Proceedings of the Fourth (1994) International Offshore and Polnr Engineering Conference
Osaka, Japan, Apnll0-15, 1994
Copynght 1994 by The International Society of Offshore and Polnr Engineers ISBN 1-880653-10-9 (Set); ISBN 1-880653-11-7 (Vol. I) Effects of Different Wave Free Surface Approximations on the Response of a TLP in Deep Water B.B. Mekha, C.P. Johnson and J.M. Roesset University of Texas at Austin Austin, Texas, USA ABSTRACT This paper is based on work conducted at the Offshore Tech- nology Research Center to Investigate the relative importance of different types of nonlInearities on the dynamic response of tensIOn leg platforms. A tIme domam solution USIng Monson's equation with several modIficatIOns and allowing to mclude or neglect various sources of nonlineantles was implemented and used in the dynamIC analysis. This paper focuses on the non- lmear effect of evaluating the wave forces up to the free surface using dIfferent approximate methods. A TLP hull model with time varymg tendon forces IS subjected to regular and irregular waves with and without current. The effects on calculating the wave kinematics up to the mean water level or up to the actual free surface, makIng use of'vanous extrapolatIOn or stretchIng techniques are then discussed. INTRODUCTION As oil exploratIOn proceeds into deeper water, the tension leg platfonn is one Viable and promising structural solution. The characteristics of the TLPs are quite challenging and mterest- Ing at the same time. Despite the considerable amount of work which has already been done to understand their behavior, there is still a great deal more research which is required. A TLP can be modeled as a six degree of freedom system. The natural fre- quencies in surge, sway and yaw are well below the range of wave frequencies. On the other hand, the heave, pitch and roll natural frequencies are above the exciting wave frequencies. There are thus two dIfferent nonlinear aspects which need to be considered m the design of TLPs, one related to the amount of offset per- mitted, the other to the limits (z.e. minimum and maxJmum) of the tendon tension forces. The fonner is mainly influenced by the slowly varying drift forces (low frequency response) while the latter IS greatly affected by the springing and ringIng phenomena associated with high frequency response. 105 Various sources of nonlinearitles are present in the analYSIS of a TLP. One of them is the evaluation of the wave kinematics up to the wave free surface (LlghthIll, 1986) Because lmear wave theory IS only valId up to the mean water level, a number of extrapolatIOn and stretchmg methods have been suggested and Wldely used In the prediction of the wave kinematics above or below the mean water level. The emphasiS on using such approx- ImatIOns can probably be attributed to the small differences in the calculation of the wave kinematics in deep water between lin- ear wave theory and Stokes' higher order wave theories. Wheeler In 1969 and ChakrabartI in 1971 suggested two different stretch- Ing methods to calculate the wave kInematIcs up to the wave free surface. Other researchers have extrapolated the results of the lmear wave theory USIng different approaches. Three of these ap- proaches, whIch are Implemented in thiS work, are: 1) hyperbolIc extrapolation, 2) unifonn extrapolation, and 3) linear extrapo- lation using Taylor's series expansion. In 1986, Gudmestad and Connor derived a nonlinear approXImatIOn which was in good agreement With experimental test results for regular waves. ThiS method was extended for JITegular waves (Gudmestad, 1990) and applied to structures where the drag forces dominate (Gudmes- tad et al., 1990). The effect of this approximation and the pre- viously mentIOned ones on inertia dominated structures such as TLPs has not been yet extenSively studied This paper focuses on the effects on the TLP response of the different methods of evaluation of the wave kinematics up to the wave free surface. A TLP model is chosen and subjected to reg- ular and irregular waves with and without current. Linear and nonlinear approximations are applIed to the linear wave theory In order to predict the wave kinematics up to the free surface. The results are discussed and compared with those obtained by Stokes' second order wave theory. The effects of the different approximations on the low and high frequency responses are in- vestigated. EQUATIONS OF MOTION The two - dimensional TLP model used for the study consists of four columns connected together by four pontoons (see Figure 1). The dynamic equilibrium equations are MU+CU+KU=P (1) where M, C and K = mass, damping and stiffness matrices, re- spectively. U, U, U = displacement, velocity and acceleration vectors, respectively. P = load vector. The TLP hull is assumed to be a rigId body with three de- grees of freedom (t.e. surge, heave and pitch) at the hull center of gravity (GGh ). The tendons are modeled as massless elastic springs which provide lateral and axial stIffness at their points of connection to the hull. Mass Matrix The mass matrix of the model COnsIsts of two parts: the struc- tural mass matrix and the added mass matnx (2) The structural mass matrix of the hull at the GGh is given by [ MOO 1 M(st) = 0 M 0 o 0 I (3) where No Np M = LMe. + LMp> >=1 s=1 Nc Np 1= L(Je. + + L(Jps + >=1 .=1 Me and Mp = mass of one column and one pontoon, re- spectively. Ie and Ip = mass moment of inertia of one column and one pontoon, respectively (Note: Ip = 0 for out-of-plane pontoon). r, = the distance between the center of gravity of a column (GGe) or pontoon (GGp ) and that of the complete hull (GGh ). Ne and Np = the total number of columns and pontoons, respectively. To maintain a constant mass matrix, the added mass matrix for the hull is calculated accordmg to the hull initial submerged depth up to the mean water level. Therefore, any change in the added mass forces due to the wave passage and structural mo- tions is applied to the force vector as corrective terms (Johnson et al., 1993). 106 The added mass matrix at the hull center of gravity is then No Np M(a) = + (4) s=1 .=1 where (5) IS the added mass matrix for a column at its center of gravity. = 1np 1 o 0 !z. 12 (6) IS the in-plane added mass matrix of a pontoon and [ 1 0 0 1 = 1np 0 1 0 000 (7) IS the out-of-plane added mass matrix of a pontoon. T [H -i; 1 (8) is the transformation matrix. Deck f 1 A Pontoon Pontoon A a) Front VIew POIItOOD b) Plan vIew - Section A-A Figure 1: TLP hull geometry. 1l'Ip = CnPwAplp .{3=!_r. 2 I. Ac and AI' = cross sectional area of column and pontoon, respectively. lo = the initIal submerged depth of a column. II' = the length of a pontoon. pw = the water mass density. Cn = the added mass coefficient. Y = the distance between the CCc and the bottom of the column. X, = x(corp). - Xh Z. = z(corp). - Zh Xh and Zh = the horizontal and vertical coordmates of the CCh . XC! and z'" = the honzontal and vertical coordInates of the center of graVIty of coiumn z. x". and z". = the horizontal and vertical coordmates of the center of gravity of pontoon z. Stiffness Matrix The hull stiffness is proVided by the tendons attached to the columns. The stIffness matrix IS calculated according to the mI- tial pretensIOrung forces in the tendons where .k -fu. x - L, No Kb = :L T, ,=1 [
K t = o 0] kz 0 o 0 and k Z - L, (9) (10) At, Et and Lt = area, modulus of elasticity and length of each group of tendons. Fto = the irutial pretensioning force of one group of tendons. T = the transformation matrix given in equation 8 with X, = Xh -Xh and Z, = Zh - Zh Xh and Zt. = the horizontal and vertical coordinates of the connection point of tendon group z to the structure. It should be noted that when the variation of the tendon forces is taken into account throughout the analysis, the dif- ference between the instantaneous and the Initial tendon forces is applied to the force vector. The hydrostatic restoring forces which could be also considered as stiffness terms are totally taken as part of the force vector. 107 Damping Matrix Two sources of damping can be present in offshore structures: structural damping and hydrodynamic damping which involves both radiation and viscous dissipatIOn of energy. The structural and the radiation damping are small for all frequencies and hard to estimate. Nevertheless, a small amount of the radiation damp- ing is included as part of the radiation forces on the bottom of the columns of the TLP hull. The main source of damping then is the viscous part of the hydrodynamIC dampmg which is be- ing considered in this work and implemented as part of the drag term in Morison's equatIOn The structural damping matnx m thIs case is merely c=o (11) Force Vector The hydrodynamic forces on the cylmders are computed using Morison's equation 7rD2 7rD2. pwCM- 4 -andl- PwCa-4-Undl . '"--v--' added m.a.ss 1 .. + ,2"PwCD Dlvn - Unl(Vn - Un)dl, (12) drag II CM = Cn + 1 = inertIa coeffiCIent II CD = drag coeffiCIent II D = dIameter of cylmder ., Vn, an = normal velOCIty (wave and current) and accelera- tion of the fiwd, respectively ., Un, Un = normal velOCIty and acceleratIOn of cylinder, re- spectIvely. The mstantaneous submerged parts of the cylinders are dI- vided mto equal length segments and both the wave and the cylinder kinematics are calculated at the centrOid of each seg- ment. The inertia and drag forces m EquatIOn 12 (excluding the added mass force) are computed for all segments and then numerically integrated to find the resultant forces and moment The added mass term is integrated along the actual submerged depth of the columns and pontoons in the hull deformed config- uratIOn to compute the resultant added mass forces. However, only the difference between these forces and the eqwvalent forces included in the added mass matrix is applied to the force vector. In addition, the hydrostatic and hydrodynamic pressures are integrated and conSIdered m the forcmg functIOns. The hydro- statIC forces are computed up to the mean water level. The hy- drodynamic forces whIch account for the hydrodynamic pressure on the bottom of the columns due to the wave elevatIOn are taken from results obtaIned with linear diffraction theory (Weggel and Roesset, 1994). The radiatIOn forces asSOCIated with the diffrac- tion forces are also considered m the form of damping and added mass terms. These terms are lmear:ly related to the velocity and the acceleration of the structure as was shown by Chakrabarti and Harma (1991). In case of time varying tendon forces, the difference between the tendon forces at any time and the Irutial forces based on the mean water level is also added to the force vector. LINEAR WAVE THEORY The simplest and most commonly used theory is the Small Am- plitude wave theory (Linear wave theory). The theory is based on the assumptions that the ratios of wave height to wave length, H / L, and wave height to water depth, H / d, are very small (Sarp- kaya and Isaacson, 1981). The velocity potential for linear wave theory in the x - z plane is given by gH coshk(z + d) . cp(x,z,t) = 2w coshkd SIn(kx-wt + cp) (13) and the wave profile by H 'Tl(x, t) = "2 cos (kx - wt + cp) (14) where d is the water depth, 9 is the gravitational acceleration, k = 2; is the wave number, w = ~ is the wave circular frequency, L, T and cp are the wave length, penod and phase, respec- tIVely, and w 2 = gk tanh kd whIch is the dispersion relation. (15) Similar to the velocity potential, the expressIOns for the dis- placement, velocity and acceleration components of a water par- ticle conSIst of three main tenns: the amplitude at the surface, independent of water depth, the vanatlOn with depth given by an attenuation In the fonn of a hyperbolic function, and a factor dependent on time and position of the particle. However, for deep water, whIch IS of interest in this work, a modified fonn of the equations is obtained by simplifymg the attenuation tenn (Sarpkaya and Isaacson, 1981). For irregular waves, the sea state represented by an energy spectral density is simulated by a com- binatIOn of many regular waves with random phases as Nw H 'Tl(x, t) = L -;f cos (k,x - w,t + cp,) ,=1 (16) where Nw is the number of waves. The heIght of each wave component is determined from the power spectral density at the corresponding frequency (Chakrabarti, 1987). Linear Free Surface Approximations To compute the hydrodynamIC forces, the wave kinematics have to be known at all positions of the structure. LInear wave theory, however, can only evaluate the wave kinematics up to the mean water level. To overcome this limitation, several extrapolatIOn methods have been suggested to evaluate the wave kinematics between the mean water level and the wave free surface. Some of the methods are: 108 It Hyperbolic Extrapolation: This method, suggested by Hogben (1974), simply extends the wave kinematics above the mean water level in a hyperbolic way. This extrapo- lation has been proved to overestimate the forces on the structure and to be very conservative (Chakrabarti, 1987). o LillleaJr Extrapolation: This method, used by Nwogu and Irani (1990) , computes the wave kinematics beyond the mean water level by expanding the expression in a Taylor's series and neglecting second or higher order tenns. As an example, the horizontal particle velocity can be expanded as 8ul u(x, z, t) ~ ulz=o + z 8z z=o (17) tI Stretching Methods: These methods shift the wave kine- matics profile from the mean water level to the free surface. Thus, the wave kinematics decay exponentially between the wave surface and the sea bed. Two methods have been sug- gested. - Wheeler's Approximation:It was recommended by Wheeler in 1969. It replaces the tenn (z + d) in the numerator of the hyperbolic function by (z + d) d ~ f / For example, the horizontal velOCIty up to the free surface becomes gkH cosh k(z + ) d ~ f / u(x,z,t)= 2w coshkd cos(kx-wt+cp) (18) - Clhakrabarti's Approximation: It was suggested by Chakrabarti In 1971 and it replaces the tenn kd in the denominator of the hyperbolic expressions by ked + 'Tl). Thus for example, the horizontal particle velocity becomes gkH coshk(z + d) u(x,z,t)=- hk(d ) cos(kx-wt+cp) 2w cos +'Tl (19) This approximatIOn satisfies the dynamic boundary condition by modifying the pressure equation. Uniform Extrapolation: This method, used by Eatock Taylor et al. (1992), assumes that the wave kinematICS in the crest region above the mean water are equal to their corresponding values at the mean water level. u(x, z, t) ~ ulz=o (20) In this work, all the above approximations have been used and their results compared to investigate the effect of the dif- ferent approximations. The implementation of Wheeler's and Chakrabarti's approximations in conjunction with deep water can readily be obtained by replacing e kz In deep water wave kinematic equations by ek(z-f/). For instance, equatIon 18 or 19 becomes u(x, z, t) = ~ w ek(z-f/) cos (kx - wt + cp) (21) Nonlinear Free surface Approximation Gudmestad and Connor (1986) utilized the free surface Wheeler's and Chakrabarti's stretching approximations by expanding them to the second order. The expansions do not ensure the local continuity enforced by Laplace's equation; however, they satisfy an averaged continuity condition to the order of the expansion. The nonlinear dynamic and kinematic boundary conditions are also satisfied at the free surface to the same order. The com- plete second order formulatIOns of Wheeler's and Chakrabarti's approXlmatIOn for any water depth are presented in detail in Gudmestad and Connor (1986) and are not repeated here. In deep water, however, Chakrabarti's second order has shown a better agreement with laboratory test results where the velOCIty under the wave trough is hIgher than the velocity under the wave crest Thus, in addition to the linear approXlmations mentioned earlIer, this approximatIOn is also adopted for use in this work. The velocity potential is <I> (x, z, t) The free surface profile IS H TJ = 2" cos (kx - wt + '1') (23) The dIspersIOn relatIOn is w 2 = gk (24) The hOrIzontal velocity of a water partIcle IS u(x, z, t) = Hw T ek(z-'1) cos(kx - wt + '1') + (T) (L) {_ek(z-r/) + _ e4k(Z-'7)} 2 cos 2(kx - wt + <p) (25) The vertIcal velOCIty of a water particle is w(x, z, t) Hw 2 ek(z-'7) sin(kx-wt+<p) + (T)(L) {_.!..ek(Z-'7) + _ 2e4k(Z-'7)} 2 2 sin2(kx - wt + <p) (26) The horizontal acceleratIOn of a water partIcle is Hw 2
__ e k (z-'7) sin (kx - wt + <p) + (--)(-) 2 P L u(x,z, t) {_2ek (z-'7) + 3e2k(z-'7) _ 2e4k(Z-'7)} sin2(kx - wt + <p) (27) The vertical acceleration of a water particle is w(x, z, t) Hw 2
--e k (z-'7) cos (kx - wt + '1') - (--)(-) 2 P L {_ek(z-'7) + 3e2k(z-'7) _ 4e4k(Z-'7)} cos2(kx-wt+<p) (28) 109 It can be observed that the higher order terms of this ap- proXlmation do not tend to zero in deep water. This nonlinear approximation was extended by Gudmestad (1990) to the case of irregular waves. SECOND-ORDER WAVE THEORY Applying the deep water simplifications to the second order ve- locities and acceleratIOns leads to the same equatIOns as for the Linear Wave in the case of deep water. The only exceptions are the wave profile which is gIven In deep water by H TJ = "2 cos (kx - wt + <p) + 4L cos2(kx - wt + <p) (29) and the dynamic pressure pgH 1 p = __ e kz cos (kx - wt + <p) + - __ e 2kz (30) 2 4 L ANALYSIS OF THE TLP MODEL Before starting the dynamic analysis, a static analYSIS is per- formed In two stages. The first stage considers the effect of buoyancy whIle the second one accounts for the current. The Imtial tensiomng forces In the tendons are computed due to the buoyancy exceeding the weIght of the structure. The buoyancy is calculated up to the mean water level. The tendons undergo an elongatIOn due to the tensIon force. This elongation is referred to as vertIcal static offset and accordingly, the structure initIal configuration is adjusted vertically by the amount of thIs vertIcal offset The second stage IS to analyze the structure statIcally due to the presence of current. The following equilibrium equation KUstattc = p (31) IS solved to obtain the initial dIsplacement vector for the three degrees of freedom. The stiffness matrix, K, given in equatIOn 9 IS evaluated upon the determination of the Initial tendon forces. The force vector, on the other hand, includes the static force due to current. It is computed usmg the drag term of Morison's equatIOn WIth no wave and structure velocities. Thus, the force on segIllent of length dl IS given by (32) where Vcr is the current normal velocity. The initial displacement vector, U stahc, is applied as initial condition for the dynamic analYSIS. The equation of motIOn given In Equation 1 is solved in the tIme domain. The constant average acceleration method was chosen for the step by step integration of the equations of motion. APPLICATIONS AND RESULTS A TLP model in 5000 feet water depth was selected to inves- tigate the effect of the different wave kinematic approximations when the forces are calculated up to the wave free surface. The data of the representative TLP is shown in Figure 2. The TLP model was subjected to regular and irregular waves. The regular wave is represented by a design wave with 100 feet height and a period of 16.95 second. A JONSWAP spectrum with significant wave height of 52 feet is used to simulate irregular waves. This significant wave height gives a peak density at a wave period of about 16.90 second which is almost the same as that of the reg- ular wave. A uniform current of 3 ftlsec along the hull is also applied m some cases. It IS worth noting again that the forces are calculated in the instantaneous pOSItion of the model at each time step. Values of CM = 2.0 and CD = 1.0 are used in the Morison's equation. For all analyses, the vertIcal static offset was 5.70 ft and that reduces the depth of the initial draft of the columns from 137 ft to 131.30 ft. At this new hull configuration, the initIal tension force of each group of tendons (2 e. 3 tendons) is 22520 kips. 285' + Weight of one cybnder = 290041o.ps WeIght of one pontoon = 3583 kIps Numbet of columns = 4 Number of pontoons = 4 + Tendon's data D1a=3167f1.Area=1534sqfl E = 4176000 Iaps/sq fI Number of tendoDS I cyhnder = 3 240' FIgure 2: TLP model - dImensions and properties. Regular Wave The TLP model was first analyzed by computing the forces up to the mean water level. The results of this analysis are then used as a basis to evaluate the effect of computing the wave kine- matics up to the free surface. Table 1 summanzes the results obtained usmg the different approximations. Inspection of these results show clearly that the approximations can be divided into two groups, one including the hyperbolic, uniform and linear ex- trapolatIOns, the other consisting of the stretching and nonlinear approximations. The amplitudes of the surge motion are very close in all cases with differences of less than 5%. This indicates that the surge amplitude is just a function of the wave height and is not affected for practical purposes by the variations in the free surface elevation. The mean drift on the other hand is extremely sensitive to the approximation used. For the first group of ap- proximations, the mean drift is nearly double of that obtained computing the forces up to the mean water level while the results for the second group are in fact slightly smaller than those of the first analysIs (mean water level). For the heave motion, both the amplItude and the offset are affected by the approXImatIOn. 110 Results using the first group of approximations show amplitudes of motIOn which are 10 to 15% higher than those computed in- tegrating to the mean water level while the amplitudes obtained with the second group of approximations are very similar to the mean water level solution. The offsets are again nearly the same using the mean water level or the second group of approXImations but decreases about 10 to 15% when using the first group. This may be attributed to the effect of the setdown of the TLP due to the large drift caused by the first group of approximations. Figure 3 shows the tIme history of the surge response using the hyperbolic and the nonlinear approximatIons as representative of the two groups. Surge (ft) Heave (ft) Ampl. Mean Drift Amp/. Mean offset MWL 38.90 38.10 0.92 5.52 Hyper. extr 39.30 69.10 1.04 5.18 Unif. extr. 39.10 63.50 1.02 5.26 Lnr. extr. 39.70 75.00 1.07 5.10 Str. app. 38.60 3020 0.92 5.68 Nonlnr. app. 37.80 30.50 0.94 5.67 Table 1: Results of the surge and heave motions usmg dIfferent approXImations. Applying Stokes' second order theory, which IS valId up to the first order wave profile, does not change the results sigruf- lcantly. As shown m Table 2, they are very close to those of the linear wave theory WIth hyperbolIc extrapolatIOn. This IS actually expected smce the second order theory approaches the lmear theory in deep water Usmg the stretching approximation WIth the second order theory writing the attenuation factor as e k (Z-'11), the results are very SImIlar to those oflmear theory WIth stretchmg. Surge (ft) Heave (ft) Ampl Mean Dnft Amp/. Mean offset Lmear-MWL 38.90 38.10 0.92 5.52 Lmear-Hyp 39.30 69.10 1.04 5.18 Linear-Str. 3860 30.20 0.92 5.68 Second 38.80 6930 1.03 5.18 Second-Str. 38.00 30.00 0.92 5.68 Table 2: Results of the surge and heave motIOns using different wave theories. Table 3 shows the results of analyses performed with and WIthout current. While the current has no effect on the surge amplitude, it significantly influences the mean drift. In the heave motion, the current mcreases the amplitude and reduces the mean offset by about 30%. As before, the reduction in the mean heave offset is related to the setdown of the modeL FIgure 4 shows the corresponding time history of the heave motion. I I Surge (ft) Heave (ft) I I Amp/. Mean Drift Amp/. Mean offset I I Str. w 10 cur. I 38.60 30.20 0.92 5.68 I I Str. wi cur. I 38.60 117.00 1.30 4.35 I Table 3: Results of the surge and heave motions with and WIthout current. Figures 5 and 6 show the spectral density of the surge and heave responses for selected analyses. The surge spectral density clearly shows the response of the model at the wave frequency and at difference frequency (low frequency). A very small peak at twice the wave frequency can also be seen. The surge spectral densIties appear to be the same for all analyses except for the amount of their energy in the low frequency range where the first group of approximations has larger amplitude consistent with their larger drift. The heave spectral density shows responses at the wave frequency as well as at the second and thud harmon- ics (high frequencies) which in thIS case are important because they become closer to the natural frequency of the model. A peak appears also at the natural frequency of the model which is about 4.50 times the wave frequency. This peak is attributed to the ringing phenomenon which is a transIent response at the structure natural frequency. The second group of approximations gave slightly higher peaks at the high frequencies. Figures 7 and 8 show the pitch response and its corresponding spectral density for the nonlinear approximation. The spectral density of the pitch response fot the MWL analysis IS also shown in Figure 8. The nonlinear effect of calculating the forces up to the wave free surface on the hIgh frequency response of the TLP can easily be seen in this figure. Besides the peak at the wave frequency, peaks appear at the second, third, fourth and fifth harmonics. The response at the pItch natural frequency of the model at about 4.25 tImes the wave frequency IS also clearly appear. The second group of apprmomations produces the hIgh- est peaks at the hIgher harmonics while the first group have the hIghest peaks at the wave frequency. 150 100 E. ; 50 e 0 '" c:.:
-50 .. = "-' -100 -150 0 v. { ... --Hyperbohc extrapolatIOn _. --.. -. Nonlmear approXllDauoa 100 200 300 400 500 Time (sec) 600 Figure 3: Time history of the surge response for regular wave. __ ____ -L ____ 4- 7 g 6 ........... --Stretcbmg app w/o CUJreDt 'Co" -------. Stretcbmg app w/current
o 100 200 300 400 500 600 Time (sec) Figure 4: Time history of the heave response for regular wave. 111 12000 10000 8000 C' 6000 r;; 4000 2000 Nonhnear approXllDabon ......... f",= 0059 Hz 3 f/f .. 4 5 6 Figure 5: Spectral density of the surge motion for regular wave. 300 250 200 l Stretchmg approXlIllalloD I l;... =0059 Hz - 150 r;; 100 50 0 \....A \. I /'.. 0 2 3 4 5 6 flf." Figure 6. Spectral density of the heave motIon for regular wave. 0004 0.003 ;:;;- 0.002 .. -=- ; 0.001 Q 0 ... :l
-0.001 ....
-0.002 it -0.003 -0.004 0 Time (sec) Figure 7: Time history of the pitch response for regular wave. __ ____ -L ____ 4- 0.35 03 025 S 0.2 "-' 0.15 0.1 0.05 2 --Nonbnear approXlIIl3t:1on - - - .. - - - Mean water level 3 f/f." fw= 0059 Hz 4 5 6 Figure 8: Spectral density of the pitch motion for regular wave Tables 4, 5 and 6, which correspond to Tables 1, 2 and :3 respectively, summarize the maximum and minimum, upstream and downstream forces for a group of tendons. The time history response of the tendon forces is shown in Figure 9 using the non- linear approximation and the corresponding spectral density is shown in Figure 10. The tendon force responses using the other approximations have essentially the same trend. However, the second group of approXImations gave slightly larger maximum tension forces for both upstream and downstream tendons. This may be explained by the increased effect of the high frequency response of the pitch motion in the second group of approxima tions. The mirumum tension forces are almost the same for all cases and they are not thus greatly affected by the approxima- tions. The current increased all tension values by about 2 to 3%. The spectral density of the tendon forces shows similarity with that of the heave spectral density, but exhibits other peaks at illgher frequencies and at the pitch natural frequency due to the effects of the pitch response_ Upstream (bps) Downstream (k:tps) Maxtm. Mzmm. Maxim Mzmm. MWL 26031 18753 25636 19039 Hyper. extr. 26278 18835 25661 19098 Umf. extr. 26301 18846 25621 19015 Lnr extr. 26301 18868 25680 19135 Str. app. 26903 19303 26249 19137 Nonlnr app. 27296 19020 26608 18654 Table 4: Results of the tendon forces per group usmg different apprmamations
o 100 200 300 400 500 600 Time (sec) Figure 9: Time history of the tendon forces for regular wave. 1.2 106+-----.\-----l-------I------+-----I-----+ 8.0 105 .j Nonhnear approlUlDaoon 0059 Hz S 6.0 "" 4.0 105
0.0 o 2 3 fir... 4 5 6 Figure 10: Spectral density of the tendon forces for regular wave. 112 [ Upstream (k:tps) Downstream (k:tps) Maxtm Mmzm Maxtm. Mmzm. Linear-MWL 26031 18753 25636 19039 Linear-Hyp. 26278 18835 25661 19098 lLinear-Str 26903 19303 26249 19137 Second 26469 18767 25777 19129 Second-Str. 26624 19418 26169 19240 Table 5: Results of the tendon forces per group using different wave theories.
["tr. wi 0 current wi current Table 6: Results of the tendon forces per group with and without current. Irregular Waves Based on the results discussed in the previous section, only two approximations, the hyperbolic extrapolation and the nonlin- ear approximation, were used with Irregular waves. The re- sults are compared with those of calculating the wave forces up to the mean water level. The model was subjected to Irregu- lar waves corresponding to JONSWAP spectrum With significant wave height of 52 feet. The time history of the irregular waves is shown m Figure 11. Figure 12 shows the surge response us- ing both approXImations and the MWL analysis while the cor- responding spectral densities for both approximations are shown m Figure 13. All analyses exillbit slowly varymg drift motions at period of about 100 second; however, the amplitudes of such mo- tions are different and the highest is associated With the hyper- boltc extrapolatIOn. Comparison of the spectral denSities With both approXImations and the MWL solutIOn (not shown here) shows that the peaks at the excitmg wave frequencies are nearly the same for all analyses (only sltghtly higher usmg the hyper- bolic extrapolation). The peaks at the low frequencIes are sub- stantuJly illgher usmg the hyperbolic extrapolation willIe those of the other two analyses are almost the same. Therefore, as for the regular wave, the dJfferent approximations have httle ef- fect on the response amplitude at the first order wave frequencies but affect mostly the low frequency content of the motion (slowly varymg dnft response) at the difference frequencies. 80 60 g 40 ., 20 0
... 0
-20 .. ... .. -40
-60 -80 0 100 200 300 400 Time (sec) 500 600 Figure 11: Time illstory of the Simulated lITegular waves. 150 100 g ~ 50 = e ... 0 .. ... =r: .. -50 f!' = "-' -100 ... _ ......... - .1 = : = ~ h C I 1 ...... -Nonhnear . -150 0 100 200 300 400 500 600 TiDle (sec) Figure 12: Time history of the surge response for irregular waves. 8000 7000 6000 5000 --Hyperbohc extrapolation -.. - NonhD.ear approX1Jllatlon S 4000 "-' 3000 2000 1000 005 0.1 0.15 0.2 0.25 03 r (Hz) Figure 13 Spectral denSity of the surge motion for Irregular waves. The time history of the heave and pitch motIOns are not shown here; however, their correspondmg spectral densities are given m Figures 14 and 15. The analyses with dIfferent approx- Imations and With MWL did not affect the heave motIOn. The time histOries and the corresponding spectral denSities are almost the same for all analyses. The peaks at the wave frequencies and the sum frequencies are sirmlar in all analyses. A peak at the heave natural frequency (around 0.26 Hz) of the model is also very clear in Figure 14. The main dIfference is m the higher peaks at the zero and difference frequencies in the hyperbolic ex- trapolation analysis due to the fact that the model undergoes a larger set down because of the large drift motion and mean surge offset. The pitch motion is on the other hand greatly influenced by the approximation used. The mam response of the model is always at its natural frequency; however, thIs response is sig- nificantly amplified using the hyperbolic extrapolation and to a lesser degree usmg the nonlinear approximation. The amplitude of the peak at the pitch natural frequency of the MWL analysis is small compared with those of the other analyses. Figures 16 and 17 show the time history of the forces of a group of tendons using both approximations. The correspond- ing response using the MWL analysis IS similar to the nonlinear approximation with slightly smaller amplitudes. It can be seen from Figure 18, which shows the corresponding spectral densi- tIes, that the rmging response of the tendon forces (as for regular waves) is mainly influenced by the pitch motion of the structure and to a lesser extent by the heave motion (WIth peaks at their 113 natural frequencies). The peaks which occur at the predominant wave frequencies are almost the same for all analyses For the hy- perbolic extrapolation, the peaks at the pitch natural frequency are the dominant ones and they are about twice the peaks at the wave frequencies. The peak at the pitch natural frequency for the nonlinear approximation is of the same order as those at the wave frequencies. The MWL analysis produces smaller response at the structure's natural frequencies. 50 --Hyperbohc extrapolatton --- ... --- Nonhnear approxunabon 40 - 30 ;;; 20 10 0.05 0.1 0.15 0.2 0.25 0.3 r (Hz) Figure 14. Spectral density of the heave motion for irregular waves. 08 06 ... ;;; 04 0.2 0 0 --Hyperbohc extrapolation -- --e--- Nonhnear approxunatJon 0.05 0.1 0.15 r (Hz) 02 0.25 03 Figure 15: Spectral density of the pitch motion for irregular waves. CONCLUSIONS Several conclusions can be drawn from the results for both reg- ular and irregular waves. Some of these are 1. If, as stated by Gudmestad and Connor (1986) the stretch- ing and the nonlinear approximatIOns YIeld results in good agreement with test data, the hyperbolic, uniform and lin- ear extrapolation methods seem to be conservative in eval- uating the wave kinematics above the mean water level. They produce larger values of drift (more than tWice the drift caused by the other two approximatIOns). Evaluating the forces up to the MWL produces acceptable results for preliminary calculations. 2. For irregular waves, as for regular waves the hyperbolIc ex- trapolation produces higher response amplitudes than the other approximatIOns in the low frequency range. ____ __ -J ____ ____ __ ____ '" .... 35000
... ::: r: 25000 g 20000 =
15000 10000 o 100 200 300 400 600 Time (sec) Figure 16: Time history of the tendon forces for irregular waves.
35000 -;;;- . . . . ... -... . .. -.' ..... I Nonlmear approlUlDllbon t .... 30000
.. 25000 e "" 0: e 20000 = ... E- 15000 10000 0 100 200 300 400 500 600 Time (sec) Figure 17: TIme hIstory of the tendon forces for Irregular waves. 5.0 105-t----'---I------L-----L----I-__ + 4.0 105 301eS o .. --HypeIbohc cxtrapolanon . ....... Nonlmear approxunanon 0.05 0.1 0.15 0.2 r (Hz) 0.25 0.3 FIgure 18: Spectral density of the tendon forces for irregular waves. 3. The evaluation of the wave forces up to the free surface does not significantly affect the response in heave while it greatly amplifies the pitch motion at its natural frequency. The pitch response at its natural frequency can be as signif- icant as the response at the wave exciting frequencies for a regular wave but is much more imp6rtant for irregular waves. 4. The effect of the current is significant on the low and high frequency motions of the TLP (mean drift in surge, ampli- tude and mean offset in heave) 114 5. The tendon forces are influenced by the heave and pitch motions of the structure. For regular waves, the second group of approximations induce hIgher tendon forces due to their hIgh frequency pitch responses. For irregular waves, the tendon forces are dOffilnated by the pitch response at its natural frequency. 6. The results of using second order wave theory tend to ap- proach those of lmear wave theory in deep water. There- fore, a suitable approximation should also be used with the second order wave theory in deep water. REFERENCES Chakrabarti, S.K. (1971). "Dynaffilc of Single Point Moonng in Deep Water," (DiSCUSSIOn), Jounralof Waterway, Harbors and Coastal Engineermg Dwiswn, ASCE, 97, August, pp. 588-590. ChakrabartI, S.K. (1987) "Hydrodynamzcs of Offshore Struc- tures," Sprmger-Verlag, New YOlk . ChakrabartI, S.K. and Hanna, S.Y. (1991) "HIgh Frequency Hydrodynamic Dampmg of a TLP Leg," Proceedmgs of the 10th Internatwnal Conference on Offshore Mechamcs and Arctzc Engmeermg, Vol. I-A, Stavanger, Norway, June 23- 28. Eatock Taylor, R., Ramey, RC. and Dru, D.N. (1992). "Non-Lmear HydrodynaIIllc Analysis of TLP's in Extreme Waves' Slender Body and DIffraction Theories Compared," Proceedmgs of the 6th Internatwnal Conference on Behav- wr of Offshore Structures, London, UK. Gudmestad, O.T. (1990). "A New Approach for Estimatmg Ir- regular Deep Warter Wave Kinematics," Journal of Applzed Ocean Research, Vol. 12, No.1, pp. 19-24 Gudmestad, O.T. and Connor, J.J. (1986). "Engjneenng Ap- proximations to Nonlinear Deepwater Waves," Journal of Applzed Ocean Research, Vol. 8, No.2, pp. 76-88. Gudmestad, O.T., SpIdsoe, N. and Karunakaran, D. (1990) "Wave Loading on Dynamic SensitIve Offshore Structures," Proceedzngs of the 9th Internatwnal Conference on Offshore Mechanzcs and Arctic Engtneerzng, Vol. I, Houston, Texas, February. Hogben, N. (1974). "Fluid Loading in Offshore Structures, A State-of-the-Art Appraisal: Wave Loads," Marztzme Tech- nology Monograph No.1, Royal Institute of Naval ArchI- tects, London, England. Johnson, C.P., Mekha, B B. and R.oesset, J.M. (1993). "Sim- plified Modelmg for the Nonlmear Response of TenSIOn Leg Platforms in Deep Water," Proceedmgs of the 3rd in- ternatwnal Offshore and Polar Engmeerzng Conference, ISOPE - 93, Singapore, June 6-1l. Lighthill, J. (1986). "Fundamental Concerning Wave Load- ing on Offshore Structures," Journal of Fluid Mechanics, Vol. 173, pp. 667-681. Nwogu, O.G. and Irani, M.B. (1990). "Numerical Prediction of Higher Order Wave Induced Loads on Tethered Platforms," Proceedings of the First European Offshore Mechanics Sym- poszum, Trondheim, Norway, August 20-22. Sarpkaya, T. and Isaacson, M. (1981). "Mechanics of Wave Forces on Offshore Structures," Von Nostrand Reinhold Company, New York. Weggel, D. and Roesset, J.M. (1994). "Vertical Hydrodynamic Forces on Truncated Cylinders," to appear zn the Proceed- zngs of the 4th mternatzonal Offshore and Polar Engineer- mg Conference, ISOPE-94, Osaka, Japan, April 10-15. Wheeler, J.D. (1969). "Method for Calculating Forces Pro- duced by Irregular Waves," Proceedings of the Fzrst Off- shore Technology Conference, Houston, Texas, OTC 1006. 115