Copyright 2004, Offshore Technology Conference This paper was prepared for presentation at the Offshore Technology Conference held in Houston, Texas, U.S.A., 36 May 2004. This paper was selected for presentation by an OTC Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the Offshore Technology Conference and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of the Offshore Technology Conference or its officers. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the Offshore Technology Conference is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented.

Abstract Atlantias global performance analysis of the Matterhorn SeaStar TLP is a unique integration of computational, analytical, and experimental techniques to design an optimal system for the development of the Matterhorn field. The key global performance issues addressed during the project are discussed. These include selection of metocean data, model test calibration, coupled analysis of hull tendon and riser systems, and mitigation of vortex induced vibration of tendons and risers.

Introduction The Matterhorn SeaStar TLP was developed by Atlantia Offshore Limited for TOTAL E&P USA, INC. The project was sanctioned September of 2001, and was installed over the summer of 2003, with first oil in November of 2003. The Matterhorn structure is the 4th SeaStar TLP designed and built by Atlantia Offshore Ltd, and is the first to include a moonpool and surface completed wells. The global analysis methods and techniques have been developed by Atlantia beginning with the Morpeth SeaStar design in 1997, and have evolved through 4 designs, multiple model test programs, and evaluation of 5 years of field data of these platforms.

Platform Overview The primary function of the Matterhorn SeaStar TLP is to serve as a production and workover platform utilizing both dry trees and remote subsea wells for extraction of oil and gas from the reserves identified in Mississippi Canyon Blocks 243 and 199. The platform design accommodates up to nine dry tree, vertical top tension production risers, and has future capacity for up to five remote subsea tiebacks with associated umbilicals. The platform supports facilities for controlling and workover of the wells, processing oil and gas production to pipeline quality, and exporting production into pipelines.

Figure 1 Matterhorn SeaStar TLP

The Matterhorn platform is a single-column TLP with a central moonpool for the dry tree risers. As shown Figure 1, the hull features a single surface-piercing column with three radiating pontoons extending from the bottom of the column. The TLP also includes an independent three-level deck supported above the column by a jacket-like structure; six tendons, with two tendons attached at the end of each of the three pontoons, and six foundation piles, mechanically connected to the bottom of each tendon. Particulars are as follows:

OTC 16609

Integrated Global Performance Analysis of Matterhorn SeaStar TLP Steve Leverette, Atlantia Offshore Ltd; Oriol Rijken, Atlantia Offshore Ltd; Mike Spillane, Atlantia Offshore Ltd; Neil Williams, Atlantia Offshore Ltd.

2 OTC 16609

Hull weight: 12279 Kips Hull displacement: 52806 Kips (@104 draft) Main column diameter: 84 feet Draft: 104 feet Pontoon radius: 179 feet Pontoon height: 27-42 feet Main Column height: 125 feet

PLATFORM EAST

FWD

Figure 2 Matterhorn SeaStar TLP - Plan

Figure 3 Matterhorn SeaStar TLP - Elevation

The water depth is 2816 ft. The main tendon system is composed of 290 ft joints of nominally neutrally buoyant 32 inch pipe. The deepest sections are thicker wall (and therefore heavier) in order to resist hydrostatic collapse. The upper 1100 ft of each tendon is covered with fairings to reduce VIV response in currents.

Table 1 Tendon Pipe Properties

Section OD Pipe Length ThicknessNo. (in) D/t (ft) (in)1 32 29.5 2350 1.08472 32 28.0 366 1.1429

The payload carrying capacity of the Matterhorn SeaStar system is 26,196 Kips. The payload is comprised of the deck structure, topside facilities, drill rig, ballast and suspended risers.

Global Design Criteria The overall TLP system is designed such that all important responses in various environmental conditions meet three primary sets of safety criteria. The environmental levels and the safety criteria correspond to: Normal Conditions, Extreme (including Reduced Extreme) Conditions and Survival Conditions.

The overall dimensions of the platform and required pretension are determined/confirmed by the global performance effort. The specific objective of the global performance work is to document the global performance analysis that demonstrates the design adequacy of the TLP system to withstand the marine environment with regard to the global parameters of:

Maximum offset Deck clearance Maximum tendon tension Minimum tendon tension Maximum acceleration

These responses cases are checked against configuration and global strength criteria (tendon going slack, maximum tendon tension, deck clearance criteria), and in most cases the responses are subsequently provided to the rest of the project team as inputs to detailed design, i.e., loads and accelerations to the hull, deck, tendon, and foundation design teams.

The design loadcases include 1,000-yr, 100-yr, and 10-yr hurricanes, and 1,000-yr, 100-yr, and 1-yr winter storms. Platform responses in the 10-yr and 100-yr loop currents are also investigated.

Environment The criteria for the project were provided by the client, based upon input from various sources, including a study by Atlantia early in the FEED portion of the design identifying response based criteria applicable to this structure at this site.

Consistent, response-based metocean design criteria are developed using probabilistic methods by associating hindcast hurricane (winter storms) and Loop-Current eddy events with modeled TLP responses. The methodology, summarized in Figure 4 for hurricanes, accounts rigorously for all possible environments known to affect the site (see Spillane, et. al., 1999). There are four separate steps involved in the process :

OTC 16609 3

1. Construction of a simplified TLP response model, calibrated to model tests and LOADCASE;

2. Cataloging of TLP responses that could occur over the available storm and Loop-Current eddy histories;

3. Statistical analyses of the responses to define N-year return period response levels;

4. Selection of metocean environments that produce design level responses.

GUMSHOE mGEM4

MetOcean Eddy States

TLP Response Model Ve

Conditioned Responses

Conditioned Probabilities

Pe

Single Random Storm Probabilities

Poisson Storm ArrivalAnnulaiized Probabilities

100-yrResponse

index

Associated Metocean Variables

Figure 4 Response-Based Criteria Selection

The hurricane and winter storm hindcasts are contained in Oceanweathers GUMSHOE and WINX metocean databases. Loop-Current eddy currents were simulated based on the CASE JIPs GEM4 software (implemented in MATLAB as mGEM4).

The primary benefit of including response-based criteria is that it ensures that critical environmental combinations which may govern the platform are checked in the design. It also provides a rational means by which combinations of un-correlated events, particularly storms and eddy currents, may be specified at appropriate return periods.

Maximum and minimum tendon tension probability distributions are essentially unaffected by eddies due to their

infrequent occurrence at the Matterhorn site and small effect on extreme tension. In contrast, the distributions of maximum offset and setdown can be strongly influenced by Loop-Current eddies. At 10-year return period offset levels, the combined effect of hurricanes and eddies is more than 50% greater than the effect of hurricanes alone. At the 100-year level, this effect drops to an approximate 10% increase. At substantially longer return periods, the effect of the Loop-Current eddies compared to hurricanes diminishes to insignificance.

Loadcases Tables 2 and 3 give the loadcases evaluated in the global performance analysis. These include both intact and damaged condition cases. The intact cases, Table 2, include hurricane and winter storm survival, hurricane and winter storm extreme (with and without eddy currents), and normal conditions. The damaged cases, Table 3, are all reduced extreme cases (10-year conditions) and include flooded compartment, tendon removed, and flooded tendon conditions.

Weight Weight control/monitoring is critical to a successful floating system project. Detailed weight estimates were generated during preliminary and front end engineering to establish realistic payload targets. The weight tracking system was based on a linked spreadsheet approach that was used from the initial project phases, enabling early estimates to be quite detailed. By building upon the initial database, each stage of engineering built upon the level of detail and accuracy of the previous phase, resulting in a stable and consistent effort throughout the project life.

The budget weight cases used in global performance are given in Table 4.

Table 4 Budget Weight Cases for Matterhorn

TLP with No Drilling Rig 35024 kips TLP with Workover Rig (750 hp) summer 36442 kips TLP with Workover Rig (750 hp) winter 37470 kips TLP with Sidetrack Rig (1,000 hp) summer 36706 kips

Ballast Philosophy. The ballast philosophy for the SeaStar is that the ballast will be changed whenever there is a major change in equipment, i.e. whenever a drilling rig is added or removed or when the subsea equipment is added. The TLP will contain both trim ballast to center the lateral CG, and permanent ballast for pretension control. All trim ballast will be carried in the pontoon tip tanks, while all permanent ballast will be carried in the base-node wedge tanks

Weight Margins and Weight Growth With a TLP, as with any floating vessel, it is important to begin the project with adequate weight margins to account for the inevitable weight growth. This growth includes both misses in the early estimates, and design change through the course of the project. During the Matterhorn project, the weight estimates and margins were based on previous SeaStar project experience, and acknowledgement by operator and designer that weight was important. Figure 5 gives the weight control estimate of total weight over the course of the project life. The global

4 OTC 16609

performance analysis was performed with the budget weight, including margins. Both a minimum and a maximum weight condition were analyzed in order to cover the possible range of operating weight.

Matterhorn TLP Weight Trend

25000.0

26000.0

27000.0

28000.0

29000.0

30000.0

31000.0

32000.0

33000.0

34000.0

35000.0

5-N

ov-0

1

13-F

eb-0

2

24-M

ay-0

2

1-Se

p-02

10-D

ec-0

2

20-M

ar-0

3

28-J

un-0

3

6-O

ct-0

3

kips

Budget Project Estimate Figure 5 Weight estimate over project duration

Loads The environmental loading on a TLP includes both static and dynamic components of wind, wave, and current, as well as water level (tide and storm surge).

Wind load The wind loads on the TLP were computed using the commercial program WINDOS developed and marketed by MARIN. The analysis involves defining a multi-component model composed of a large number of elements that approximate the geometry of the superstructure. Wind load coefficients for each element are assigned within the WINDOS program, taking into account the particular element geometry, the proximity of neighboring elements etc. Figure 6 shows the WINDOS model of the SeaStar superstructure used in the present study (no rig case). In the Matterhorn modeling, the wind areas of the workover rig and sidetrack rig are assumed to be equal.

For a particular structural configuration, wind speed, and direction, the WINDOS program then produces a set of global wind load coefficients. In the present study, wind loads at other wind speeds were obtained by multiplying these coefficients by an appropriate factor based on wind velocity-squared load dependence. The resultant force and center of pressure for a number of loadcases are presented in Table 5.

Table 5 Wind Force and Center of Pressure for Selected Load

Cases Using 1-Minute Wind Speed Loadcase # Environment Rig Configuration Force Direction Center of Pressure

(kips) (deg) (ft, above keel)1001 1000 year hurricane without rig 416 140 2051101 100 year hurricane without rig 333 140 2041103 100 year hurricane with rig 461 140 2271105 100 year winter storm with rig 233 120 2251301 1 year winter storm without rig 70 120 2051303 1 year winter storm with rig 93 120 224

Wind tunnel tests for the wind load were considered. However, the configuration of the SeaStar, with very little of the hull exposed, resembles a conventional jacket structure, and programs such as WINDOS are fairly well established for estimating wind loads on such structures. The fact that the TLP does not pitch or roll makes this simplification possible.

Figure 6 WINDOS Model for Matterhorn Wind Loads

Current Load Current loads on the TLP hull and the tendons are calculated based on the current force formulation in API RP 2T. Although it is known that the drag coefficient for the TLP hull and tendon system is a function of current speed and heading, in the present analysis, representative constant drag coefficients have been used. The drag coefficients for the columns and pontoons have been taken as Cd = 1.0, and 2.0, respectively. The drag coefficient for the tendons was taken to be 0.8 for the first 1470 ft beneath the water surface (covered with fairings) and increased to 1.0 thereafter. The current forces for selected cases are presented in Table 6.

Table 6 Breakdown of Current Force for Selected Load Cases

Case # Environment From Hull From Tendon From Riser Total Force Direction(kips) (kips) (kips) (kips) (deg)

1001 1000 year hurricane 302.7 58.2 58.3 419.2 1131101 100 year hurricane 239 39 38.8 316.8 1041105 100 year winter storm 46.8 7.6 7.5 61.9 1201301 1 year winter storm 5.2 2 2.1 9.3 120

Towing tests on an earlier SeaStar established that the drag due to current is not overly sensitive to current direction, that the platform response in various current directions is stable (no galloping), and that there is only a small component of hull VIV which leads to slightly larger drag coefficients covered by the above values.

Wave Loads The wave-frequency hydrodynamic loads are calculated in the frequency domain using the three-dimensional radiation/diffraction panel program OSAWAVE. In OSAWAVE, the TLP is treated as a rigid body oscillating in six degrees of freedom, subjected to regular Airy waves. The program gives the hydrodynamic pressure distribution, the exciting forces and moments, and the added-mass, and radiation damping coefficients in each of the six degrees of freedom. The panel mesh adopted consists of 1,188 panels and is shown in Figure 7. The hydrodynamic coefficients obtained using this mesh are convergent to within 1-2%.

OTC 16609 5

-100

0

100

200

-200-150

-100-50

050

100150

200-150

-100

-50

0

x (ft)

osamesh.dat

y (ft)

z (ft

)

Figure 7 OSAWAVE Panel Model for Matterhorn Wave Loads

The resultant hydrodynamic load, mass, and damping information is used directly in the frequency domain load and motions analysis (LOADCASE), and are processed to obtain the infinite frequency added mass matrix, and the excitation and retardation functions used in the time domain coupled analysis TLPSIM.

Mean Wave Drift / Slow Drift The mean wave drift loads and slow drift responses of the TLP are estimated based on the Matterhorn TLP model test data. The mean drift loads are taken to be functions of wave heading, significant wave height (Hs), and spectral peak period (Tp). A cubic polynomial in (Hs/Tp)2 has been fitted to the data for various wave headings. The experimental data and the polynomial fit for wave heading of 0 degrees is shown in Figure 8.

0

20

40

60

80

100

120

140

0.00 2.00 4.00 6.00 8.00 10.00 12.00

(Hs/Tp)^2

Mea

n D

rift F

orce

(kip

s)

0 degrees

Figure 8 Mean Wave Drift Coefficients from Model Tests

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

0.00 2.00 4.00 6.00 8.00 10.00 12.00

(Hs/Tp)^2

RM

S Sl

ow D

rift R

espo

nse

(ft)

0 degrees

Figure 9 Slow Drift Coefficients from Model Tests

In a similar manner, the rms of the slow drift response is also estimated from the model test data using a cubic polynomial fit in (Hs/Tp)2, again for different wave headings. The experimental data and the polynomial fit for wave heading of 0 degrees is shown in Figure 9.

VIV/VIM The Vortex Induce Vibration / Vortex Induced Motion responses of the system are dealt with in various ways. For the hull, as mentioned above, the VIM response has been shown to be small and of little consequence to design. Results from the model tests are used to modify the drag coefficients for maximum offset calculations.

For tendons, the SHEAR7 program is used to evaluate the fatigue due to VIV response. Because of previous experience of other SeaStars in high currents (see Leverette, et. al, 2003), it was decided at the beginning of this project to install fairings on the upper portion of the tendons. During the course of this project, further evolution in the cold core current criteria led to the decision to extend the fairings to 1400 water depth. As a consequence, VIV of tendons, and any subsequent effects on the hull and deck, became an insignificant issue for Matterhorn. VIV of the production risers on Matterhorn is dealt with in Jordan, et. al. 2004.

Model Tests A scaled version of the Matterhorn TLP hull was tested at the OTRC wave basin in College Station during late Fall of 2001. These series of experiments were performed with a focus on the hydrodynamic response of the TLP under a wide range of sea states. The experimental observations are subsequently used to calibrate the hydrodynamic modules of numerical tools such as time domain simulations, and to validate the design methodology.

The TLP is modeled with full length tendons which resulted in a scale factor of 1:54 based on the depth of the test basin pit. This scale implies a main column diameter of approximately 18.7 inch, the height of the hull is approximately 27.8 inch and the deck and hull weight is approximately 210 lbs (Figures 10 and 11). The wind load on the TLP is simulated as an offset force acting on the whole TLP, thefore the deck is represented by a simple rectangular box with aluminun decks and aluminum girders. Key issues in modeling the tendons are to model the axial stiffness properly, provide very light axial damping (measured heave damping is less than 1% of critical) and to provide the correct offset setdown curve. All tendons are modeled individually. Each tendon is made from inch OD aluminum tube with an axial spring at the bottom to achieve the desired stiffness characteristics. The central moonpool has a scaled diameter of 8 in. The keel plate in the moonpool is replaceable, and several were provided during the test program. Each keel plate is a different design for the lower riser guide frame. The dimensions of this scaled model are all reasonable values in terms of model construction and experimental setup. Further, the scaled sea states, with prototype significant wave heights ranging from 6.5 to 52 ft, all lay within the capacity of the wave generator.

The experiments simulated platform response under a variety of different environmental conditions. The platform is subjected to a total of 14 sea states ranging from near operational conditions to seastates beyond the survival

6 OTC 16609

condition (Table 7). Most of the sea states, in particular the larger ones, are were selected to be at the steep range of the criteria. The steep waves have been shown to provide the most non-linear TLP response, and the desire was to define the limits of this for design. This non-linear response can typically only be generated experimentally, and that non-linear response can be incorporated in the numerical formulations by calibrating the numerical formulations to the model test results.

Figure 10 Underwater View of Model

Each sea state is calibrated prior to installing the model in the basin. The wave elevation along the response center line is measured at specific locations without the presence of the model. Those wave elevation records are checked for spectral content and trough, crest and wave height distributions. Each undisturbed wave elevation record is used to develop experimental response transfer functions which are used as reference values during the calibration of the numerical formulations.

Each sea state is characterized by a particular combination of wind and current. The wind and current load on the platform result in a mean offset position. It is this mean position about which the response occurs. By using an

externally applied force which places the platform at the sea state dependent offset location, the platform geometry (setdown, tendon angle, pretension distribution) is correct for this conditins and measured platform dynamic response is solely due to hydrodynamics, and not complicated with time varying current and wind force effects. The magnitude of the wind and/or current force is established using numerical methods as described above.

Figure 11 Matterhorn Model in Basin

Table 7 Environmental Conditions For Tests Sea State

Hs Tp Wind Current (ft) (s) (-) ft/s ft/s

S01 6.5 5.2 1 0 0.0 S02 10.0 6.4 1 0 0.1 S03 14.5 7.8 1 0 0.5 S04 21.5 9.6 1 0 1.2 I01 26.0 10.8 1 83.1 1.6 I02 26.0 12.6 1 83.1 1.6 I03 32.0 12.4 2.2 99.7 2.2 I04 35.0 13.0 2.2 108.0 2.5 I05 38.0 14.0 2.2 116.3 2.8 I06 41.0 13.8 2.2 124.6 3.1 I07 42.8 14.6 2.2 129.5 3.3 I08 47.0 14.5 2.2 141.1 3.7 I09 48.5 15.2 2.2 145.3 3.8 I10 52.0 15.8 2.2 155.0 4.1

The platform and tendons are instrumented to obtain the response behavior under the different sea states. The motion of the TLP is established using an optical tracking system and several accelerometers which provide the six degree of freedom displacement and acceleration vectors respectively. The airgap at four locations around the platform is measured as is the water level at two locations in the moonpool. Tendon tensions are recorded at the lower end of each tendon. The force representing the wind and current loading is applied though a constant tension winch, the applied force is recorded.

OTC 16609 7

The test program included testing the TLP under two weight conditions which reflect the presence (high VCG) and absence (low VCG) of the drilling equipment and associated ballast distribution. Three different keel plate configurations were installed on the model to examine the effect of keel plate porosity on the TLP response. Some of those findings will be discussed later in this paper. Experiments were performed with an apparent wave heading of 0o, 60o and 90o which is achieved by rotating the foundation. Symmetry considerations can be used to obtain apparent wave heading ranging from 0o to 360o at 30o increments. Static offset and decay tests are performed at each wave heading to check the tendon system and TLP mass properties. Static offset tests are also used to obtain drift force estimates. Additional special experiments are also performed, including various test of the TLP with a tendon removed.

The model tests are an integral part of the global performance design and analysis of the Matterhorn TLP. Rather than comparing a combined response estimates, the individual components which make up the response are compared (and calibrated). For example, the code equation for maximum dynamic tension is computed as a combination of wave-plus-high-frequency and slow drift tension effects :

highwaveslowdyn TTT ++= The wave-plus-high-frequency tension part, which includes higher order tension responses, is defined as : waveewavedyn QKKT =, where wave is the theoretical wave frequency-only tension RMS and Q takes into account the probability of occurrence and is based on the ratio of the environmental exposure time and the mean zero crossing time of the dynamic tension. The factor eK relates the most-likely extreme dynamic tension based on best-fit extreme value estimate (e.g. Weibull or Gumbell distribution) to the Rayleigh distribution based wave frequency-only estimate. The factor K relates the measured wave frequency tension RMS to the theoretical tension RMS. An alternative way to interpret the factor K is that it takes into account the difference between the measured and observed tension RAOs :

( ) ( )( ) ( )

=

WFm

WFe

dTS

dTSK

2

2

2

where WF indicates the wave frequency range, ( )eT the experimentally obtained transfer function (RAO) and ( )mT the theoretical transfer function (RAO). The function ( )S represents the measured wave spectrum. The factors eK and

K are unique for each tendon and, as the model tests have indicated, are sea state specific.

Response Calculations The TLP global analysis is performed primarily using the LOADCASE spreadsheet system. The use of the LOADCASE spreadsheet system to carry out the global analysis yields a convenient method to consistently calculate and document the TLP responses for tension, offset, acceleration, and deck clearance under the environmental conditions and platform configurations specified in each load case. The LOADCASE spreadsheet system is a series of linked EXCEL worksheets that carry out the frequency domain global analysis of the SeaStar TLP. An overview of the calculation methods is presented herein.

The basic methodology for calculating design global performance loadcases is based on API RP 2T.

The pretension in each of the TLP tendons is calculated based on a detailed description of the weights of the various components of the platform deck and hull, and the mean offset induced by wind, current and steady wave drift loads. The dynamic tendon tensions at the wave-frequencies are obtained from the platform response RAOs. The TLP rigid-body response is calculated from the linearized equations of motion in the six modes of oscillation using the hydrodynamic data (exciting forces, added mass, radiation damping) supplied by OSAWAVE and treating the tendons as linear springs. A portion of the mass of each tendon is added to the vessel mass matrix for the dynamic response calculations. For a given input wave spectrum, the extreme wave-frequency tendon tensions in the up-wave and down-wave tendons may then be calculated. These values are then corrected for nonlinear wave effects using correction factors obtained from model tests (as discussed earlier). The final part of the LOADCASE spreadsheet system utilizes these quantities in code equations to compute the design responses of maximum and minimum tension, maximum offset, maximum accelerations, and minimum deck clearance.

Design Responses The design responses of maximum and minimum tension, maximum offset, and minimum deck clearance are computed in LOADCASE. The general approach for the prediction of the extremes of each response is conservatively based on the worst combinations of wind, wave, current, and tide specified for each load case, summed in the most onerous fashion. Statistical methods are used for the prediction of extreme responses. The project standard for the statistical analysis of extremes is the most probable extreme for narrow-band Gaussian processes with Rayleigh-distributed peaks. When appropriate (in most cases), these estimates are modified based on statistical correction factors determined from model tests in order to account for nonlinear (non-Gaussian) effects. For tension, the factors vary by tendon, by wave direction, and by seastate. Different factors apply for minima and for maximal.

Maximum Tension The LOADCASE analysis assumes a non-collinear wind, wave and current environment, therefore the identification of the tendon which experiences the maximum tension depends on the full specification of wind, wave, and current values and directions. The maximum tension will usually occur in an upwave or downwave tendon. Exceptions to this may occur if a tendon is removed for

8 OTC 16609

maintenance or if a compartment is flooded due to hull damage. Therefore, the maximum tensions in all tendons are calculated, and the largest value reported. The maximum tension is one of the design parameters that control the strength requirements for the tendons and tendon porches, and for portions of the hull.

The maximum tension in LOADCASE is calculated as the sum of the static pretension, the setdown-induced tension from the static offset calculations, the tension induced by the moments of the wind, current and steady drift forces, and the wave-induced tension from the tendon tension RAO. The steady offset of the platform is calculated iteratively based on a horizontal force balance involving the wind, current and steady wave forces, and the tendon tension induced by the resultant offset (assuming straight-line tendons). The total offset for the maximum tension case is then obtained by taking the largest resultant mean plus dynamic (wave frequency plus slow drift) offset vector. The maximum tension occurs at the top of the tendon.

It should be noted that the tendon tension contributions due to higher-order wave responses (ringing and springing) are relatively small for the SeaStar TLP and are incorporated into the design through the extreme statistical factors derived from the model tests.

Minimum Tension The LOADCASE analysis assumes non-collinear wind, wave and current, therefore the identification of the tendon which experiences the minimum tension depends on the full specification of wind, wave, and current values and directions. The minimum tension will usually occur in an upwave or downwave tendon. Exceptions to this behaviour may occur if a tendon is removed for maintenance or if a compartment is flooded due to hull damage. Therefore, the minimum tensions in all tendons are calculated, and the smallest value reported. The minimum tension is one of the design parameters that control the pretension requirements for the TLP.

The minimum tension in LOADCASE is calculated as the sum of the static pretension and the setdown-induced tension from the static offset calculations, the tension induced by the moments of the wind, current and steady-drift forces and the wave-induced tension from the tendon tension RAO. The steady offset of the platform is again calculated iteratively based on a horizontal force balance involving the wind, current and steady wave forces, and the tendon tension induced by the resultant offset (assuming straight-line tendons). The total offset for the minimum tension case is then obtained by taking the smallest resultant mean minus dynamic (wave frequency plus slow drift) offset vector. The minimum tension occurs at the bottom of the tendon.

The contribution to minimum tension due to ringing and springing are handled through the model test statistical factors, in a similar manner to the maximum tension calculations.

Maximum Offset The maximum offset of the platform occurs in response to the action of wind, wave and current forces acting on the hull, tendons and risers. The mean offset of the platform is calculated iteratively based on a horizontal force balance involving the wind, current and steady wave (drift) force vectors, and the tendon tension induced by offset

(assuming straight-line tendons). The maximum dynamic offset is calculated as the most probable maximum of the combination of wave-frequency (surge) offset and slow drift values. The maximum total offset is then obtained by vectorially adding this dynamic offset to the mean offset.

In order to check the simplified analysis which is implemented as frequency domain code equations in the LOADCASE spreadsheet, a time domain simulation utilizing a fully coupled hull/tendon/riser numerical model was performed. This simulation includes linear diffraction results implemented as a retardation function, V2 drag due to wave, current, and motion relative velocities, and finite wave height effects based on a strip theory approach in the splash zone. The simulation was validated against the model tests. One very satisfying result from this validation was the comparison of the resultant slow drift peak. For both model tests and simulation result, it is clear that the slow drift of the platform includes both resonant and non resonant components, with the dominant energy being non-resonant. The simulation was performed with the measured wave history from the model test. Figure 12 gives the comparison, with the resonant peak being the peak at approximately 0.008 Hz.

0 0.02 0.04 0.06 0.08 0.1 0.120

2000

4000

6000

8000

10000

12000

14000

frequency (Hz)

tlpsi

m S

urge

((ft)

^2/H

z)

100drif9.out

tlpsim Surgemodel test Basin Surge

Figure 12 Comparison of Model Test Surge spectrum with

TLPSIM simulation results Minimum Deck Clearance The deck clearance is calculated as the difference between the elevation of the lowest deck top-of-steel at maximum setdown and the wave crest elevation, including tide and wave amplification effects. The wave amplification factor is to account for the increase in water surface elevation as the wave interacts with the platform (i.e., wave upwelling and runup effects). The minimum deck clearance is assumed to occur at the maximum offset.

The design water level, which includes the effects of tide, storm surge and subsidence, has a significant influence on the platform dynamics. This interaction is such that the high design water level will not always produce the minimum deck clearance. This occurs because although at the higher water level there is less initial clearance, the platform offset and resulting setdown are also less. Therefore, the deck clearance is checked at both low and high design water levels. The frequency domain analysis draft for the low design water level (LDWL) is defined as:

OTC 16609 9

still water draft - tidal amplitude + setdown + storm surge

The frequency domain analysis draft for the high design water level (HDWL) is defined as:

still water draft + tidal amplitude + setdown + storm surge

In computing deck clearance, the effects of subsidence and the possible correction to the deck height due to deck leveling during installation are taken into account. The deck clearance is therefore computed according to

Still water deck height - max. setdown wave crest height

subsidence deck leveling allowance

Based on model test observations, wave amplification due to runup is negligible for this structure. The disturbances to the wave field due to the presence of the TLP are small and localized in extent. Combined with the fact that the TLP is moving with the wave crest during crest passage through the structure (resulting in lower relative velocities), this results in a minimal effect on the structure due to local crest enhancements. Therefore, the deck height is set to ensure positive clearance (5 margin) in design extreme conditions, and non-negative (i.e. no impact other than the above-mentioned local effects), in survival conditions.

Loads on Keel Plate Near the bottom of the moon pool there is a Keel Plate which is the support structure for the lower lateral restraint of the risers (Jordan, et. al., 2004). The loading on the keel plate required a novel analysis. The moon pool level was observed to oscillate at resonance (11-14 seconds) during the model tests. The total motion of the moon pool free surface included direct wave forcing, overtopping fill/venting at the top of the column, as well as the resonant motion. The keel plate partially obstructs the flow in and out of the moon pool. During the model tests, alternate designs for this structure were tested to determine an optimum effective closure to optimally damp the moon pool resonance.

In order to calculate design loads on the keel plate, maximum flow velocities were estimated from model test data of the moon pool level, processed to remove fill effects that do not represent flow though the keel plate, and treated statistically to estimate extreme values. A load model was generated from techniques normally used for steady flow through a perforated plate (see Blevins, 1984, and Leverette and Spillane, 1994). The resulting loads were used for the hull structural design. A subsequent analysis for smaller seastates was used to generate fatigue loads for this design.

Response Results The results of the global performance analysis include extemes for each major response for each loadcase. Table 8 gives a summary of the results for the intact loadcases. Finally, utilizing the LOADCASE analysis with the tension margins which account for the operational weight variablity removed, a series of analyses are performed with varying weight and center of gravity. These runs are used to define weight limits which match the design calculations with margins. The weight limits are then used to define an

operational weight triangle, which provides the operational envelope which the operator uses to control weight of the platform. All points within the triangle are acceptable from global performance and structural strength criteria. Different triangles are provided for different seasons (hurricane versus non-hurricane) and for different VCG conditions. A sample triangle for the no-rig VCG during hurricane season is given in Figure 12.

0

1

2

3

4

5

6

32000 32500 33000 33500 34000 34500 35000 35500 36000

Weight (kips)CG

Offs

et (f

t) 34,400 kips @ 3.0 ft offset

35,300 kips33,000 kips

MASS VCG @ 110 ft AKL

Figure 12 Weight Triangle of Allowable Weight Condition based on Global Performance Results

Installation The installation of the TLP involves pre-installation of the driven anchor piles, installation of the tendons using temporary buoyancy so that they are free standing, and floatover of the hull and ballasting down to install the hull over the pre-installed free standing tendons. The deck is then installed by single piece lift in a manner of setting a deck on a fixed jacket.

For the Matterhorn installation, high current conditions were encountered during the tendon installation, and operations had to be halted with three tendons in place with currents exceeding 2 knots. This exceeded the original design value set for installation of the tendons. Analysis of the free standing tendons, using global performance models, was used to guide operations. Active steps taken during this period included use of a tug attached to the tendon top assembly to keep two free standing tendons separated, and to limit the bottom angle. The fairings which were installed to prevent fatigue damage to the tendons and structure during intact eddy events proved to be invaluable in allowing the partially installed system to survive this event. No significant vibrations of the free standing tendons were observed, and bottom angles remained within allowable limits. Conclusions The overall conclusion of the global peformance evaluation of the Matterhorn SeaStar is that the SeaStar TLP has been designed to safely meet all reasonably expected environmental events, meets customer and applicable code requirements, and has already proven its robustness by proving capable of severe current conditions during installation

10 OTC 16609

Acknowledgements The authors would like to acknowledge TOTAL E&P USA, INC. for supporting this project, and for encouraging publication of the results.

References Blevins, R. D., Applied Fluid Dynamics Handbook, Van Nostrand

Reinhold Company Inc, 1984. Jordan, Otten, Trent, and Cao, Matterhorn Dry-Tree Production

Risers, Paper 16608, Offshore Technology Conference Proceedings, May 2-6, 2004.

Leverette, Rijken, Dooley, Thompson, Analysis of TLP VIV responses to Eddy Currents, Paper 15289, Offshore Technology Conference Proceedings, Houston, May 5-8, 2003.

Leverette and Spillane, Hydrodynamics of the Roseau Tower Added Mass Trap, in BOSS94, Behaviour of Offshore Structures, Cambridge, Mass., Pergamon Press, 1994.

Spillane, Haring, Irick, and Chen, Response-Based Criteria Development for the GB260 Compliant Tower, Paper 10916, Offshore Technology Conference Proceedings, Houston, May 3-6, 1999.

Table 2 Intact Loadcases for Matterhorn

MATTERHORN TLP - INTACT

Description1001 1000 yr hurricane - HDWL x x x x x1002 1000 yr hurricane - LDWL x x x x x1003 100 yr hurricane - HDWL x x x x x1004 100 yr hurricane - LDWL x x x x x1005 1000 yr winter storm - HDWL x x x x x1006 1000 yr winter storm - LDWL x x x x x

1101 100 yr hurricane - HDWL x x x x x1102 100 yr hurricane - LDWL x x x x x1103 100 yr hurricane - HDWL x x x x x1104 100 yr hurricane - LDWL x x x x x1105 100 yr winter storm - HDWL x x x x x1106 100 yr winter storm - LDWL x x x x x1107 100 yr hurricane plus LCE - HDWL x x x x x1108 100 yr hurricane plus LCE - LDWL x x x x x1109 100 yr hurricane plus LCE - HDWL x x x x x1110 100 yr hurricane plus LCE - LDWL x x x x x1111 100 yr winter storm plus LCE - HDWL x x x x x1112 100 yr winter storm plus LCE - LDWL x x x x x

1201 100 yr LCE - HDWL x x x x x1202 100 yr LCE - LDWL x x x x x1203 100 yr LCE - HDWL x x x x x1204 100 yr LCE - LDWL x x x x x1205 100 yr LCE - HDWL x x x x x1206 100 yr LCE - LDWL x x x x x

1301 1 yr winter storm - HDWL x x x x x1302 1 yr winter storm - LDWL x x x x x1303 1 yr winter storm - HDWL x x x x x1304 1 yr winter storm - LDWL x x x x x1305 1 yr winter storm - HDWL x x x x x1306 1 yr winter storm - LDWL x x x x x

EnvironmentPlatform Condtion Safety CriteraDrill Rig StatusWeight Status

OTC 16609 11

Table 3 Damaged Loadcases for Matterhorn

M A T T E R H O R N T L P - D A M A G E D

D escrip tion1401 10 y r hu rricane - LD W L x x x x x1402 10 y r hu rricane - LD W L P 2A flooded x x x x x1403 10 y r hu rricane - LD W L P 2B flooded x x x x x1404 10 y r hu rricane - LD W L B 2 flooded x x x x x1405 10 y r hu rricane - LD W L B 31 flooded x x x x x1406 10 y r hu rricane - LD W L C 1 flooded x x x x x1407 10 y r hu rricane - LD W L C 2 flooded x x x x x1408 10 y r hu rricane - LD W L C 3 flooded x x x x x1409 10 y r w in te r s to rm - LD W L x x x x x1410 10 y r w in te r s to rm - LD W L P 2A flooded x x x x x1411 10 y r w in te r s to rm - LD W L P 2B flooded x x x x x1412 10 y r w in te r s to rm - LD W L B 2 flooded x x x x x1413 10 y r w in te r s to rm - LD W L B 31 flooded x x x x x1414 10 y r w in te r s to rm - LD W L C 1 flooded x x x x x1415 10 y r w in te r s to rm - LD W L C 2 flooded x x x x x1416 10 y r w in te r s to rm - LD W L C 3 flooded x x x x x1417 10 y r LC E - LD W L x x x x x1418 10 y r LC E - LD W L P 1A flooded x x x x x1419 10 y r LC E - LD W L P 1B flooded x x x x x1420 10 y r LC E - LD W L B 1 flooded x x x x x1421 10 y r LC E - LD W L B 23 flooded x x x x x1422 10 y r LC E - LD W L C 1 flooded x x x x x1423 10 y r LC E - LD W L C 2 flooded x x x x x1424 10 y r LC E - LD W L C 3 flooded x x x x x

1501 10 y r hu rricane - H D W L tendon 3 o r 4 rem oved x x x x x1502 10 y r hu rricane - LD W L tendon 3 o r 4 rem oved x x x x x1503 10 y r hu rricane - H D W L tendon 3 o r 4 rem oved x x x x x1504 10 y r hu rricane - LD W L tendon 3 o r 4 rem oved x x x x x1505 10 y r w in te r s to rm - H D W L tendon 3 o r 4 rem oved x x x x x1506 10 y r w in te r s to rm - LD W L tendon 3 o r 4 rem oved x x x x x1507 10 y r LC E - H D W L tendon 1 o r 2 rem oved x x x x x1508 10 y r LC E - LD W L tendon 1 o r 2 rem oved x x x x x

1601 10 y r hu rricane - H D W L tendon 3 o r 4 flooded x x x x x1602 10 y r hu rricane - LD W L tendon 3 o r 4 flooded x x x x x1603 10 y r hu rricane - H D W L tendon 3 o r 4 flooded x x x x x1604 10 y r hu rricane - LD W L tendon 3 o r 4 flooded x x x x x1605 10 y r w in te r s to rm - H D W L tendon 3 o r 4 flooded x x x x x1606 10 y r w in te r s to rm - LD W L tendon 3 o r 4 flooded x x x x x1607 10 y r LC E - H D W L tendon 1 o r 2 flooded x x x x x1608 10 y r LC E - LD W L tendon 1 o r 2 flooded x x x x x

E nv ironm entP la tfo rm C ondtion S a fe ty C rite raD rill R ig S ta tusW e igh t S ta tus

Table 8 Intact Loadcase Results for Matterhorn

MATTERHORN TLP - INTACT

Load

case

#

Description

Maximum Tension (kips)

Minimum Tension (kips)

Maximum Offset

(ft)

Minimum Deck Clearance

(ft)

Maximum Tendon Angle

(degrees)

Maximum Deck

Acceleration (g's)

1001 1000 yr hurricane - HDWL 5522 273 187 1.4 3.9 0.2761002 1000 yr hurricane - LDWL 5435 78 190 4.1 4.0 0.2741003 100 yr hurricane - HDWL 5142 251 170 9.4 3.6 0.2431004 100 yr hurricane - LDWL 5050 65 174 12.2 3.7 0.2421005 1000 yr winter storm - HDWL 4686 671 99 27.5 2.1 0.1771006 1000 yr winter storm - LDWL 4579 491 101 30.4 2.1 0.176

1101 100 yr hurricane - HDWL 5146 780 150 10.6 3.2 0.2481102 100 yr hurricane - LDWL 5049 592 153 13.4 3.2 0.2461103 100 yr hurricane - HDWL 5142 251 170 9.4 3.6 0.2431104 100 yr hurricane - LDWL 5050 65 174 12.2 3.7 0.2421105 100 yr winter storm - HDWL 4397 1287 87 33.0 1.8 0.1601106 100 yr winter storm - LDWL 4285 1117 90 36.0 1.9 0.1591107 100 yr hurricane plus LCE - HDWL 4494 2160 182 29.1 3.8 0.1701108 100 yr hurricane plus LCE - LDWL 4342 2011 173 32.7 3.7 0.1691109 100 yr hurricane plus LCE - HDWL 4384 1891 196 28.2 4.1 0.1731110 100 yr hurricane plus LCE - LDWL 4274 1747 201 30.8 4.2 0.1721111 100 yr winter storm plus LCE - HDWL 4011 1940 145 38.9 3.1 0.1261112 100 yr winter storm plus LCE - LDWL 3892 1792 150 41.6 3.2 0.125

1201 100 yr LCE - HDWL 3960 2862 201 46.9 4.2 0.0581202 100 yr LCE - LDWL 3838 2732 206 49.5 4.4 0.0581203 100 yr LCE - HDWL 3698 2336 222 45.3 4.7 0.0631204 100 yr LCE - LDWL 3580 2205 228 47.7 4.8 0.0631205 100 yr LCE - HDWL 3866 2512 215 45.8 4.5 0.0641206 100 yr LCE - LDWL 3757 2376 221 48.3 4.7 0.064

1301 1 yr winter storm - HDWL 3773 2541 38 49.2 0.8 0.0801302 1 yr winter storm - LDWL 3637 2391 39 52.2 0.8 0.0791303 1 yr winter storm - HDWL 3515 2062 44 49.1 0.9 0.0821304 1 yr winter storm - LDWL 3381 1911 46 52.1 1.0 0.0821305 1 yr winter storm - HDWL 3638 2116 43 49.1 0.9 0.0831306 1 yr winter storm - LDWL 3503 1956 45 52.1 0.9 0.083