(31)

(32)

(3) As the buoyancy dependent force on the model barge depend on the underwater displacement and density of the medium in which it is floating, for buoyancy similarity it must have

For both weight and buoyancy similarity together, put (29) into (32), which gives 1 I : " Xa 11/: (33)

i,n=-~t lop---~p('2op--~p)j

(4) For draft similarity, which is XT = Xt the model barge weight (let,) must satisfy (24)

In light of (33), one has to choose a model material, i.e. Xa such that ~ or,
_

Xd (34) Xp

Xwb = xsb = x] Xp

(25)
(ip/+op) =/> 1 - - -

and also the distribution of let, should be such that (24) is satisfied under similar conditions. (5) Tank boundaries of the barge must be located at geometrically similar positions in the model. (6) Major structural fittings like launch beams, rocker arms etc. in the model should also be dictated by Xz. (7) Other less important functional fittings like boUard eyes, structural members, rocker arm hinges, deck openings, winch assembly etc. may be designed only from practical considerations of convenience. Now coming to the jacket, which is the focus of this study, it is seen that the modelling requirement in load-out and upending are not the same. Though geometrical similarity is required in both the operations, in load-out only weight similarity will suffice, while in upending both weight and buoyancy similarity must be simultaneously satisfied. The latter will be considered as it automatically satisfies the former. Overall geometrical similarity of both the jacket and the barge must be given by the same Xt. For the jacket, clearly, Xw = XB = X~Xp Xcc = ~ B = ~'t
~k I ----Xf Xp

M
Putting ip = Cop -- 2tp into (34) it follows that,

~)tp
Cop\
i-

Cop

l kp <--4 Xa

(35)

This inequality may not be satisfied for all tubular sizes for all ;Mr. For example, for Xp = 1.025 (usually), if model material is perspex (dm = 1.2), then (35) gives -Z ~> 24.39

tp

(36)

(26a) (26b)
(27)

taking only the practical inequality. Equation (36) may not be satisfied by all sizes of tubulars. However, fortunately for this useful model material, this inequality is satisfied by most jacket tubular sizes. So, it can be stated that given a model material, (29) and (33) will dictate the model tubular sizes subject to (35). Now considering (33), it may at times be impracticable to have a Oim as suggested by it even when (29) is satisfied. To get a better leverage on the choice of ~ira let us consider that the tubular hollow space is filled up by a liquid of specific gravity d~n. In such a case,
2 t,.d,. = X~ Xp (2o,. - ~ ) l . d , . + ~m

(~op -

~p) lz,dp

(37)

The jacket structure is largely made of tubulars. Ideally (26) should be satisfied for each individual member. This requirement will now be looked into in detail. For a tubular, one has, for buoyancy similarity, by (26) XaXtXp = X~Xp as a = lr/4,1 ~'o = k, For weight similarity, again by (26), one has (29) (28)

On simplification, this gives

'['
Two sets of conditions are possible: 1. (a) \
Using ~)ip = ~)op --

+i.0,,+}]%
dr,, > d,n
dp

(39)
--

~pl

Xp

i>

(40)

XaXtkd = XaIXp
or. (30)

2tp in (40) and simplifying, tp( ") 1Xp


-

1--tp

<.---

(41)

(~op

dC)Op

kd

AoDlied Ocean Research. 1984

Vol_ 6 N n

2~3

Application of similitude to installation operations of offshore steel jackets: S. K. Bhattacha~_ .va


So (39) and (41) forms the first set. 2. (a) am > am
2
or.

(42)

~ i = Xt (P~n/Pm )1/2
!

(48)

(b)

t
tp ~1
I

dp

dp
- - ~< 0 (43)

which on similar simplification gives


-----I/>

Clearly, suitable balance between Oim and Pra is possible. In general, P~n > Pm will provide a practical solution. For a plate in the jacket structure, condition for buoyancy similarity is Xt = Xt (49)

tp

1 Xp

Cop "

Cop"

4 Xa

(44)

and that for weight similarity is

So (42) and (44) form the second set. Clearly between these two sets all tubulars can be t modelled by all materials subject to correct choice of din. This way the theoretical difficulty posed by (35) may be overcome. From (38) it can be seen that a suitable balance between elm and d m is possible, subject to Oim < Oom.

xt = x~xp/xa

(50)

Example 1
Say Xt = 50, d m = 1.2, pp = 1.025, Pm= 1, Oop = 1384 mm, t p = 4 4 m m . So, (~orn= 2 7 . 6 8 m m and (~op/tp = 31.44. So (35) is satisfied. By (33)

Oim
Say, one requires

----

12.86 mm

~l,n = 20 mm Using this value in (38) d~n = 0.704 Such a liquid is possible.

Using a steel plate in the model is clearly the best course because this gives the ratio of Xp and Xa quite close to unity. Once the jacket model is designed based on such a procedure, the CG and CB will be geometrically similar and (26), (27) will be automatically satisfied. Now a good look is needed to find if there is any 'less than ideal' method to model the jacket in order to get reasonable results from the model tests. Ooviously, the fundamental equations as given by (26) and (27) must be satisfied. Depending upon the material chosen and sizes available, one can choose either (47) or (48) in order to arrive at a suitable Xt. It remains to be decided on a scheme of member design with relation to the overall geometry such that firstly (26) and (27) are satisfied and secondly, the model jacket may be considered as a 'close' copy o f the prototype which can be expected to produce reasonable data leading to reasonable prediction of prototype parameters of interest. A possible scheme is developed below step by step. (1) Overall geometric similarity given by Xt to be maintained. (2) For any member, length must be as suggested by Xt but (29) and (30) need not be satisfied and therefore (33) and (34) and subsequent equations also do not come into the picture. (3) However, deviation from (29) should not be too great and the members which cannot be modelled even by the smallest tube sizes available (i.e. deviation from (29) is large) may be omitted altogether. (4) The jacket is made of discrete members and in its geometry, the levels (at different elevations) and the faces are clearly discrenable as discrete but major contributors to weight and buoyancy (depending upon the mode of floatation). Therefore, if weight and buoyancy of the individual levels and faces satisfy (26) and (27) and so does the jacket in overall terms, then it can be assumed that the model represents the prototype closely enough. The scheme described above requires iterative process to be carried out for each level and each face until a satisfactory configuration is reached. It is likely that in order to maintain (26) and (27) at each level and face, certain additional members may be included in the form of flat plates, tubes or solid rods, and certain members may be omitted due to their smallness in model scale. It is advisable to include additional members only at the levels and not at the faces. It must be realised that even after proper matching of levels and faces, there may be some discrepancy in overall weight, buoyancy, inertia, CG and CB. This can only be

Example 2
Other things being the same as in example 1, the tubular size is given by o p = 6 6 0 m m , t p = 3 2 m m . So (~o,n= 13.2 mm and (~op/tp = 20.625. So (35) is not satisfied. Adopt the condition set given by (42) and (44). In this case (44) is satisfied. So d~n > dm must be used. From (38)

~,.=
=

36.66

1.2 - d ; ,
r

Select ~im 7.2 mm (say) which gives dm = 1.907. Such a liquid is possible. It can be readily seen that, for a particular member, if Xa = XP - 2 -_ _ (45)

~op e~ip
2

then the member needs to be a solid rod of circular cross section. For upending one further requires

XFi = X~ Xp

(46)

for ballast legs. Ballast CG also must be geometrically similar. Now if liquid used in the model for ballast has specific gravity Pm then, Xeo = Xei = Xl (47)

This is a condition which may be difficult to satisfy in practice. Alternatively if a liquid of different specific gravity (P~n) is used for model ballast, one has

X6i =
224

~klPp/Pr

a =

X~Xp

Applied Ocean Research, 1984, Vol. 6, No. 4

Application of similitude to installation operations of offshore steel jackets: S. K. Bhattacharyya


removed by lumping of suitable weights at some convenient locations. The objective, clearly, should be to use minimum number of such lumped weights. One general requirement of such an iterative procedure is that for a member (most of them, in fact), requirements of which cannot be met simultaneously. Relative importance of Froude and Reynolds forces can be adjusted (through selection of jacket velocity), a well known technique in many problems, so that launch modelling becomes possible. Launch phenomenon is essentially a slow speed one and thus may be expected to result in laminar flow condition, the regime in which Reynolds force cannot be ignored. But fortunately, by the very nature of jacket geometry, turbulent flow is generated. This of course can be made doubly sure by accepted methods of turbulence simulation as done in ship models. In such a case, the effect of viscosity can be considered to have only a very modest effect on trajectory, the flow being above critical value of Reynolds number. The resulting trajectory can be considered accurate enough for all practical applications. Based on these considerations one can write, Tx(r), T y ( r ) = f ( F ) subject to Clearly one must have, in light of (53) 2 ~
.i~2

W/a < 1

(51)

so that lumping of weight to achieve (26) and (27) for the jacket in overall sense can be effective. Equation (51) can be written as

x. < X~Xp/Xa
in light of (28) to (30). In the present approach let

(52)
(53)

Corn = 6op
Then (52) reduces to

elm <

622op

--

4tp(op -- tp) )t~)tO Xa

(54)

(57)

Rm > Rcr

(58)

It is seen that (54) and (55) can easily be satisfied in practice because of the very nature of the inequalities. SIMILITUDE IN LAUNCHING Modelling of jacket trajectory is of fundamental importance in launching. Firstly the trajectory must be geometrically similar and secondly the attitude of the jacket along the trajectory in both the prototype and the model must be identical. It is well known that any dynamic modelling of motion of a body in fluid depend upon a number of force fields. These are intertial, viscous, gravi. tational, surface tension, flexural, static pressure and dynamic pressure forces and forces of compressibility. It is known that the phenomenon cannot be modelled for dynamic similarity of all these forces. The relative importance of these forces will be briefly reviewed here. In general one can write,

The 3'-e relationship as pointed out in (7) and (8) can only be found by long term correlation with prototype measurements. This is probably not necessary from application point of view. DESIGN OF EXPERIMENTS Simulation of load.out, launching and upending must be in accordance with principles of similitude. This means that apart from the models, the other imposed experimental parameters should also be guided by similitude. In load-out one must have
~tX = )tT =- )tt ~- )tl

(59) (60)

/~m = gp In launching one must have Xv = XY2

(61) (62)

~im = ~ip
In upending one must have

Tx, Ty = f ( F , R , o,S,M,)tt)

(56)
)th e ~" )th = )kl

(63)

based on dimensional analysis. As there is a virtual lack of importance of surface tension forces on the shape of ocean surface, S, the Weber number can be eliminated from (56). This problem is not in the domain of hydroelasticity and therefore the flexure number need not be considered. Considering the velocity range of the jacket (order of 5 m s-1) it can be said that forces of compressibility will not become predominant due to near absence of gravitational forces. Thus, M, the Mach number does not come into play in modelling. Because of the shape of the entry face of the jacket, there is no trace of an encircling cavity formed on impact (which grows in size subsequently) as happens in the case of a hydroballistic missile, a The force equilibrium of cavity, therefore need not be simulated in launch modelling without causing significant error in trajectory prediction. Both viscous and gravitational forces are compared with inertial force because inertia is always important in the velocity range to be considered. These comparisons are through R (Reynolds number) and F (Froude number),

PREDICTION EQUATIONS Once the basis of modelling for load-out, launching and upending has been established and the models are made in accordance to the laid down design criteria, one can write down all possible prediction equations in the form of (5). Some of the more important ones are given below. Others can be written on similar lines. Some prediction equations of load-out are,
)tF w = )tFfi = )tFli = )tFr = )tFb = )t~ )tp

(64)

Some prediction equations of launching are, )trx(r) = )try(r) = )tt (65) (66) (67) (68) (69)

~r(Tx) = )tr (Ty)= XY2


)tv (r) = )t}/2 )tb (r) ,= 1

Xo(r ) = 1 = )t~ (r)

Applied Ocean Research. 1984 Vol 6 No d

~9~

Application o f similitude to installation operations o f offshore steel/ackets: S. K. Bhattacha~ya X$ (7") = A}"1/2 Xg(r) = X71 Some prediction equations for upending are XHl = XFj = X~Xp XI- = 1 (72) (73) (70) (71) F Fb Ff Ft F/ Fr Fw h Ht he 1 1 m M p R Rcr S T Ty Tx t tvt v /J W Wb X x y 7 e Xk Froude number barge ballast fender reaction mooring line forces jacket ballast rocker arm reaction winch pull water depth hook load hook elevation moment of inertia of jacket characteristic length subscript denoting model Mach number subscript denoting prototype Reynolds number critical Reynolds number Weber number draft depth of jacket (trajectory) below the water surface horizontal distance of jacket (trajectory) thickness of tubular tide internal volume of tanks velocity of jacket acceleration or deceleration of jacket weight of jacket weight of barge jacket travel trajectory axis pointing vertically downwards horizontal trajectory axis prediction factor distortion factor ratio of parameter 'k' of prototype to model coefficient of friction, static and dynamic p specific gravity of the floating medium p' specific gravity of the material filling model tubular o cavitation number r time go outer diameter of tubular ~i inner diamer of tubular generalised angle of the barge in launching ~i initial barge trim angle in launching generalised angle representing jacket orientation in upending 0 generalised angle in jacket trajectory ~) generalised angular velocity generalised angular acceleration

CONCLUDING REMARKS Use of scaled models in simulation of installation operations based on principles of similitude can be considered both as an adjunct to numerical modelling and as an independent tool of investigation. The role of similitude and the associated problems have been discussed. It can be used with advantage for the special problems like optimising the upending operations, effect of initial conditions on trajectory in launching etc. in very short time as compared to that required in computer simulation. The performance of numerical modelling can also be improved in light of such a model study. The most attractive feature of any such study for the practising engineer will probably be the understanding gained by visual observation of the operations. It is quite likely that the visual nature of observation will lead to appreciation of some finer points of the operations which is not possible otherwise.

REFERENCES 1 Hossdorf, H. Model Analysis of Structures, Van Nostrand Reinhold Co., 1974 2 Graph, W. J. Introduction to Offshore Structures, Gulf Publishing Co., Houston, Texas, 1981 3 Sedov, L. I. Similarity and Dimensional Methods in Mechanics, MIR Publishers, Moscow, 1982

NOMENCLATURE ti a B Ba CB CG d f cross sectional area of tubular lot calculating buoyancy cross sectional area of tubular for calculating weight buoyancy of jacket buoyancy of barge centre of buoyancy centre of gravity specific gravity of model material a function

226

Applied Ocean Research, 1984, Vol. 6, No. 4

Вам также может понравиться