Managing Free Spans on Dynamic Seabed

Transcription

1 Managing Free Spans on Dynamic Seabed Piotr Krawczyk, Halny Ltd, Romke Bijker, Witteveen+Bos/ACRB, Zhiwen Chen, BMT Argoss, Hans Boersma, BBL Company, 1. Abstract This paper presents the strategy adopted to manage scour induced free spans on the BBL pipeline. The discussion focuses on two subjects: predicting span evolution in time and quantifying the sheltering effect of the scour trenches. Both these subjects are only partly covered by the existing span analysis methods,whichoften leads to overly conservative results being obtained and unnecessary remediation work being undertaken. It is demonstrated here that taking into account the sheltering effect of scour trenches and the span evolution in time can result in considerably longer spans being acceptable. This reduces the need for offshore intervention work and, in some cases, it may eliminate it altogether. This work is based on field measurements of scour trenches and span evolution. While the discussion is presented in the context of the experience with the BBL pipeline, it is directly applicable to any offshore pipeline susceptible to scour and dynamic seabed conditions. Keywords: pipeline, free span, scour, VIV, fatigue, span evolution 2. Introduction Pipelines laid on erodible seabed are prone to develop long free spans. This is caused by two factors: seabed scour around the pipelineand/or global seabed evolution irrespective of the pipeline (i.e.: sand wave migration).both these phenomena are of particular concern in relatively shallow water applications, like the North Sea, where the combined loading by waves and currents may result inthe span being exposed to unacceptably high environmental loads. This paper considersintegrity management of the free spans caused by seabed scour. The discussion follows two parallel developments made for the BBL pipeline. The first one focuses on improving our understanding the sheltering effect of scour trenches. The second one focuses on developing analysis techniqueswhich would allow predicting spanevolution in time. When Page 1

2 combined, these two developments resulted in significantly reduced amount of offshore intervention work required to maintain the pipeline in an acceptablecondition. This work is driven by the experience with BBL pipeline since the start of its operation in the 2006, where: Spans identified during spring time surveys and concluded to be unacceptable in the as-found configuration had disappeared, or significantly reduced in length,by the time remediation works could be undertaken in autumn the same year.this suggests that standard acceptance criteria for scour induced free spans may lead to unnecessary remediation work being undertaken. Most scour trenches are observed as consistently more narrow than the scour trenches reported in the literature and being the basis for the standard assessment methods. Thus, a dedicated assessment has been undertaken to quantify the sheltering effect of narrow scour trenches. This paper describes the present status of the studies. Further development is on-going. 3. Past developments The process of seabed scour around pipelines has been extensively investigated over the last three decades. A valuable summary of these developments can be found in an overview paper by Sumer and Fredsoe, ref.[1], which attempts to summarise the various research directions. However, despite this extensive research effort, there is still little practical guidance available to the engineers managingscour induced spans. The DNV Recommended Practice F105, ref.[2], being the most comprehensive guideline for free span analysis, is clearly geared towards assessing the free spans in the as-found configuration and it remains silent on how to address span evolution in time. A nominal safety margin is recommended for scour induced spans (to cater for future span growth), but this is rather simplistic approach and it has been concluded to lead to unnecessary remediation work being undertaken. Furthermore, the sheltering effect of scour trenches is only partly included in the DNV-RP-F105 methodology. It is defined for cross-flow VIV and direct wave action, but it is missing entirely for in-line VIV, which tends to govern free span fatigue estimates. It will be shown here that the sheltering effect of scour trenches is also applicable to in-line VIV and that taking this effect into account can further reduce the requirement for span rectification. The work presented in ref.[4]introducesprobabilistic analysis techniques to determine the acceptability of scour induced spans developing as a result of pipeline self-lowering. This can be used to enhance to the default DNV-RP- Page 2

3 F105 methodology. However, it should be realised that the method relies on a number of important assumptions such as: (1) fatigue limit state is assumed to govern at all times and (2) it is assumed that the pipeline develops a large number of spans,all of which evolve rapidly and behave in a similar manner. These assumptions are generally appropriate for effectively self-lowering pipelines (pipelines self-lowering in relatively short time). However, it should not be assumed that all scour induced spans behave in this manner. Experience with the BBL project suggests that some of the most onerous scour induced spans behave in an entirely different manner, for which statistical analysis may not be the most appropriate approach. An alternative approach is therefore developed for these spans. 4. Sheltering Effect of Scour Trenches The sheltering effect of scour trenches is investigated here using an advanced Computational Fluid Dynamics (CFD) model presented inref.[6]. This model has been specifically developed to tackle large amplitude pipeline Vortex Induced Vibrations (VIV) near the seabed. This is achieved through the use of mesh-deformation and re-meshing techniques introduced in recent years in Ansys CFX software. Furthermore, the model allows for both, in-line (horizontal) and cross-flow (vertical) pipeline motion across the fluid domain. It is therefore capable of resolving pipe motions without the requirement for making any arbitrary assumptions with regards to the interaction between the two vibration modes, or the relationship between the hydrodynamic forces and the vibration amplitudes. All these interactions are automatically calculated from the first principles (i.e.: fluid mechanics and structural engineering principles are used to resolve fluid-structure interaction). Calibration results presented in ref. [6]suggest that the CFD model is capable of representing experimental results at the Reynolds number and the reduced velocity levels commonly encountered in offshore pipeline applications. Figure 1 shows a snap-shot of the CFD model. The pipeline can be seen located in a typical scour trench for the BBL pipeline (sheltering 50% of the pipe diameter and with a width of approximately two pipe diameters). The plot shows vortices being shed behind the pipeline in a steady flow simulation. These are the red and the blue oval patterns to the right hand side of the figure. As the flow is affected by the proximity of the seabed, the vortices are being shed in an alternating pattern, which results in a combined cross-flow and in-line VIV response at all times (there is no pure in-line VIV response stipulated by DNV-RP-F105 methodology). Page 3

4 Figure 1: CFD model typical vortex shedding in a steady flow from pipe sheltered by a narrow scour trench Figure 2 shows an example of the sheltering effect observed for the pipeline. Compared here are the motion patterns for the pipeline located over flat seabed and in a typical scour trench sheltering 50% of the pipe diameter (all other conditions remain the same). Pipeline located in the scour trench can be seen to experience significantly lower vibration amplitudes. It can also be noted that the motion pattern has changed considerably, from a flattened eight-shaped pattern over flat seabed to an oval vibration pattern in a scour trench. This sort of oval motion pattern has been consistently observed for all investigated trench depths and profiles. Figure 2: Pipeline motion patterns over flat seabed and in a scour trench Page 4

5 Figure 3 shows two of the scour trench profiles investigated for the BBL pipeline. Both trench profiles are a schematisation of the in-situ trench profiles recorded using Multi Beam Echo Sounding survey equipment.figure 3a shows a moderately deep, slightly asymmetric, scour trench profile sheltering 50% of the pipe diameter on the upstream side and approximately 60% of the pipe diameter on the downstream side. Figure 3b shows a deep scour trench, sheltering 65% of the pipe diameter. These two trench profiles are representative of some of the longest spans observed on the BBL pipeline, all of these spans are in the order of 50-60m long. Figure 3a Figure 3b Figure 3:Scour trench profiles Figure 4 and 5 show the response models generated based on the CFD simulations for the pipe in moderately deep scour trench (sheltering 50% of the pipe diameter). It can be noted that the cross-flow VIV response is fairly consistent with the DNV-RP-F105 predictions. However, the onset of in-line VIV is considerably delayed when compared to the predictions based on the Recommended Practice methodology. That is caused by the fact that DNV- RP-F105 methodology does not incorporate the sheltering effect of scour trenches for in-line VIV. This limitation is not explained in the Recommended Page 5

6 Practice. However, reviewing the background literature, ref.[3], it can be noted that the effect is missing simplybecause it has not been investigated at the time when therecommended Practice was published. Figure 4: In-Line VIV in scour trench sheltering 50% OD Figure 5: Cross-Flow VIV in scour trench sheltering 50% OD Page 6

7 Figure 6 and 7 show the response models generated based on the CFD simulations for the pipe in the deep scour trench (sheltering 65% of the pipe diameter). Here, the pure in-line VIV response is essentially non-existent. Appreciable VIV amplitudes are only observed for reduced velocities in excess of 3.5. Interestingly though, in-line VIV for higher flow velocities (V r >4.5) is notably higher than that predicted by DNV-RP-F105. This effect, associated with so-called cross-flow induced in-line VIV, has been consistently observed for all investigated scour trench profiles. Cross-flow VIV response is fairly consistent with DNV-RP-F105 predictions, albeit the amplitudes are lower. That is probably the result of BBL scour trenches being somewhat narrower than those investigated in ref.[3]. Figure 6: In-Line VIV in scour trench sheltering 65% OD Page 7

8 Figure 7: Cross-Flow VIV in scour trench sheltering 65% OD The above results suggest that fatigue predictions for scour induced free spans can improve considerably when taking into account the sheltering effect of scour trenches. This effect is not fully captured by the existing methods of analysis, which results in overly conservative results being obtained. BBL Company used advanced Computational Fluid Dynamics simulation techniques to quantify the sheltering effect of in-situ scour trenches observed during survey and to refine the DNV-RP-F105 span analysis method by introducing custom response models for scour induced free spans. The result of this improvement is illustrated in Table 1, where fatigue predictions are compared for the default DNV-RP-F105 response models and for the custom BBL response models. It can be noted that considerably longer spans can be tolerated when the trench effect is fully accounted for. Longer acceptable span lengths translate to less remedial work being required to maintain the pipeline in an acceptable condition. In some cases, entire clusters of spans have been demonstrated not to require any remediation at all. Page 8

9 Table 1: The effect of scour trench on span fatigue predictions Scenario Span over Span in a moderate scour Span in a deep scour flat seabed trench, sheltering 50% OD trench, sheltering 65% OD Response Model DNV-RP- DNV-RP- DNV-RP- BBL F105 F105 F105 BBL Span Length (L/D) Fatigue estimate for the free span horizontal direction (in-line VIV + Direct Wave Action) yr 6.1 yr 2,699 yr 6.1 yr 5,448 yr yr 0.6 yr 872 yr 0.6 yr 1,700 yr yr 0.1 yr 305 yr 0.1 yr 578 yr yr 0.0 yr 98 yr 0.0 yr 188 yr Span Length (L/D) Fatigue estimate for the free span vertical direction (cross-flow VIV) 46 7,175 yr 100,000+ yr 100,000+ yr 100,000+ yr 100,000+ yr yr 4,990 yr 89,919 yr 9,818 yr 100,000+ yr yr 347 yr 1,830 yr 654 yr 100,000+ yr yr 93 yr 124 yr 174 yr 13,562 yr The results presented in Table 1 are truncated at non-dimensional span lengthl/d = 59, as longer scour spans are not likely to occur. A span with L/D ratio of 59 is long enough to develop vertical sag of circa 0.8 pipe diameter.this is more than the scour trench depth (typically between 0.4 and 0.7 pipe diameter). The pipe can therefore be expected to touch the bottom of the scour trench and to self-remediate the span. As the pipeline would now be located in a very deep trench, sheltering circa 80% of the pipe diameter, it would no longer constitute an obstacle to the sea-water flow. The scouring process would therefore cease and natural backfilling of the trench would commence. 5. Span Evolution Acceptability of pipeline free spans is generally assessed using guidelines and tools, which are developed assuming stationary span configuration.such an approach is adequate for the spans caused by large, persistent seabed features (e.g.: glacier scars, boulders, or rocky outcrops), or in case of cohesive soils, which are generally less dynamic (i.e.: less susceptible to scour). For non-stationary spans ona dynamicseabed, however, the situation is very different, as the criticality of the free span is strongly linked to the persistence of the span and the prevailing environmental conditions. A realistic and reliable assessment must therefore account for:(1) the history of the span and (2) the predictions into its future development. Several studies have been carried out in the past on free span evolution, including Hansen et al (1991), Klomp et al (1995), Chen & Bijker (2001) and Page 9

10 Drago et al (2007), ref. [8], [9], [5] and [7] respectively. In these studies, free span evolution was modelled as an integral part of the pipeline self-lowering process. Model validation was generally carried out in terms of pipeline selflowering and no direct validation of the free span evolution was carried out. This was mainly due to the absence of reliable field data measurements on the evolution of the span lengths. As a result, there is still noreliable tool to actually predict free span development in time. This paper reports on the first results of an ongoing study to develop a tool for predicting span evolution in time. Thistoolis being developed on the basis of the formulations derived for self-lowering pipelines and isfurther upgraded with new formulations, to realistically account for pipe-soil interaction processes that have been observed in recent years of BBL pipeline span management. The following discussion gives an overview of the approach, the field data used for model validation and some preliminary results.the overall model structure is presented in Figure 8. Figure 8:Overall structure of the span evolution model Span celerity is calculated based on the following formulations, ref. [8] and [9]: C sn ( qcn q0n ) D (1 p) ed Page 10

11 C st q 0t D (1 p) ed Managing Free Spans on Dynamic Seabed where: C sn = span celerity due to sediment transport normal to pipeline (m/s) C st = span celerity due to sediment transport parallel to pipeline (m/s) q cn = sediment transport normal to pipeline in the corner of span (m 3 /ms) q 0n = sediment transport normal to pipeline upstream of pipeline (m 3 /ms) q 0t = sediment transport parallel to pipeline upstream of pipeline (m 3 /ms) p = porosity of seabed (-) D = pipe outer diameter (m) e = pipe embedment (m) β = coefficient for the length of span support area (-) This model was initially calibrated using experimentallaboratory data, ref. [9], and it is now refined using field measurement data for the BBL pipeline. The field data includes: (1) free span length measurements based on BBL pipeline surveys and (2) the wind and wave height measurements at the K13 platform. Free span development in time is established based on the comparison of the following surveys: the 2009spring time survey the 2010spring time survey the 2010autumn re-survey the 2011spring time survey the 2012spring time survey The wave data at the span locations are derived from the measurement data at the K13 platform. The measured winds and waves are transformed to the relevant pipeline locations using a dedicated hindcast model. Figure 9 shows typical wave height variations at a location along pipeline route where a free span of interest has been detected. The wave height distribution shown in Figure 9 is typical for the Southern North Sea, where storms with significant wave height of circa 6m are observed annually in winter time. Summer time is generally calmer, with the significant wave height rarely exceeding 3m. This wave climate dictates pipeline inspection/maintenance program, where the pipeline is surveyed in spring time, any detected free spans are assessed in summer time and remediated (if required) in autumn, before the winter storms could affect pipeline integrity. Page 11

12 Wave height Hs (m) Managing Free Spans on Dynamic Seabed Date Figure 9:Historical wave data at the span of interest The span evolution model is calibrated using the available field data for a few selected locations,where free spans of interest have been detected. These spans are typically characterized by moderate growth rates (in the order of a few meters per year). Thus, they are neither stagnant, nor fast evolving, as assumed in ref. [4]. A statistical approach to span evolution is therefore not justified and an alternative (deterministic) approach seems more appropriate. An example of calibration results for span evolution model is provided in Figure 10.The horizontal axis is the time, the blue curve shows significant wave height variation, the black rectangles show the span lengths observed during pipeline inspections and the red, dashed line shows the predications from the span evolution model. It can be noted that span growth is not uniform in time. The scour induced spans develop and grow rapidly during larger storms. During calm weather periods the spans tend to remain stagnant, or may even start backfilling. The backfilling process is not considered in current span evolution model, as this backfill material consists of very loose sand and is rapidly washed away during the next large storm. Thus, the span evolution model predicts a stagnant span in-between the storms. Page 12

13 Figure 10:Span evolution model calibration The calibration assessment presented above considered historical developments at the span. The same methodology can however be used to predict future behavior of the existing spans. This would normally consider the following two scenarios: Span development during typical winter storms, up to the next scheduled survey. Span development during an extreme, or design storm. Examples of such predictions are provided below. Span evolution for the typical winter conditions is predicted by assuming previous years environmental conditions. Span evolution during an extreme event is predicted by considering a 100-year design storm condition with a typical build-up time. Page 13

14 Figure 11:Predicted span growthassuming the 2011/2012 wave data Figure 12:Predicted span growthassuming the 100-year wave data Page 14

15 Assuming the same wave conditions as in previous years, the span evolution model predicts that the span will increase in length by approximately 7m. This is significant increment, yet the final span length is considered acceptable, due to the sheltering effect of scour trench. Considering extreme storm conditions, it can be noted that the span is expected to remain stagnant up to the point where the significant wave height would reach 4.5m. Beyond this threshold, rapid span growth is expected. It should be noted however that the span would self-remediate before the peak storm conditions could occur. This is motivated by the fact that an 80m long span would be long enough for the pipe to sag down into the scour trench and thus split the original span into two shorter ones. In fact, a 50-65m long span is already predicted to develop sufficient sag to facilitate self-remediation of scour induced spans. Thus longer spans are not expected. As mentioned before, the span evolution model is under development. The model still has the following limitations: The reliability of the model obviously depends on the reliability and accuracy of the survey data and the specialist interpretation. Uncertainties in the span data, in particular the support configuration, result in uncertainties in the span predictions. More detailed analysis of the field data is presently being performed. In addition discussions are being held with the specialists of the survey company to further improve the data quality of the free span surveys. The process is schematized and enhancements are presently being introduced. We already know form previous studies of actual self-lowered pipelines that the prediction of thebackfilling process needs improvement. From the available field data we also know that more work is required on the schematization of the seabed data, to incorporate the significant effects of compactness of sand and changes in cohesiveness due to small changes in the content of silt or clay. The model is not yet able to predict self-remediation of the spans. Additional function would need to be introduced to limit span length based on the structural span model predicting pipe deflection into the scour trench, and the scour trench depth prediction based on the storm conditions, the geotechnical data and the initial pipe embedment into the mean seabed. Further development and additional calibration of the model is therefore ongoing. Page 15

16 6. Conclusions and recommendations It is concluded that the sheltering effect of scour trenches is not sufficiently covered by the traditional span analysis methods. Taking this effect into account results in considerably longer free spans being acceptable. Thus, the requirement for remedial work at scour induced free spans can be reduced, or eliminated. The results presented in this paper confirm the observation presented in ref. [3], that the presence of a scour trench delays onset of pipeline VIV. At higher flow velocities the trench effect is less pronounced and large amplitude VIV may still be possible. Previous work on the sheltering effect of scour trenches, ref.[3], concluded that the trench effect is not sensitive to the scour trench profile. Recent work by BBL indicates that this may not always be the case. The sheltering effect of scour trench was observed to be sensitive to the trench width and profile. Further work is however required, to quantify this effect and to come up with a generalised guideline. Recent advances in modelling of turbulent flows allow for using Computational Fluid Dynamics approach to refine our understanding of VIV in general, and for studying the sheltering effect of scour trenches in particular. However, for the method to be fully accepted by the industry, it is recommended to perform further validation of the CFD predictions versus experiments. A Joint Industry Project will therefore be launched to obtain further confirmation of the CFD predictions and to come up with a generic guidance note which could be used for a wider range of pipeline applications. Limiting span lengths for pipelines are generally calculated assuming stationary free spans. This seems to be rather conservative for dynamic seabed conditions, where the criticality of free spans is strongly linked to the persistence of the spans and the prevailing environmental conditions. A realistic and reliable assessment must therefore account for (1) the history of the spans and (2) the predictions into their future development. A span evolution model is therefore being developed and the first version has successfully been calibrated for a number of free spans that have been monitored in detail for the last three years. Although limitations of the model are eminent, the predicted trend is in good agreement with the field data. Further work is on-going to enhance the reliability of the span evolution model. The new approach to managing scour induced free spanstakes into consideration the most important aspects of the erodible and dynamic seabed, being (1) the development of narrow scour trenches providing shelter to the pipeline, and (2) the erosion of free span supports and interaction with Page 16

17 dynamic bedforms. Both these aspects resulted in significant improvement of methods and models. They are combined and integrated in the free span management procedure for the BBL pipeline, which has already resulted in less remediation without jeopardising the integrity of the pipeline system. Present work is based on field measurements for the BBL pipeline. Similar approach could however be applied to any offshore pipeline susceptible to scour and dynamic seabed conditions. 7. References [1] A Review of Wave/Current-Induced Scour Around Pipelines; B.M. Sumer, J. Fredsoe; Coastal Engineering Proceedings, 1992, Chapter 217 [2] Free Spanning Pipelines, DNV-RP-F105, 2006 [3] Vibrations of a Free Spanning Pipeline Located in the Vicinity of a Trench; E.A. Hansen, M. Bryndum, K. Mork, R. Verley, L. Sortland, H. Nes; OMEA 2001, Paper No [4] A Strategy for Assessment of Non-Stationary Free Spans, K. Mork, O. Fyrileiv, H. Nes, L Sortland, Proceedings of the Ninth International Offshore and Polar Engineering Conference, 1999 [5] Interaction of Offshore Pipelines and Dynamic Seabed, Z. Chen, R. Bijker, International Association for Hydro-Environment Engineering and Research (IAHR) conference, Beijing, 2001 [6] Numerical Simulation of Pipeline VIV Near the Sea-Bed for Steady and Unsteady Currents; P. Bhattacharjee, K.K. Nielsen, G. Stewart;Proceedings of the ASME 28 th International Conference on Ocean, Offshore and Arctic Engineering; OMAE 2009 [7] Drago, M, M.Pigliapoco and T.Ciuffardi (2007), Analysis of Pipeline Fatigue Damage for Scour Induced Freespans, Proceedings of the Sixteenth (2007) International Offshore and Polar Engineering Conference Lisbon, Portugal, July 1-6, 2007 [8] Hansen E.A. Staub C. Fredsoe J. and Summer B.M. (1991) Time development of scour induced free spans of pipelines Proc. of Offshore Mechanics and Arctic Engineering Conference [9] Klomp,W.H.G., E. A. Hansen, Z. Chen, R. Bijker and M.B. Bryndum (1995), Pipeline Seabed Interaction, Free Span Development. ISOPE 1995 Page 17