STABILITY OF TENSION LEG PLATFORMS WITH DAMAGED TENDONS

OBJECTIVE: The primary objective of this project is to investigate the static and dynamic stability of various classes of TLPs under extreme hurricane and Loop current conditions where, for whatever reason, one or more tendons have been lost due to damage or disconnect. The investigation of dynamic stability will be performed using time domain numerical simulations, which will necessarily involve a number of simplifying assumptions and procedural extensions to handle wind and greenwater loading under heeled platform conditions. Consequently, a secondary objective of this project is to develop rational procedures for simulating platform responses under heavily damaged conditions. This study will not attempt to rigorously address the risks of tendon failure nor, for practical reasons, will it attempt to find a basis for direct comparison of the inherent tendon-damaged stability of various TLP concepts.

BACKGROUND: One of the main advantages of the TLP concept is its inherent stability, primarily in the vertical modes (heave, pitch and roll), which makes it most suitable for supporting vertical top-tensioned risers with dry trees, as well as for supporting steel catenary risers to tie in subsea flowlines and export pipelines. The early TLP designs for major hub developments consisted of four columns connected by a ring pontoon at the base and supporting the outer corners of a rectangular deck at the top. Attached to each column was a minimum of three tendons. Later TLP designs for marginal fields and smaller topsides payloads required less hull buoyancy and less total tendon pretension. Similar intact stability as for the early TLP designs was achieved with more efficient distribution of buoyancy and relatively larger tendon footprints, but also with fewer tendons per “corner”. For example,

• the E-TLP hull concept retains the four separate corner columns, but incorporates a radial pontoon extension at each column for supporting two tendon porches at its extremity,
• the MOSES TLP hull concept concentrates the column buoyancy closer to the center of the platform, still with four separate columns of relatively smaller cross-sectional area, but incorporates four large radial hull extensions at its base (tendon support structures), each one supporting two tendon porches at its extremity,
• the SeaStar hull concept concentrates the column buoyancy in a single column at the center of the platform and incorporates three radial hull extensions at its base, each one supporting two tendon porches at its extremity.

The current practice for design of TLPs for the Gulf of Mexico, as documented in the API Recommended Practice for Design of Tension Leg Platforms (RP 2T), does not distinguish between the various classes of TLPs. All design analysis procedures and safety factors are equally applicable to all concepts. Questions have been raised as to whether the design practice should recognize what appears to be inherently different levels of stability for damaged tendon conditions between the various classes of TLPs. The underlying concern is whether the design practice ensures that all classes of TLPs will be designed with adequate robustness to guard against total loss of the platform in the unlikely event of tendon failure due to overload.

The analysis of TLP platform response under damaged tendon conditions is exceedingly difficult for practical cases of interest. This is in part because of the large number of possible damage scenarios, including effects of inter-dependent progressive component failures. Another major source of difficulty lies in the modeling of highly nonlinear phenomena typically in play during progressive failure situations. This applies to analysis using either numerical or experimental modeling techniques. In the present context, as the platform loses tendon restraints and progressively heels over, phenomena such as bottom stroking and snap loading in remaining intact tendons, variable projected deck areas exposed to wind and wave loading, large platform yaw rotations and out-of-plane loading on catenary risers, snagging and fouling effects from damaged components, etc. can be difficult to model in the general case. The designer strives to achieve a structural system that is well behaved and straightforward to analyze in the intact condition so that its performance can be reliably predicted. Mother Nature is generally not so kind when she decides to take it apart.

Because of these difficulties, it is not common for TLP designers to perform dynamic simulations of tendon missing or failure scenarios in extreme (100-year return period and beyond) environmental conditions. Apart from the relatively straightforward simulations of a single tendon removed (for planned maintenance or replacement) with ballast compensation under exposure to moderate strength hurricane sea states or extreme winter storm conditions, we are not aware of any prior studies in the public domain that have addressed dynamic stability of TLPs with damaged tendons.

While the dynamic analysis of progressive tendon failure scenarios can be considered a daunting task, technology for numerically simulating nonlinear platform loading and responses has continued to steadily improve with time. Improvements in computer hardware, better understanding of the mechanics of fluid-structure interaction, and more robust software for vessel-mooring-riser coupled dynamic analysis of floating systems subject to large amplitude motions make feasible a much more in-depth level of analysis than would have been possible, say, 15 -20 years ago when TLP design analysis procedures were first being established. This study will employ well-validated software packages such as StabCAD for static stability analysis, WAMIT for hydrodynamic loading, and WINPOST for coupled time domain platform/tendon response calculations.

APPROACH: The stability of an intact TLP is mainly provided by its tendon system. TLPs are very stiff in pitch and roll, which means that they develop very large righting moments to counteract overturning moments generated by environmental forces. When tendons are removed from the system (either due to damage or deliberate removal), the burden of developing stabilizing righting moments shifts from the tendon system to hydrostatic stiffness of the hull itself, specifically, through the pendulum effect due to the separation of the centers of gravity and buoyancy, and through the waterplane effect associated with the surface piercing columns. With continued loss of tendons, at some point the hydrostatic stability of the hull itself becomes critical to the ultimate survival of the platform.

For purposes of this study, we will define and consider four classes of TLPs from a tendon/hull hydrostatic stability perspective:

1. Conventional TLP – 4 corner columns with a ring pontoon, 3 tendons per corner
2. E-TLP – 4 corner columns with a ring pontoon, 2 tendons per corner extension
3. MOSES TLP – 4 inner columns with 4 tendon support structures, 2 tendons per support structure
4. SeaStar TLP – 1 central column with 3 tendon support structures, 2 tendons per support structure.

Static stability analyses will be performed for various tendon damaged conditions, with and without ballast compensation, for all four classes of TLPs. Dynamic stability analyses, which are inherently much more time-consuming, will be performed for three classes of TLPs (excluding the E-TLP class).

The static stability analysis will be performed using the commercial software StabCAD. Standard righting moment curves will be developed for the TLP in the intact, one tendon missing and two tendon missing cases, where the missing tendons are removed from the upweather corner. In the case of the conventional TLP the stability analysis will also be performed for the three tendon missing case (i.e. all tendons missing from the upweather corner). For the damaged cases, the stability analysis will be performed relative to a static equilibrium configuration that reflects a balance between the mean environmental loads associated with the storm condition under consideration and the restoring forces providing by the remaining intact tendons and the hydrostatics of the hull. Static stability analyses will be performed for 100-year and 1000-year return period hurricane seastates, and for 100-year loop current conditions, as defined by API’s proposed new metocean criteria currently under review by the industry.

The dynamic stability analysis will be performed through time domain simulations using the in-house software WINPOST and the commercial software WAMIT. Program WINPOST has been verified against numerous experimental and field reports and used extensively on OTRC projects funded by industry and government sponsors.

The dynamic simulations will require a substantial amount of preparation and set-up. It will first be necessary to determine the equilibrium configuration of the TLP associated with each tendon damaged condition and seastate. Given this information, the hydrodynamic coefficients (first-order exciting forces, added mass, radiation damping, second-order drift forces) will next be determined using WAMIT. Simple, but rational models for the wind and current loading under heeled platform conditions will be devised and applied to supply the needed coefficients. A rational model for deck loading associated with immersion in waves will also need to be devised. With all the force coefficients in hand, the WINPOST simulations can be initiated.

A number of assumptions will be made to simplify the analyses:

• all catenary risers will be removed and their force effect at the attachment points will be replaced by an equivalent ballast weight
• regardless of whether tendon failure is indicated by overload or loss of tension leading to disconnect, tendon damage will be represented by completely and instantaneously removing the tendon from the system at the time of failure (as if it had parted at the tendon porch)
• loads associated with immersion of a flare boom as the platform heels over will not be considered.

In this way we believe we can achieve a balance between retaining an appropriate level of detail so that meaningful results are generated and compromising the validity of the results through excessive use of questionable modeling practices.

DEPLOYMENT OF RESULTS: The results obtained in this study will be communicated to TLP designers, owners and regulatory agencies through conference presentations and publications, and through dissemination of project reports and student theses through normal OTRC and MMS channels. The results may provide insights that lead to new guidelines for the design of TLPs to be deployed in the Gulf of Mexico.

PROJECT PLAN FOR 2007 - 2008: The project will be configured in 7 separate tasks, described below:

Task 1: Definition of TLP Particulars
In order to perform the static and dynamic stability analyses described above, it will be necessary to define the various TLP cases in terms of their

• hull geometry
• deck dimensions
• tendon locations, sizes, weights and pretensions
• top tensioned riser locations, sizes, weights and pretensions
• platform mass, center of gravity and moments of inertia

It is not at all clear how to “normalize” existing designs for the various classes of TLPs in order to facilitate a direct comparison of performance with damaged tendons, nor is it clear how to generate new designs for the four classes of TLPs that are in some way normalized to a common basis. Therefore we will not attempt to generate new designs. Instead we will review the publicly available information on existing TLPs and request needed additional information from designers and operators in order to come up with four candidate TLPs for which we have the most information at our disposal. Where necessary we will estimate particulars, based on engineering judgment, for which we do not have a dependable source. As discussed above, the need to simplify the analysis by neglecting the effects of catenary risers will already require compensating adjustments to mass properties. In this way we will ultimately arrive at four TLP designs that are representative of each class, modeled to the extent possible after existing TLPs, yet are not precise replicas of existing TLPs.

Task 2: Definition of Metocean Conditions
Existing Gulf of Mexico TLPs have been designed to metocean criteria that are now considered insufficient for future designs. Proposed new metocean criteria for the Gulf of Mexico are defined on the basis of location. For purposes of this study we will adopt the proposed API metocean criteria for the most severe region in the Gulf. As the new API recommended practice documenting the metocean criteria is currently under review by the industry and subject to change, at the time this task is scheduled to begin we will interrogate the current version of the API document and use it to fix the metocean criteria for 100- and 1000-year return period events for the remainder of this study.

Task 3: Assembly of StabCAD, WAMIT and WINPOST Models for TLPs with Intact Tendons
Numerical models for all TLPs to be analyzed will be assembled in order to ensure that all needed information is available for the remainder of the study. The various models will be reviewed for internal consistency and documented.

Task 4: Static Stability Analyses – Intact and Damaged Conditions
Static stability analyses will be conducted for all four TLPs for the following conditions:

• intact tendon system, one, two and three (for conventional TLP) upweather tendons missing
• calm water, 100-year Loop current, 100-year hurricane and 1,000-year hurricane
• for missing tendon cases, without and with ballast compensation.

For the last item, the cases with ballast compensation will serve as a reference to quantify the incremental loss of stability experienced in actual damaged conditions where it may not be possible to shift ballast in order to mitigate the risk of further adverse effects. It is recognized that, for real situations, there will be a limit on how much ballast can be shifted but, again, these cases are just for reference purposes. Since the StabCAD simulations will be relatively simple to execute, it is feasible to execute a larger number of cases in order to generate additional information that may add insight.

Task 5: Dynamic Simulations – Intact and Damaged Conditions
Dynamic simulations will be conducted for three TLP configurations (conventional, MOSES and SeaStar) for the following conditions:

• intact tendon system, one, two and three (for conventional TLP) upweather tendons missing
• 100-year Loop current, 100-year hurricane and 1,000-year hurricane
• no ballast compensation for missing tendon cases.

For the three seastate conditions, separate simulations will be performed assuming two different tendon failure scenarios:

• starting with an intact tendon system, failure of a single upwave tendon occurs at a random time not related to overload or loss of tension (due perhaps to fatigue, for example), with subsequent tendon failures triggered by overload or loss of tension
• starting with an intact system, failure of a tendon does not occur unless an overload or loss of tension situation is detected.

Since it is possible that, following failure of the first tendon, subsequent tendon failures are triggered by the induced transient response from the first tendon failure, it will be important to properly simulate the transient response mechanism.

Following a tendon failure event, the simulation will be continued until the next tendon failure is triggered or for a maximum of 3 more hours, should the platform reach and remain in a steady equilibrium configuration (albeit damaged). The 3-hour duration should be sufficient to allow determination of the probability distribution of individual responses of interest. This information can then be used to assess the probability of subsequent tendon overload or loss of tension, assuming the seastate remains constant.

In cases where the simulated tendon tension responses indicate an overload or disconnect due to loss of tension, the simulation will be repeated with an instruction to instantaneously fail (i.e. remove) the affected tendon at the appointed time, and the transient responses of the TLP will be simulated, again checking for overload and loss of tension conditions.

Task 6: Review and Interpretation of Results
The results of the static and dynamic analyses will be reviewed with the objective of distilling general findings that may be used to inform TLP designers, owners and regulatory agencies. To the extent that the TLP designs and associated analyses will not have been configured to allow a precise one-to-one comparison between classes, it is recognized that some conclusions may be need to be based on subjective judgment in interpretation of results.

Some of the target areas to be investigated are:

• the general configurations that the TLP will adopt as upweather tendons progressively fail (platform heel angles, mean tendon tensions, platform offset),
• the extent to which the stability of the various classes of TLPs under damaged tendon conditions can be characterized as significantly different,
• the expected time to reach the next tendon failure (overload or disconnect) assuming the seastate remains constant,
• the degree to which conventional static stability analysis applied to damaged tendon configurations in extreme seastates can be used to gage platform survivability or robustness against total loss.

Task 7: Final Report Preparation
A final report will be prepared that documents all assumptions, methods and procedures used, the TLP configurations analyzed and the results generated as the basis for the conclusions that are drawn.

PRINCIPAL INVESTIGATOR(S) & OTHERS INVOLVED IN PROJECT:

Principal Investigators: Drs. Moo-Hyun Kim, Jun Zhang, and Richard S. Mercier

Support Staff: Two graduate students to be determined