Вы находитесь на странице: 1из 15

OTC 15168 Guidelines for Compression Modeling in Flexible Risers for Deepwater Applications

D. McCann/MCS, F. Smith/MCS, P. OBrien/MCS


Copyright 2003, Offshore Technology Conference This paper was prepared for presentation at the 2003 Offshore Technology Conference held in Houston, Texas U.S.A., 58 May 2003. This paper was selected for presentation by the 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.

significant, compression forces can arise at the touchdown point that, when large, can cause global buckling of the pipe section significantly reducing the dynamic pipe minimum bend radius. Thus, if compressive response arises, global buckling of the pipe cross section must be verified. In general, the catenary configuration will not experience compression unless the vertical motion of the hangoff is severe. This is not a problem that normally affects traditional Gulf of Mexico SPAR or TLP deepwater development vessels as they do not have significant heave motions. A Floating Production Storage and Offloading Vessel (FPSO) has a considerably larger heave response. In addition, the vertical motion is a combination of both the heave and pitch. Therefore the hangoff motion is a function of distance from the center of motion. Figures 1 and 2 comprise Response Amplitude Operators (RAOs) of vertical motion and acceleration at the hangoff point of an FPSO respectively. The risers are assumed to hangoff at distances of 0m, 50m 100m, 150m and 200m horizontally along the vessel from the motion center. As the hangoff point moves away from the center of motion, pitch amplified heave motion increases considerably. Therefore, catenary risers attached to FPSO's, in particular those where the riser hangoff are located away for the vessel motion center, are more likely to experience of compressive loading. The majority of the work detailed herein is performed on two generic flexible risers, namely a 6 production and 4-1/2 water injection riser. An extensive amount of analyses including sensitivities to element length, analysis timestep, bending stiffness and drag coefficient have been performed in an attempt to establish trends. These are then used to produce modeling guidelines when effective compression loading is observed at the touchdown point region.

Abstract There is not currently in industry a consensus for modeling compression in flexible risers, particularly for deepwater applications where compression is a key design driver. This paper gives guidelines on the particular type of behavior experienced by risers under compression and on the requirements for accurate modeling of this behavior. For example, the assessment of post buckling curvature requires a full understanding of the events occurring and it is only with tight guidelines on modeling this issue that accurate, consistent solutions can be delivered. At present one of the critical design drivers for catenary risers, either flexible or metallic, is compression of the riser at the touchdown location and the associated buckling. An accurate assessment of this behavior is fundamental to the design process. The guidelines developed and described in this paper will help in galvanizing the understanding of this behavior and on how to accurately predict the response when using a global finite element modeling technique. 1.0 Introduction During the 1990s and up to the present day, flexible risers have been installed in a number of deepwater developments offshore Brazil, in the Gulf of Mexico and elsewhere around the world. Where individual risers are selected as economic over a hybrid design, it is preferred to configure the flexible pipes as simple free-hanging catenaries. This configuration represents both the simplest and cheapest option in terms of installation and procurement in deepwater. The catenary configuration couples the mean, low frequency offset and first order motion of the vessel to touchdown point response. Where vertical motion of the hangoff point is

2.0 Compressive Failure Mode of Flexible Risers A flexible pipe may be subject to two types of compression, namely axial (or true wall) compression and effective compression (negative effective tension). Axial compression can cause birdcaging of the tensile armor layer that can result in pipe failure with rupture of the external polymer sheath.

OTC 15168

Effective compression causes the global stability of the pipe to be significantly reduced consequently resulting in significant deformation of the pipe through global buckling. A snapshot of a numerical riser model under buckling is illustrated in Figure 3. The latter is the focus of this study. Buckling of flexible pipe is not a new issue. Typically, shallow water Lazy-S systems experience certain level of compressive buckling at the touchdown region. However, response is a function of the catenary length. In shallow water systems, the length are sufficiently small that local stiffening of the cross section can alleviate problems should they occur. In deeper waters, catenary length are considerably longer and make it more difficult to alleviate compressive over bending should it occur. This will be examined in more detail later in the paper. 3.0 Existing Guidelines on Modeling Buckling A review was conducted of the recommended practice and code documents that provide guidance for numerical analysis of offshore structures. These include API RP 17B [1], DNV OSF201 [2] and API RP 2RD [3]. API RP 17B Recommended Practice for Flexible Pipe gives guidance on criteria to use when designing pipes that are subject to compressive loading; clause 5.4.9.2 states that the effective compression should be less than which would cause MBR criteria to be violated and that Bar (Euler) buckling should not occur. Clause 5.4.9.3 states that The maximum axial compression should be calculated as the value which causes a gap between the tensile armor wires and the underlying layer equal to half the thickness of the armor wire. It also states that The allowable axial compression should be calculated from the maximum axial compression using a safety factor not less than 1.0. Clause 8.2.3.1.4 says that Very large changes in the riser configuration require a nonlinear solution procedure. For dynamically sensitive structures, nonlinear time domain solutions are required. API RP 2RD Design of Risers for Floating Production Systems (FPSs) and Tension Leg Platforms (TLPs) talks in general about finite element discretization in clause 6.4.3.1. The riser must be discretized carefully to avoid numerical errors and The level of discretization that is ultimately acceptable depends on the numerical representation of tension variation, the special variation in physical properties of the riser and in the magnitude of applied load, frequency content of the applied load and the accuracy of the desired results. It also suggests that near a boundary, element length should not exceed the square root of the bending stiffness divided by the corresponding effective tension [3]. DNV-OS-F201 Dynamic Risers, Section 5 discusses global buckling but does not address analysis techniques and considerations. One of the difficulties here is that a large variation in post buckling behavior can arise depending on how the problem is numerically modeled and these can lead to difficulties in comparing against a fixed MBR criterion. The guidelines

currently used do not sufficiently address compression modeling, buckling and post buckling behavior of risers. This is something that requires further research. This paper aims to address some of the issues that are important. 4.0 Theoretical Representation of Buckling Inherently, buckling is a type of stability problem. Much of the theory of buckling has developed from the static analysis of structures, primarily from a structural engineering viewpoint. This theory also forms the background to many of the numerical methods and packages that are used to examine buckling behavior. Considering a numerical example, an idealized model of a straight vertical strut is illustrated in Figure 4. The strut has a length L and is assumed to have a linear bending stiffness EI. When the applied load is small, the column remains straight and subject to only axial compression. By taking a sectional view and resolving the individual force and moment components, the axial bar compression is related to the bending and deflection as follows [4]: 2 [1] EI d v 2 = M = P.v dx Assuming k 2 = P , Equation 1 may be re-written in the EI form: v '' + k 2 v = 0 [2] The general solution to this equation has the form : v = C1 Sinkx + C 2 Coskx . [3] By applying appropriate boundary conditions, the critical buckling load (or Euler buckling load) is determined. [4] Pcr = 2 EI 2 L In terms of structural stability, below Pcr the strut is stable and lateral deflections are small under lateral loads. When Pcr is reached, the column is in neutral equilibrium. In this state, small changes in lateral deformation may be achieved without affecting the applied force. For loads in excess of Pcr the deflections tend to infinity and the strut fails. This point is conveyed in Figure 5. Fundamentally, Equation 4 identifies that the buckling load is directly proportional to the bending stiffness and inversely proportional to the square of the effective length. It is noted that this only applied where the bending stiffness is linear. The implication of this can be directly applies to riser behavior. For shorter riser lengths, such as the lower catenary of a shallow water Lazy-S system, an increase in the bending stiffness can cause a considerable increase in the compressive capacity of the pipe. As the water depth become considerably deeper, the influence of potential increases in EI are over shadowed by the effective length of the catenary and it is difficult to gain a considerable benefit from purely increasing the bending stiffness. Ideally, in design, it is preferred to avoid buckling. In cases where this is not possible, it is preferable to utilize a largest bending stiffness to increase PCR and limit the amount of deflection.

OTC 15168

Numerical Modeling of Buckling There are a number of structural analysis packages available for the analysis of compliant offshore structures, such as marine risers. Many of these use finite element solution schemes and are based on beam bending equations that have Equation 1 as a cornerstone. As a result, deflection in the element will increase with increasing compressive load until the theoretical buckling load is reached. At the buckling load, the deflection tends towards infinity (Figure 5) and the solution crashes. In reality, although the structure deforms at the buckling load, under postbuckling behavior, the structure can accommodated increasing level of load as it become increasingly distorted. This poses some difficulty in modeling problems experiencing significant compression. In reality, when buckling occurs the structure exhibits large displacements. Therefore, the numerical solution is also required to account for the post-buckling behavior in addition to the correctly identifying the buckling load. The program used by the authors for dynamic riser analysis is Flexcom-3D. We shall now look at how this program handles buckling. Buckling in Flexcom -3D Flexom-3D is a general purpose non linear 3D finite element analysis package for the analysis of compliant offshore structures, and is developed by MCS. The program uses a hybrid beam-column element which solves for fully coupled bending, axial and torsional effects. In the program, the structure is discretized using a series of inter connecting beam elements. The elements are C1 continuous, i.e. both deflection and slope at the nodes are continuous across the elements. The program uses a Convected Coordinate System approach (CCS) to calculate and track the large displacement and rotations that typically occur under dynamic motions. Under the convected coordinate approach, each element is assigned a local element coordinate system. Deformation of the element is calculated relative to the local element axis before being applied transferred back to the global equations of motion where the equations are solved. The behavior of the structure is characterized by a linear stiffness and a geometric stiffness. The linear stiffness, KLINEAR, controls the behavior within the bounds of each element. The geometric stiffness, KGEOMETRIC, controls how the elements deform relative to one another. It is the combination of these stiffness that permit Flexcom-3D to model post buckling behavior. Under normal circumstances, both the deformations and relative element displacement are small. In such cases the KLINEAR dominates the behavior of KGEOMETRIC. However, in a situation with large deformations, KGEOMETRIC become large and dominates over KLINEAR.

There is one particular point of interest that follows from this. As previously outlined, when the displacement is based on the beam bending equations with linear bend stiffness, deflections tend to infinity and the numerical solution becomes unstable when the load approached the theoretical Euler buckling load, Equation [4]. Therefore, when the effective compression in an element achieves the theoretical element compression limit, the solution becomes numerically unstable and will terminate. The corollary from this is to reduce the element lengths sufficiently to ensure that each element can with stand the maximum compressive load expected. Although it is common practice to reduce element length, this approach allows a good estimate of an appropriate element length to be made once the magnitude of the compression is known. Flexcom-3D and Static Buckling A previous study has shown that Flexcom-3D can accurately predict both the onset of buckling instability and post buckling behavior, provided sensible values of element length and timestep are used, and provided the Euler buckling load within each individual element is not exceeded. A simplified structure, shown in Figure 6, is used to demonstrate this. This is a cylindrical slender column subjected to a vertical compressive load of 20kN. A small lateral load of 20N is applied to induce lateral displacement in the solution. The theoretical Euler buckling limit for the strut is 17.66kN. The effect of a range of parameters was examined on the Flexcom-3D response. Special consideration was given to both the selected element length and timestep. Element lengths were varied in a series of analyses varying from 1m to 0.1m in length. Additionally, 3 timestep sizes of 0.01s, 0.005s and 0.001s were examined. In all cases a static analysis from 0 to 20s was run, with both the horizontal and vertical loads ramped on over the full simulation length. For all of these combinations Flexcom-3D correctly predicted the onset of buckling instability. Also, all of these analyses ran to successful conclusion although the column is in effective compression throughout. It is important to note though that the Euler buckling load within an individual element is never exceeded. For the maximum element length of 1m, this critical load is approximately 0.635MN; for an element of length 0.1m, the load is 6.35MN. For the model with 60 elements, a further analysis with both horizontal and vertical loads doubled in magnitude was performed; this means the maximum compressive load is 40kN, or about 2.25 time the critical buckling load. The intention was to examine how accurately Flexcom-3D can model postbuckling behavior. A verification analysis was performed with the general-purpose finite element program ANSYS. The Flexcom-3D and ANSYS results are compared. Figure 7 shows superposed snapshots of the column for increasing load magnitudes up to the maximum of 40kN. The corresponding ANSYS snapshot are detailed in Figures 8, 9 and 10. A comparison of calculated displacement at the top node is shown in Figure 11. Results from the two programs show complete

OTC 15168

agreement when predicting the buckling load and post buckling displacement behavior. The conclusion from this study is that Flexcom-3D can accurately predict both the onset of buckling instability and post-buckling behavior. 5.0 Dynamic Buckling In the previous section, the ability of Flexcom3D to accurately account for post buckling behavior was demonstrated for a simple static case. In this section, the focus is moved to buckling in dynamic problems, and in particular for deepwater catenary risers. Dynamic problems are considerably more complex than that statics for a number of reasons. Firstly, the structure is no longer confined to buckle in a single predefined direction. Secondly, the loading on the structure, both from the applied vessel response and the hydrodynamic sources is continually changing. Finally, as the structure deforms by a considerable amount between analytical iterations, significant reduction in timesteps are required to consistently and accurately account for the response. The computation time required for this is expensive. Consequently, irregular sea analyses are sometimes replaced in favor of deterministic regular wave analyses. The major benefit with the regular wave analyses being the quicker solution time. The objective of this paper is to try to identify parameters that govern the buckling response at the touchdown point and use this information to form design guidelines when buckling is an issue. Buckle Initiation To begin, it is useful to consider the nature of the response when buckling occurs. Buckling in catenaries is driven by the prescribed hangoff motions of the vessel, in particular heave. During downward heave, the pipe attempt to follow the applied vessel motion. As the pipe displaces through the water column, drag forces act against the direction of motion. The effect of drag force is to retard the motion of the pipe. Consequentially, vessel heave is translated into a reduction in effective tension. Effective tension essentially acts to prevent bending. When the tension is low or the pipe enters compression, the pipe relies on the cross sectional bending stiffness to limit curvature. The pipe bending stiffness is arbitrarily small in comparison with the distances and forces involved in deepwater. Consequently, buckling occurs. Reviewing the train of events described above, it is obvious that buckling is sensitive to applied vessel motions/accelerations, drag loading, apparent pipe weight, catenary length, water depth and bending stiffness. In order to establish a relationship between the parameters it was decided to conduct a number of analyses on both the 4.5 and 6 risers. A snapshot of the riser is shown in Figure 12.

Both risers are hung off the turret at a mean vertical departure angle of 7. The 4.5 riser is assumed water filled (1025kg/m3) and the 6 Production is filled with oil of density 850kg/m3. The drag to apparent weight ratios for the 4.5" and 6" riser are 5.95 and 5.66m2/tonne respectively. In total 6 individual wave heights were examined in 3m bands, namely 3m, 6m, 9m, 12m, 15m and 18m. For each wave height a minimum, mean and maximum wave period were calculated based on characteristic range of wave steepness [6]. Waves were applied from both the Near and Far directions. A Near wave causes the vessel to offset towards the riser base and a Far wave causes displacement away from the riser base. As the vessel is turret moored, the vessel weathervanes to head in the direction of applied environment. A normal drag coefficient of 0.9 is assumed throughout and a deterministic approach based on regular waves is used. A summary of the results for the 4.5 riser are presented in Table 1. For seastates of 9m or below, the hangoff motions do not induce compression and buckling is not an issue. As the seastate become larger, tensions become increasingly more negative and curvature become larger. The near offset case enters compression first. This is expected, as there is lower tension due to the shorter catenary length involved. Figures 13 and 14 show the distribution of both maximum reported curvature and minimum effective tension with maximum vertical hangoff velocity for both 4.5 and 6 pipes in both Near and Far environmental directions. In all cases an increase in curvature and a reduction in tension is noted with increasing hangoff velocity. The results presented thus far confirm that the hangoff motion primarily drives the response. However, no assumption can be made about either bending stiffness or applied drag coefficient. For this reason, further sets of sensitivities are conducted to examine these effects. Sensitivity to Pipe Bending Stiffness A sensitivity is performed to investigate the effect of changing the pipe bending stiffness by +/-20%. The base case analysis assumed is 15m wave height with a wave period of 14.33s with a Near environment. As the pipe become stiffer, it is expected that the curvature should reduce and vice versa. This trend is followed for the 4.5 pipe, however, the 6 pipe show a reduction in reported curvature with both increased and decreased curvature levels (Table 2). This result caused some surprise. More detailed examination revealed that this was related to the way in which the pipe deforms. Buckling is very much dependant on the hangoff motion, but also on the deformed shape of the riser as the tension reduces. As the EI is uniform in all directions, the pipe will buckle in the plane in which it has the largest deformation. The change in bending stiffness caused the mode of buckling to change and this induce the curvature change.

OTC 15168

First and foremost this indicates that the buckling behavior is sensitive to bending stiffness. However, more fundamentally, it highlights that apart from being a rate dependant problem, the way in which buckling progresses is most strongly influenced by the deformed position of the risers as compression arises. Sensitivity to Normal Drag Coefficient Sensitivity is performed to the effect of normal drag coefficient. Once again these analyses are performed on the 15m Near environment case with a 14.33s wave period. A base case drag coefficient of 0.9 is used. Sensitivity to normal drag of 0.7 and 1.2 is performed and results are summarized in Table 3. Both the level of compression observed and the resulting curvatures vary greatly with the change in normal drag. Again, we can relate these results to a change in the initiation of the buckling behavior. As in the case of the bending stiffness, it is not possible to make any general conclusion based solely on the basis of the drag coefficient to be used. The response to drag coefficient does indicate that drag load is critical in determining post-buckling behavior. A recommendation from this would be to conduct sensitivities on the normal drag coefficient to ensure that the level of buckling response is conservative. A more convenient way of achieving this may be to use a normal drag formulation that varies with Reynold number as Keuglan-Carpenter numbers will be small for the depth typically considered for catenary solution (typically >500m). The preceding sensitivities help to clarify that buckling behavior is a function of the instantaneous deformed shape of the structure as compression occurs. For this reason, it is not possible to form any concrete guidelines on parameters that control the magnitude of bending, such as bending stiffness or drag coefficient, after the negative tension has occurred. Therefore, we must limit our investigation more so to the cause of the problem, the drag force on the structure. Revised Parameter for Examining Buckling Although buckling is attributed to vessel motion, we have not isolated whether velocity or acceleration drive the touchdown point response. It is intuitive that if the response were applied infinitely slow, buckling would not occur. Nondimensionalized plots of hangoff velocity, acceleration and the corresponding curvature at the touchdown point, for a particular case, are shown in Figure 15. In these plots a negative velocity represents that the hangoff is heaving down. Examining the curvature distribution, it is observed that maximum curvature lags behind the maximum downward velocity of the vessel. This represents the time at which the vessel is horizontal about to heave downward. The maximum curvature occurs before the maximum hangoff acceleration. Therefore, it is hangoff velocity, not acceleration that drives the buckling at the touchdown point. Buckling primarily stems from the inability of the riser pipe to travel through the surrounding water at sufficient speed. This is caused by the drag force acting against the direction of motion.

An analogy can be drawn between this type of problem and the problem of a grain of sand falling in a fluid. Initially the sand particle accelerates until the action of the gravitational acceleration. As the velocity increases, the restoring drag force increases until it cancels out the acceleration and the particle reaches the terminal velocity. At the terminal velocity, the acceleration of the particle is balanced by the restoring drag force; hence: Gravitational Force = Drag Force

Wa = m.g =

C d . .Ddrag .VTer min al

[5]

VTer min al = 2.m.g

Cd . .Ddrag

[6]

Where: m = mass, g = gravity, Cd = normal drag coefficient, Ddrag = Drag Diameter, r = Water Density, Vterminal = Terminal Velocity This may be directly attributed to the flexible pipe cross section (Figure 16), where mg is the apparent weight of the pipe including internal fluid. Thus, an equivalent terminal velocity for the riser may be calculated. It is not possible for the pipe to fall faster than the terminal velocity. Ignoring any deformation of the pipe, the hangoff velocity can be directly compared with the terminal velocity. Alternatively, a non-dimension parameter of VHangoff/VTerminal can be used to indicate where buckling may be a problem. Buckling is expected where the parameter is 1.0 or greater. Tables 4 and 5 summarize the maximum compression, maximum curvature and velocity parameters for the 4.5 and 6 risers respectively. Figures 17 and 18 show the distribution of tension and curvature with the velocity parameter. Tension is negative for all cases where VHangoff/VTerminal is greater than 1. For the Near case, some cases have negative tension where VHangoff/VTerminal is less than 1. This is most likely attributed to the lower tension distribution along the riser due to its shorter distance to touchdown. However, it is noted that the compressive tensions in these cases are low and not driving. A more fundamental observation is that the maximum value of curvature and tension occur when VHangoff/VTerminal is largest. Both the 4.5 Water Injection riser and the 6 Production riser have similar drag to apparent weight ratios. To test the assumption further, a third pipe size was examined, namely a 10 production riser with a mass per unit length of 200kg/m, internal diameter of 0.254m and an outer diameter of 0.3717m. The drag to apparent weight ratio for this pipe is 2.87m2/tonne. Results from the analysis are summarized in Table 6. Once again, the most onerous curvature and compression occurs when the velocity parameter is largest. This lends confidence that the velocity parameter may be generally used to identify critical buckling cases.

OTC 15168

Based on the preceding results, the non-dimensional parameter is a useful measure of buckling response. Although response changes depending on the riser catenary length and static tension, in the unmodified form, this parameter comprises a simple yet effective indicator of where buckling response is most onerous. The non-dimensional velocity parameter can be used in two ways. Firstly, for a given set of vessel RAOs, it may be used to determine whether a riser is likely to experience compression. Secondly, if compression is expected, then the parameter may be screened to identify the period at which most onerous buckling may occur. Sensitivity to Element Length and Timestep The final set of sensitivities conducted focused on the effect of both element length and timestep on the results from the numerical analysis. Time step. The basecase model has an element length of 0.5m in the region of the touchdown point. A series of analyses were conducted with timestep sizes of 0.04s, 0.03s, 0.02s, 0.01s, 0.0075s and 0.005s. Regular wave analysis was used throughout. A large variation in resulting curvature was observed with changing timestep. This is due to very slight variations in the postbuckling behavior of the pipe. The sensitivity indicates that using a deterministic approach in selecting the most onerous curvature could lead to large errors. Following this analysis, plots of standard deviation of curvature were prepared. A plot of the curvature distribution at the touchdown point is shown in Figure 19. The standard deviations are more uniform than the corresponding curvature distribution. Therefore the use of a statistical extrapolation based on the curvature deviation provides a more comprehensive and robust method of deriving the maximum curvature. A closer agreement is noted between corresponding results as the timestep becomes smaller. It is suggested that sensitivities be performed until the difference in standard deviation is considered acceptably small between cases. Element Length Two additional basecase models were prepared with element lengths of 1.63m and 2.35m in the touchdown region. All analyses were run using a timestep of 0.01s using a regular wave approach. A plot of the touchdown curvatures are illustrated in Figure 20 and the corresponding curvature deviation is shown in Figure 21. Again, a noticeable difference in resulting curvature is observed with increasing element length, which again reinforces that a deterministic approach to buckling may result in incorrect curvatures. Differences in the curvature deviation plots are also observed. To ensure confidence in the derived curvature, it is suggested that the element length be reduced until the difference in curvature deviation between meshes is considered acceptably small.

6.0 Guidelines and Conclusions In summary, it is demonstrated that the vertical hang off motions and in particular the vertical velocity drives the riser touchdown point response. Where vertical motions at the hang off point are large, compression forces can arise at the touchdown point that can cause global buckling of the pipe section. This behavior is most significant on vessels that exhibit large heave and pitch induced heave motions. This is characteristic of catenary risers attached to FPSO's. In particular, compressive buckling becomes more severe with increasing lever arm from the vessel motion center and is most onerous for a bow mounted turret design. The overwhelming conclusion from the preceding discussions is that post buckling behavior is particularly difficult to capture. The authors have highlighted that many parameters affect the global buckling response of the riser and these may cause a large variation in the post buckling response. However, there are some broad guidelines that may be formed on the basis of the work contained herein: 1. The non-dimensional velocity parameter (VHangoff/VTerminal) is both a simple and useful measure in assessing the likelihood of compressive buckling for a range of hang off responses. Where the parameter is greater than 1, buckling is expected. In addition, larger magnitudes are associated with more onerous riser curvatures. The nondimensional velocity parameter can be used in two ways. Firstly, for a given set of vessel RAOs, it may be used to determine whether a riser is likely to experience compression. Secondly, if compression is expected, then the parameter may be screened to identify the period at which most onerous buckling response may occur. 2. Large variation between the results from deterministic analysis may exist. Analysis is particularly sensitive to both the time step and element length. It is recommended that a stochastic method be employed to extrapolate extreme curvature based on the standard deviation of curvature in the touchdown region. This process applies equally to both regular airy wave analysis and irregular sea analysis. 3. Element length sensitivities should be performed. The element length should be reduced until variations between the standard deviations of curvatures are considered acceptably small. 4. Element length should be selected sufficiently small that the buckling capacity exceeds the maximum expected compressive force. Where the stiffness of the element is linear, non-convergence will occur when the compressive force exceeds the maximum compressive element force, based on the element length and stiffness. 5. Sensitivity to timestep should also be performed. Time step should be selected sufficiently small that variation between the standard deviations of curvature from successive runs are considered acceptably small.
6. It is recommended that sensitivity be performed to the applied coefficient of normal drag. A reasonable variation of normal drag coefficient should be selected cognizant of both Reynolds number and Keuglan Carpenter number. Alternatively, use of a Reynolds number varying approach may be adopted.

OTC 15168

7.0 References 1. API RP 17B (1998), "Recommended Practice for Flexible Pipe, Second edition". 2. DNV (2001), "Dynamic Riser, DNV-OS-F201" 3. API RP 2RD (1998), "Design of Riser for Floating Production Systems (FPSs) and Tension Leg Platforms (TLPs)", First edition. 4. Gere J.M & Timoshenko S.P. (1999) "Mechanics of Materials", 4th Edition, Stanley Thornes Ltd. 5. O'Brien P. & McNamara J.F., "Improvments to the Convected Coordinates method for prediction large deflection extreme riser response", OMAE2002. 6. Baltrop & Adams (1991), Dynamics of Fixed marine Structures, 3rd Edition, Butterworth Heinemann, The Marine Technology Directorate Limited.

OTC 15168

Table 1 Summary of Dynamic Analysis Results for 4.5" Water Injection Riser
H (m)
3 3 3 6 6 6 9 9 9 12 12 12 15 15 15 18 18 18

Table 5 6" Production Riser


H (m) 3 3 3 6 6 6 9 9 9 12 12 12 15 15 15 18 18 18 T(s) 4.26 5.33 6.41 6.03 7.54 9.06 7.38 9.24 11.1 8.52 10.67 12.82 9.53 11.93 14.33 10.4 13.07 15.7 Near Offset Maximum Max Min Terminal Hangoff Velocity Tension Curvature Velocity (1/m) (kN) (m/s) (m/s) 0.01 1.94 20.6 0.024 0.02 1.94 20.6 0.025 0.02 1.94 20.7 0.024 0.04 1.94 20.2 0.025 0.18 1.94 18.7 0.027 0.43 1.94 13.7 0.036 0.24 1.94 19 0.027 1.13 1.94 -3.3 0.066 1.04 1.94 2.5 0.075 0.73 1.94 15 0.033 2.59 1.94 -28.3 0.294 3.24 1.94 -30 0.393 2.62 1.94 -28.2 0.339 1.75 1.94 -17.5 0.216 5.91 1.94 -25.7 0.686 4.29 1.94 -27.9 0.452 5.63 1.94 -30.1 0.438 7.30 1.94 -30.7 1.23 Far Offset Min Tension (kN) 90.5 90.5 90.2 88.7 84.4 72.9 85.5 36.8 59.3 67.8 -35.3 -42.4 -38 0 -79 Max Curvature Vhangoff/ V (1/m) Terminal 0.00 0.005 0.01 0.005 0.01 0.005 0.02 0.005 0.09 0.006 0.22 0.008 0.13 0.006 0.58 0.022 0.54 0.012 0.38 0.009 1.34 0.125 1.67 0.413 1.35 0.26 0.90 0.052 3.05 0.483 2.21 2.90 3.77

T(s)
4.26 5.33 6.41 6.03 7.54 9.06 7.38 9.24 11.1 8.52 10.67 12.82 9.53 11.93 14.33 10.4 13.07 15.7

Near Offset Min Max Tension Curvature (kN) (1/m)


12.1 12 12.1 11.4 11 8.1 11.2 -1.9 1.4 9.1 -12.1 -14.4 -12.4 -9.7 -16.2 -14.8 -15.5 -16.3 0.027 0.027 0.027 0.027 0.03 0.04 0.029 0.112 0.086 0.035 0.361 0.468 0.424 0.25 0.544 0.459 0.574 1.713

Far Offset Min Max Tension Curvature (kN) (1/m)


55.8 55.8 55.6 54.9 52.7 47.6 553.4 28 38.4 45.3 -20.9 -30.2 -24.3 2.4 -78.4 0.005 0.005 0.006 0.006 0.006 0.008 0.006 0.018 0.011 0.008 0.169 0.35 0.287 0.058 0.484

Table 6 Sensitivity for 9.5" Production Riser


Near Offset H (m)
15 15 15

T(s)
9.53 11.93 14.33

Maximum Terminal Hangoff Velocity Velocity (m/s) (m/s)


2.62 1.75 5.91 2.72 2.72 2.72

Min Tension (kN)


-78.0 -9.0 -114.6

Max Curvature Vhangoff/ (1/m) VTerminal


0.237 0.093 0.350 0.96 0.64 2.17

Table 2 Sensitivity to Pipe Bending Stiffness


Pipe 4.5" Water Injection 15 6" Production H (m) T(s) EI 21,270 25,524 17,016 57,820 69,384 46,256 Min Tension (kN) -16.2 -16.6 -14.9 -25.7 -32.4 -27.1 Max Curvature (1/m) 0.544 0.478 0.82 0.686 0.451 0.496

Basecase 20% -20% 14.33 Basecase 20% -20%

Table 3 Sensitivity to Normal Drag Coefficient


Pipe 4.5" Water Injection Hmax (m) 15 T(s) 14.33 Cd 0.7 0.9 1.2 0.7 0.9 1.2 Min Tension (kN) -18.3 -18.1 -17.8 -42.3 -39 -39.9 Max Curvature (1/m) 0.642 0.575 0.62 0.489 0.532 0.566

6" Production

15

14.33

Table 4
H (m) 3 3 3 6 6 6 9 9 9 12 12 12 15 15 15 18 18 18 T(s) 4.26 5.33 6.41 6.03 7.54 9.06 7.38 9.24 11.1 8.52 10.67 12.82 9.53 11.93 14.33 10.4 13.07 15.7

4.5" Water Injection Riser


Near Offset Maximum Terminal Min Max Hangoff Velocity Tension Curvature Velocity (m/s) (kN) (1/m) (m/s) 0.01 1.90 12.1 0.027 0.02 1.90 12 0.027 0.02 1.90 12.1 0.027 0.04 1.90 11.4 0.027 0.18 1.90 11 0.03 0.43 1.90 8.1 0.04 0.24 1.90 11.2 0.029 1.13 1.90 -1.9 0.112 1.04 1.90 1.4 0.086 0.73 1.90 9.1 0.035 2.59 1.90 -12.1 0.361 3.24 1.90 -14.4 0.468 2.62 1.90 -12.4 0.424 1.75 1.90 -9.7 0.25 5.91 1.90 -16.2 0.544 4.29 1.90 -14.8 0.459 5.63 1.90 -15.5 0.574 7.30 1.90 -16.3 1.713 Far Offset Min Tension (kN) 55.8 55.8 55.6 54.9 52.7 47.6 553.4 28 38.4 45.3 -20.9 -30.2 -24.3 2.4 -78.4 Max Curvature Vhangoff/ V (1/m) Terminal 0.00 0.005 0.01 0.005 0.01 0.006 0.02 0.006 0.09 0.006 0.23 0.008 0.13 0.006 0.60 0.018 0.55 0.011 0.39 0.008 1.37 0.169 1.71 0.35 1.38 0.287 0.92 0.058 3.12 0.484 2.26 2.97 3.85 -

OTC 15168

P
Sensitivity of Vertical Motion to Riser Hangoff Positioning
2.2 2
Hangoff 0m Forward of COM Hangoff 50m Forward of COM Hangoff 100m Forward of COM Hangoff 150m Forward of COM Hangoff 200m Forward of COM

RAO of Vertical Motion (m/m)

1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 0 1

Deflected Shape

10

11

12

13

14

15

16

17

18

19

20

Period of Response (s)

Figure 1 RAO of Hangoff Point Vertical Motion depending on distance from Center of Motion
Sensitivity of Vertical Acceleration to Riser Hangoff Positioning
0.38 0.36 0.34 0.32 0.3 0.28 0.26 0.24 0.22 0.2 0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 0 1
Hangoff 0m Forward of COM Hangoff 50m Forward of COM Hangoff 100m Forward of COM Hangoff 150m Forward of COM Hangoff 200m Forward of COM

Figure 4

Idealized Strut in Compression

RAO of Vertical Acceleration (m/s^2m)

Pcr Compressive Force, P

10

11

12

13

14

15

16

17

18

19

20

Lateral Deflection, v

Period of Response (s)

Figure 2 RAO of Hangoff Point Vertical Acceleration depending on distance from Center of Motion

Figure 5

Sample Compressive Force vs. Deflection plot

20kN
Column Properties
Outer Diameter : 50mm Esteel : 210 MPa

6m

20N

Figure 3

Sample Snapshot Illustrating Compressive Buckling

Figure 6

Schematic of Problem for Verification.

10

OTC 15168

Figure 7

Snapshots Illustrating Progressive Buckling of a vertical Column (Flexcom3D)

Figure 10
45000

Snapshot of Post Buckling Behavior from ANSYS (3 of 3)

40000

35000

30000

Load (N)

25000

20000

15000

10000

5000

0 -7000

-6000

-5000

-4000

-3000

-2000

-1000

Vertical Displacem ent (mm)

Figure 8

Snapshot of Post Buckling Behavior from ANSYS (1 of 3)

Figure 11

Flexcom3D and ANSYS : Comparison of Top Node Displacement

Figure 12 Figure 9 Snapshot of Post Buckling Behavior from ANSYS (2 of 3)

Snapshot of Riser System

OTC 15168

11

0.9

0.8

0.7

Maximum Resultant Curvature (1/m)

0.6

0.5

0.4

0.3

0.2
4.5" Water Injeciton Near 6" Production Near 4.5" Water Injection Far 6" Production Far

0.1

0 0.00

1.00

2.00

3.00

4.00

5.00

6.00

7.00

Maximum Hangoff Velocity (m/s)

Figure 13

Summary Plot of Maximum Curvature vs. Hangoff Velocity for Basecase analysis

Figure 14

Summary Plot of Minimum Tension vs. Hangoff Velocity for Basecase analysis

12

OTC 15168

1.5

Non Dimensional Parameter ( Positive Increasing, Negative Decreasing)

0.5

0 50 55 60 65 70 75 80 85 90

-0.5

-1

Velocity Acceleration Curvature Time (s)


Figure 15 Variation of Hangoff Acceleration and Maximum TDP Curvature with Seastate

-1.5

Drag Force = .Cd..Ddrag.V2

Gravitational Force Fd = m.g


Figure 16 Snapshot of Riser Configuration

OTC 15168

13

4.50

4.00

3.50

3.00 Max Hangoff Velocity/ Pipe Terminal Velocity

2.50

2.00

1.50

1.00 4.5" Water Injection Near 6" Production Near 4.5" Water Injection Far 6" Production Far 0.00 -100 -80 -60 -40 -20 0 20 40 60 80 100 Minimum Effective Tension (kN) 0.50

Figure 17
4.50

Distribution of Minimum Effective Tension vs. Non dimensional Velocity Parameter

4.00

3.50

Max Hangoff Velocity/Pipe Terminal Velocity

3.00

2.50

2.00

1.50

1.00 4.5" Water Injeciton Near 6" Production Near 4.5" Water Injection Far 6" Production Far 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8

0.50

0.00

Maximum Resultant Curvature (1/m)

Figure 18

Distribution of Maximum Resultant Curvature vs. Non dimensional Velocity Parameter

14

OTC 15168

0.225 Time = 0.005s 0.200 Time = 0.01s Time = 0.02s Time = 0.03s 0.175 Time = 0.04s

0.150 Standard Deviation of Curvature (1/m)

0.125

0.100

0.075

0.050

0.025

0.000 200

250

300 Length Along Riser (m)

350

400

Figure 19

Standard Deviation of Curvature in The Touchdown Region


Elem length = 0.5m ent Elem Length = 1.63m ent Elem Length = 2.35m ent

Resultant Curvature at Touchdown Point (1/m)

0 65

0.1

0.2

0.3

0.4

0.5

0.6

0.7

70

75

80

85

90

TIME (S)

Figure 20

Sensitivity of Curvature to Element Length (Timestep of 0.01s)

OTC 15168

15

Tim = 0.01s, 0.5mElem e ent Tim = 0.01, 1.63mElem e ent TIm = 0.01s, 2.35mElem e ent

Resultant Curvature Standard Deviation

0 150

0.05

0.1

0.15

0.2

200

250

300

350

400

Curvilinear Distance along the set riser

Figure 21

Variation of Standard deviation of Curvature with Element Length

Вам также может понравиться