STATION KEEPING

EXTENSIVE MODEL TESTING OF A DRY-TREE SPREAD-MOORED BARGE IN BRAZILLIAN WATERS Arjan Voogt (MARIN) and Mamoun Naciri (SBM) Deep Offshore Technology XIV (DOT-2002) ABSTRACT This paper describes the overall behavior of a new build barge to be spread moored offshore Brazil in 1,200 m water depth. The barge supports 12 dry trees located in a well bay near amidships. Experiments have been carried out on a scaled model of the barge in survival environments ranging from 4.7m significant beam waves to 7.8m significant bow quartering waves in the presence of wind and strong current. This model test campaign has been partly funded by five oil companies. Extensive instrumentation was used to provide information on vessel and risers during these tests. The barge motion behavior was monitored to allow assessment of the floater sea-keeping properties and of the riser/hull clearance. Wave gauges fixed to the barge measured the moonpool agitation and its effects, if any, on the risers. Riser top tensions were monitored to determine their dynamics and to verify the over-tension factor, which can be compared to other dry tree solutions. Finally, various loads in the tensioning system were monitored and provide valuable information for the hull structural design. The extensive model test campaign has confirmed 1) the feasibility of the Tension Leg Deck (TLD) system to pretension risers offshore Brazil; 2) the seakeeping performance of the spread moored barge supporting these risers; 3) the riser/tendon clearance and 4) the wave agitation in the moonpool. 1 INTRODUCTION With oil exploration and production moving to ever-deeper waters with few or no infrastructures, dry tree concepts providing the full range of functionalities i.e. production, drilling, storage and offloading could decrease the lead-time to first oil and therefore be very attractive. One such concept has been developed in Single Buoy Moorings starting in 1998 and is called the Tension Leg Deck (TLD). The TLD concept is based on the use of gravity – instead of buoyancy in existing dry tree concepts – as a means to pretension risers. In this concept, risers are connected to a deck structure sheltered inside a moon-pool. This deck is suspended above the mean sea level by cables running over sheaves and connected to weights. These weights can either be under water or inside dedicated hull compartments. Development of the TLD system with weights in water and its experimental verification for West of Africa environments have been reported in [1], [2], [3] and [4]. Figure 1 below shows a schematic of a TLD system with weights inside the vessel hull. These weights are located at the end of articulated lever arms free to roll with respect to the vessel hull. When the vessel heaves up, the lever arms roll up while the deck structure in the moon-pool remains stationary.

Figure 1 Schematic of the TLD system with weights in dedicated hull compartments Further developments have recently been undertaken to investigate the feasibility of a 12-well TLD system in a new build barge spread moored offshore Brazil in 1200 m water depth [5]. The purpose of the present paper is to report on recent experimental work confirming this feasibility. The main objectives of this new model test campaign were as follows: • To confirm the motion/acceleration performances of a barge concept with horizontal skirts near the keel in typical 100-year Brazilian environments. These appendices effectively shift the roll natural period above the range of wave energy [6] [7], • To confirm the feasibility of putting the weights at the tip of articulated lever arms in dedicated hull compartments, • To confirm the calculated over tension factors in the TLD cables and in the risers, • To verify clearances between the deck and the barge moon-pool walls, • To assess the wave agitation in the moon-pool. This model test campaign took place during February and March 2002. Model construction and instrumentation is covered in section 2. In section 3, static and decay tests are discussed followed by the analysis of the FPDSO-TLD system motion response in section 4. Section 5 covers the load response of the TLD system and while section 6 addresses a qualitative and quantitative description of the moonpool agitation, the comparison of some salient model test results with the predictions of our numerical software is discussed in section 7. Conclusions are drawn in section 8. 2 MODELS AND INSTRUMENTATION The model tests were performed at a 1:60 scale in MARIN's Offshore Basin. At this scale an accurate modeling of the barge and TLD system was possible. Owing to depth limitations of the basin, the modeled water depth was reduced from the prototype 1200m to 570m. Therefore truncated risers, tendons and mooring system were designed. Barge The barge has shoebox geometry (see Figure 2) and is fitted with skirts along the side to minimize the roll and heave motions and accelerations. The particulars of this barge are in Table 1. This hull form also includes a moonpool with walls vertical from keel to main deck. The moonpool walls were made of Plexiglas in order to allow visualization of the lever arm system. Blocks of wood and foam were constructed to model the topside and achieve realistic wind forces. Figure 2 shows a photo of the barge.

Figure 2 Barge model TLD system For the sake of simplicity, four pretension weights have been used instead of eight in reality (Figure 1 and Table 2). Four large sheaves were mounted above the barge main deck near amidships. Four cables run over these sheaves to connect the pretension weights to the deck supporting the risers and tendons. Each of these pretension weights is guided in its vertical motion by a hinged arm that is free to roll relative to the barge hull (see Figure 3). In addition to the above transverse system a longitudinal system was modeled with weights guided by hinged arms free to pitch relative to the barge hull. During the tests discussed in this paper the latter setup was used. The particulars are shown in Table 2.

Table 2 Particulars of TLD system

Figure 3 Schematic view of TLD deck system (transverse arms)

Figure 4 shows a photo of the TLD system with longitudinal arms connected. In the center of the Plexiglas moonpool, the black rectangular TLD deck can be seen. The risers and tendons were connected below the deck. At the four corners on top of the deck a wire running over the sheaves and connecting the weight to the TLD deck can be observed. The hinges of both arms are close to one another at the centerline of the vessel.

Figure 4 TLD system showing the longitudinal hinged arms, weights, cables, sheaves and TLD deck

Risers and Tendons The prototype TLD deck will support twelve risers and four tendons. The required riser spacing is 3m at deck level and 8m at the seabed. This requires flaring in both the longitudinal and transverse directions. In order to simplify the model and later the numerical calibrations, four equivalent risers replaced the twelve individual risers. The four tendons and four equivalent risers in the reduced water depth were modeled to represent the mass, submerged weight and axial stiffness of the prototype system (full water depth). This results in a larger diameter to get the

same buoyancy over the truncated depth of 570m. Attention was paid to distribute the axial stiffness of the risers/tendons along their length within the given physical constraints. To correct for stiffer materials at model scale, springs were used at deck level as well as tank bottom. The bending stiffness was not a main concern in these tests owing to water depth truncation and was therefore not modeled to scale. Mooring The emphasis of these tests was on the Tension Leg Deck (TLD) system. Consequently, the mooring system was predominantly designed with TLD considerations in mind. The mooring system was designed for the truncated depth of 570 m and consists of four bundles of four semitaut lines. Each line has a top and bottom chain segment and a polyester rope in between. This polyester rope was modeled with a steel wire (to achieve weight and elasticity) with a silicon tube around it to get the correct drag forces on the mooring line. The mooring lines were connected to the corners of the barge with a pre-tension of 2150 kN. Instrumentation The motions of the barge and the TLD deck were measured with two independent targets containing each three light sources. The position of these light sources is accurately determined with an optical system and the motions in six degrees of freedom were derived at the respective centers of gravity. Relative wave motions were measured inside the moonpool with 3 wave gauges mounted on the barge and one gauge secured to the TLD deck to measure the airgap at all times. Top tensions were measured with ring shaped strain gauges in all mooring lines, risers and tendons. The same transducers were used to measure the axial loads in the TLD cables at both sides of the sheaves. Finally six-component frames were installed below two out of four hinged arms to measure the reaction forces in six degrees of freedom. 3 STATIC TESTS AND NATURAL PERIODS Static offset tests were performed by displacing the barge in surge towards its stern. The displacements of the TLD deck due to the barge offset are shown in Figure 5. The absolute pitch and heave displacements and the relative surge displacements of the TLD deck are shown as a function of the barge excursion. From the figure it is clear that the relative TLD deck surge is small (less than 2% of the barge excursion). The TLD deck lags behind the barge.

Figure 5 TLD deck motions in static offset tests

Figure 6 shows why the TLD pitches down (positive pitch angles) when the barge surges towards aft. The riser flaring causes this pitch displacement. The set-down (heave displacement) is typical of vertically restrained floaters such as TLPs.

Figure 6 TLD deck pitch due to riser flare Decay tests on the barge as well as on the TLD system were performed in calm water. Table 3 shows the barge natural periods in different configurations. Note the rather large natural period in roll due to the presence of skirts near the barge keel. Table 3 Barge natural periods in seconds.

A special set up was built to perform the decay tests on the TLD deck. It ensures that the initial offset for the decay tests is applied in one degree of freedom and that the barge is motionless. Note however that the hinged arms are free to pitch during these decay tests. For each degree of freedom the test is done twice. Time traces of the TLD deck motions are presented in Figure 7.

Figure 7 Decay tests on TLD deck From Figure 7 it is clear that the relative damping for the TLD deck motions is high. Therefore only a few oscillations occur and the standard analysis tools can not be used. To estimate the relative damping the overshoot is manually determined and the following equation is used:

Mooring Set-up Surge Sway Heave Roll Pitch Yaw

free floating barge 14.2 20.6 12.6 moored, no TLD 207 160 14.2 20.6 12.6 85moored with TLD 204 162 14.2 20.6 12.6 89

The large relative damping also prevents an accurate estimation of the natural periods. This is especially true for surge and sway. The resulting estimated relative damping values are shown in Table 4:

4 MOTION RESPONSE Tests were carried out for Brazil conditions with and without current in head seas, bow quartering seas and beam seas. The environmental conditions are shown in Table 5. In this paper the results of the test in bow quartering and in beam environments are discussed. Figure 8 shows a photograph of a test in head environment.

Figure 8 Barge in head environment The barge and TLD deck motions were measured in wind only, current only, waves only and in the full environmental conditions of Table 5. The motion response of the barge shows two different frequency regimes: a low frequency (LF) and a wave frequency (WF) regime. The wave frequency regime is defined as frequency interval associated with first order wave energy. The low frequency regime response occurs at frequencies ? < 0.2 rad/s. In a similar way a high frequency (HF) regime can be defined as the frequencies above the WF regime, in this case ? > 1.0 rad/s. This regime includes the TLD deck natural modes (see Table 4). To study the coupling between the barge and the TLD deck motions, all motion signals were filtered with low pass, band

bass and high pass filters. Standard deviations of all motions in the LF, WF and HF regimes have then been computed. TLD-barge clearances The TLD deck moonpool clearance is 4 meters longitudinally (40m-long moonpool and 32m-long deck) and 8 meters in the transverse directions. The relative horizontal motions of the TLD with respect to the barge are small and did not result in any clearance problems for the deck in this moonpool. Table 6 shows that the horizontal barge motions are mainly in the low frequency range. It is further clear that the relative motions are small thus leaving enough clearance between the deck and the moonpool walls. Finally it is interesting to point out that most of the relative surge and sway occurs at the wave frequency regime. Owing to the combined inertia of the TLD deck, risers and tendons, the deck cannot follow instantly the first order surge/sway barge motions.

Vertical TLD deck motions The absolute vertical TLD deck motions (heave, roll and pitch) induce the load variations on the risers and are analyzed in more detail below. Table 7 shows that the heave motion of the TLD deck occurs mainly in the low frequency range. As the heave motions of the barge itself are mainly in the wave frequency range, this means that quasi-static set-down effects due to the low frequency surge motions are the main contributors to the deck heave motion. The wave frequency TLD deck motion is caused by the inertia force of the pretension weights and by the elasticity of the risers.

For the roll motions of the barge and TLD deck in beam seas, Table 8 shows contributions from both the low frequency and the wave frequency range. From the TLD roll spectrum in Figure 9 a clear distinction between the low frequency and wave frequency contributions can be observed. In contrast, most of the barge roll energy is in the wave frequency region at the roll natural frequency and around the wave spectrum peak frequency. This breakdown in well-separated forced and resonant roll responses is due to the presence of the skirts and has been observed before [6], [7].

From the roll spectral densities of Figure 9 it is clear that the TLD roll response at the barge roll natural period and around the peak spectrum frequency is significantly reduced. The bulk of roll energy is concentrated at the barge sway natural frequency in this beam sea condition. When the barge is pushed to portside, the transverse flaring of the risers pulls the deck down at starboard side.

Figure 10 illustrates the frequency response in pitch of the barge and TLD deck. TLD deck pitch motions appear at the surge natural period of the barge and pitch natural periods of both barge and TLD deck.

Table 9 shows the standard deviation for the different frequency regimes. The barge surge motions induce the low frequency pitch motions of the deck. The longitudinal flaring of the risers is the cause of this pitch motion as was observed in the static tests. The pitch motion of the barge mainly results in variations of the arm angles. However, due to inertia and friction forces, these motions cause tension variations in the cables and therefore excite the TLD deck in pitch as well. The comparison of the pitch motions of the barge (standard deviation = 0.73 deg) with the pitch motions of the deck for this frequency regime (standard deviation = 0.11 deg) shows that the TLD system is experiencing only 15% of the barge motion in pitch. Finally the TLD deck pitch motion around its natural period is induced by the direct wave loading on the risers and tendons; the barge response at these frequencies being very small.

In summary, it was observed that all deck vertical motions are small compared to the corresponding barge motions. Furthermore the main contribution to these TLD motions is low frequency due to the low frequency surge and sway motions of the barge. The wave frequency regime is the second most important contribution to the TLD vertical motions. Barge vertical motions cause the lever arms to pitch thus inducing inertia loads and tension variations in the cables attached to the TLD deck. This is investigated in more details in the next section.

5 LOAD RESPONSE The vertical barge motions relative to the TLD deck induce rotations of the arms with the tensioning weights. For instance when the barge rolls such that the starboard sheaves heave up by one meter, the starboard tensioning weights heave up by two meters. The associated accelerations and thus inertia forces cause tension variations in the cables. In Table 10, the standard deviation of the cable tension is computed on both sides of the sheaves for the bow quartering Brazilian environment. The main contribution to the tension variations clearly comes from the wave frequency regime. The corresponding standard deviations are almost identical to the standard deviation of the raw signal. The second largest contribution corresponds to high frequencies i.e. to the natural response of the TLD deck. The low frequency regime expectedly does not contribute at all to the tension variations, as these motions are quasi-static in essence.

It is interesting to point out the large difference existing between the standard deviation computed on both sides of a sheave (357 kN on the deck side and 270 kN on the weight side for the Starboard Aft (SBA) sheave). This difference can be explained by the inertia of the sheave in rotation. The Dynamic Amplification Factor (DAF) is defined as the ratio of the standard deviation of the tension in the TLD cable to its mean value (equal to the pretension weight). From the above table, this DAF is equal to 5% for the most severe bow-quartering environment. Head and beam environments led to smaller DAF values. Table 11 shows the statistics of the measured tensions in the cables. The mean and standard deviations are added to facilitate comparison. Table 11 Statistics of the measured tensions in the TLD cables

Compared to the tension variations in the TLD cables, the inertia forces associated to the small TLD deck motions can be ignored. Therefore the tension variations in the risers and tendon are comparable to the tension variations in the TLD cables. This is confirmed by the dynamic amplification factors derived from the measured top tensions in the risers (see Table 12).

6 MOONPOOL AGITATION To obtain an impression of the risk of slamming events on the underside of the TLD deck, the relative wave elevation was measured at three locations fixed to the barge. Furthermore one wave-probe was mounted to the center of the TLD deck. The airgap is defined as the amount of space between crest of the waves and bottom of the TLD deck. If the airgap becomes zero this means that the TLD deck is exposed to wave impact loads, which can influence the tensions in the TLD cables, risers and tendons. The maximum relative motions below the TLD deck were observed during the tests in bow quartering condition. For this tests an airgap of 8.5 m was sufficient to prevent slamming. The relative water motions inside the moonpool can be caused by the pumping mode resonance (vertical motion of the mass of water contained in the moonpool) or by sloshing inside the moonpool (horizontal motions). The response of the relative water motions to the incoming wave shows that resonant vertical oscillations occur at a period of 10.8 seconds. From the measured relative wave motions at three locations inside the moonpool it was found that mainly the pumping mode was present. By drawing a plane through these three measurement points, the relative wave motions in the center of the moonpool can be calculated as well and compared to the airgap. Figure 11 shows a spectral comparison for the wave motions relative to the barge and relative to the TLD. Both spectra show a maximum response at 10.8 s. This is the natural frequency of the pumping mode in the moonpool. In addition to this a large response is found for the wave motions relative to the deck. This peak corresponds to the wave frequencies. At this period the relative wave motions with respect to the earth-fixed deck are much larger than with respect to the barge.

7 CALIBRATION/COMPARISON WITH NUMERICAL CALCULATIONS The purpose of this section is to compare the numerical predictions made prior to [8] and during the model test campaign with some salient experimental results. Numerical model The numerical analysis has been performed with the AQWA suite of programs maintained by Century Dynamics (Horsham, UK). Diffraction/radiation calculations have been performed with the mesh of Figure 12 i.e. including

the amidships moon-pool and the longitudinal skirts at the keel level. The calculated coupled natural periods of the barge are favorably compared below with the periods measured during the decay tests in free-floating condition:

Second order drift force Quadratic Transfer Functions (QTF’s) have been computed with the pressure integration method [9] in order to have all six components of the drift load QTF and, in particular, the second order roll moment.

An AQWA model of the FPDSO-TLD system has been constructed. The model includes the following components: • Six-body model (barge, TLD deck and four articulated arms), • 16-line mooring system, • 12m-diameter sheaves modeled as two zero-diameter sheaves set 12m apart, • Articulated beams free to pitch relative to the barge, • Risers and tendons modeled by finite elements with correct mass, weight in air and axial stiffness. Figure 13 hereafter shows a close-up of the moon-pool area.

Static tests with TLD system connected Static tests have been performed with the moored barge with TLD system connected. Figure 14 hereafter illustrates the changes in TLD pitch angle with barge excursion. As mentioned previously, the riser flaring causes these changes. Measured and calculated values agree well.

Similar agreement is found for horizontal forces and for TLD deck vertical set-down.

TLD deck natural periods The measured natural periods of the TLD deck are now compared with the results of numerical decay experiments in Table 14:

The largest discrepancy between measurements and calculations is found for surge and sway. However, consideration of the decay test time series in Figure 7 shows the difficulty in deriving accurately natural periods. Apart from these freedoms, a good agreement is found.

Motion and load response For the sake of brevity, only the design bow-quartering environment (208010, see Table 5) is discussed here. A time domain simulation with low and wave frequency effects has been performed. A time step of 0.05s is used to accurately integrate in time the high frequency TLD deck dynamics. The measured and calculated barge and TLD deck motion responses are compared in Table 15 below:

Discrepancies are found in the low frequency motions with the barge surging more and swaying less than in the basin. A very good agreement is found in the heave response. Measured roll and pitch RMS barge responses are smaller than calculated. This is due to our conservative assumptions regarding viscous damping. The heave response of the TLD deck – the main driver of riser dynamic tensions – is well captured by the numerical model. Horizontal motions of the deck are, expectedly, almost identical to those of the barge. The measured and calculated dynamic tensions in the cables connecting the TLD deck to the tip of the articulated arms are compared in Table 16 below. The right hand side of the table shows the dynamic tensions at the riser top end.

Agreement between measurements and calculations is very good (relative error less than 6% for the cables and less than 8% for the risers).

8 CONCLUSIONS An extensive model test campaign has been performed to demonstrate the feasibility of spreadmooring a new build barge with 12 dry trees in 1,200m water depth offshore Brazil. This test campaign was also aimed at verifying a new configuration of the Tension Leg Deck system with weights (to pretension the risers) stored in dedicated hull compartments. Additional tests were also performed for West African environments. The model tests results confirm that the TLD configuration with weights inside the hull is feasible with a barge fitted with 7m-wide skirts and for environments up to Hs=7.8m bow quartering waves. In this 100-year environment, the DAF (Dynamic Amplification Factor defined as the ratio between the tension standard deviation to its mean) in the TLD cables is only 5%. The test campaign has confirmed (for the second time [6]) the seakeeping performance of a newbuild rectangular barge equipped with longitudinal skirts at keel level. Clearances between the tendons/risers and the moonpool walls near the keel have been verified experimentally in the design environment. Finally, the tests have helped quantity the moonpool agitation levels in four locations thus providing insights and guidance for the choice of the TLD deck height above mean sea level. Results of the model tests (static tests, decay tests, irregular wave tests) have been successfully compared to numerical predictions. In particular, an excellent agreement is found for the DAF in the TLD system i.e. for the dynamic tensions in the risers. The fundamental reason for this excellent agreement (which has been confirmed for the 100-year beam and head environment conditions) is that the DAF can be estimated by the following extremely simple formula [5]: 1/ 3 DAF a g = where 3 / 1 a is the significant single amplitude heave acceleration at sheave location and g is the acceleration of gravity. The heave motions being 1st order and highly damped, it can easily and accurately be estimated even in the frequency domain once the hull shape is defined.

REFERENCES [1] “An FPSO with dry well heads located on a Novel Tension Leg Deck as Stand Alone Deepwater Field Development” by L. Poldervaart, J. Pollack. Proceedings of the Deep Offshore Technology Conference, 19-21 October 1999. Stavanger, Norway. [2] “A Surface Tree Riser Tensioning System for FPSOs” by J. Pollack, K. Poldervaart and M. Naciri. Proceeding of 2000 Offshore Technology Conference. OTC paper #11902. [3] “Tension Leg Deck (TLD) SBM test campaign” Main report MARINTEK, November 1998. [4] “ The Tension Leg Deck, From Drawing Board to Numerical Design Tools” by J. Pollack, M. Naciri and L. Poldervaart. Proceedings of the 2000 Offshore Mechanics and Artic Engineering Conference. Paper # OMAE-00-4001; February 14th –17th, 2000; New Orleans, Louisiana. [5] “A Shallow Draft Dry Completion Unit (DCU) with Flexible Deck Load and Storage Capacity for a Stand Alone Ultra Deepwater Field Facility” by M. Naciri, L. Poldervaart and J. Pollack. Proceedings of the Deep Offshore Technology Conference, 17th –19th October 2001. Rio de Janeiro – Brazil.

[6] “Model testing of a rectangular barge with different skirt configurations” MARIN report No 16562-1-OB December 2000. [7] “ Non-linear low-frequency roll excitation of a rectangular barge” by M. Naciri and N. Lledo. Proceedings of the 2001 Offshore Mechanics and Artic Engineering Conference Paper OMAE- 01-1247; June 4th-8th, 2001; Rio de Janeiro – Brazil. [8] FPDSO-TLD model test JIP – Technical appraisal. Document RD52420-CRM91005-Rev C1 – January 2002. [9] "Low frequency second order wave exciting forces on floating structures" by J.A. Pinkster. PhD thesis, Technical University of Delft, 1980