Proceedings of the ASME 2011 30th International Conference on Ocean, Offshore and Arctic Engineering OMAE2011 June 19-24, 2011, Rotterdam, The Netherlands OMAE2011-49 SQUALL RESPONSE BASED DESIGN OF FLOATING UNITS IN WEST AFRICA Juan Alvarez Bureau Veritas Neuilly Sur Seine, France Pierre Orsero UTC Compiegne, France Valerie Quiniou-Ramus TOTAL Paris, France Michel François Formerly Bureau Veritas Neuilly Sur Seine, France Anne-Gaëlle Moysan TOTAL Paris, France Didier l’Hostis TOTAL Paris, France ABSTRACT Squalls are one of the main issues for the design of West Africa floating units mooring systems. At the present time and due to the lack of more relevant information and models, squalls are represented by on site time series of time varying wind speed and relative heading. The first FPSO units were designed on the basis of a reduced Squall database. Nowadays, the number of squall records has been significantly increased and a response based analysis can be carried out. The present paper is focused on the Gulf of Guinea environment. The area has been divided into two zones: North (Nigeria…) and South (Congo, Angola…). This approach enabled us to deal with 90 Squall events for North zone and 115 Squall events for South zone. Two different mooring systems, with quite different natural periods, have been investigated in order to cover the range of already installed spread moored FPSO’s. For every Squall of the database, time domain and modal simulations have been carried out in order to obtain the maximum values of the axial tension in mooring lines and of the offset of a standard spread moored unit. Then a statistical procedure is applied a) to estimate 100-year return period values for these parameters and b) to assess overall trends besides the differences between results from both zones and both mooring systems. Alain Ledoux TOTAL Paris, France classical design procedures in order to evaluate the potential design margins for extreme responses. Finally, areas needing further investigation are identified. INTRODUCTION In West African waters, squalls cause strong and sudden winds. Although major progress has been made in the past few years for describing squall winds, there are still many uncertainties in calculating their effects on the mooring system of floating structures, and the lack of design guidelines leads to time consuming and costly engineering work without a clear understanding of safety level reached. The present study was initiated with the objective to define “Response-Based Design” (RBD) methodologies for the analysis of mooring systems under squall conditions that could be applicable to different types of systems, such as spread and turret moored FPSO’s, or buoy moored tankers. The final objectives of the project are: a) To provide the designers with methodologies to design mooring systems for squall winds, that are suitable for both early design stages (quick and conservative approach) and detailed design (accurate assessment of system response) b) To understand the level of conservatism behind each methodology. A comparative study has also been carried out to relate the 100year return period extrapolations with the values derived from 1 Copyright © 2011 by ASME BACKGROUND Analysis of wind measurements offshore West Africa clearly highlights the presence of two wind velocity populations, differing in intensity, direction stability and frequency. The normal wind regime is dominated by persistent southerly trade winds, driven by large-scale atmospheric pressure systems, and blowing predominately from the southsouthwest. Extreme winds are caused by squall events. Squalls are associated with the leading edge of multi-cell thunderstorms, generated on land, which tend to form along lines separating air masses and which usually travel westward over the sea. The squall seasonality is linked to passages of the inter-tropical convergence zone (ITCZ) on the way north in northern hemisphere spring then again on the way south in the northern hemisphere autumn. Squalls trigger a sudden and large increase of wind speeds (a squall event usually lasts less than an hour at one given location), with often (but not always) large variations even a full reversal, of wind direction. In-situ wind measurements were analysed during Phase 1 of the West Africa Gust Joint Industry Project (WAG JIP, [1]), using more than four years of blended wind records at various fields operated by Oil Companies (Chevron, ExxonMobil, Shell, Statoil, Total, Woodside). This work gave very useful insight to a number of aspects, such as : • Physics of phenomenon, • Criteria for squall event identification from records, • Geographical, seasonal and diurnal variability, • Directional distribution, • Extreme values, • Gust Factors. WAG Phase 1 was extremely beneficial to JIP participants, who could significantly improve the reliability of their met-ocean design criteria. Knowledge of the squall physical phenomenon was also developed thanks to Phase 3 of the WAG JIP [3], [4]: in-situ measurements with a vertical and horizontal array of anemometers were conducted on Total E&P Congo platform “Likouala” for two years. This provided information about the spatial distribution and coherence of squall winds, and thus gave an insight about how squalls apply on large structures such as Floating Production Storage and Offloading vessels (FPSO). design condition for some deepwater FPSO mooring systems, although to date no failure has been observed in reality. Therefore, it is of upmost importance to understand the level of conservatism behind the current design methodology and alternatives that are proposed in the present paper. DATA PRESENTATION To reach the present objectives, the authors have made use of in-situ wind records acquired by TOTAL S.A. in the frame of the WAG JIP, or by TOTAL affiliates, in particular Total Upstream Nigeria, Total E&P Angola and Total E&P Congo. WAG Phase 1 From the WAG1 JIP [1], TOTAL, as the Operator of fields offshore Nigeria, Congo and Angola, had supplied time series of wind recorded in these three regions, see Table 1. Site Elevation of sensor Start End Nigeria 84m 03/04/2003 09/03/2004 2s 17 Congo 30m 01/01/1997 31/10/1998 5s 13 Angola 60m 27/05/2002 12/01/2004 1s 4 Sample Intervals Events Table 1: WAG1 data sets WAG Phase 3 From WAG3 [3] [4] , TOTAL as a JIP partner has access to the global database, see Table 2 . Site Elevation of sensor Start End Sample Intervals Events Congo 10m->38m 01/12/2006 01/05/2007 3s 41 Congo 10m->38m 01/12/2007 01/05/2008 3s 25 Table 2: WAG3 data sets TOTAL own database TOTAL has also access to operational data in order to build their own database and to improve their understanding of Squall phenomena. Those continuous operational records were taken on floating units offshore Congo, Angola and Nigeria, see Table 3 . Site Elevation of sensor Start End Sample Intervals Events CONGO 51m 28/05/2008 31/03/2009 120s 12 ANGOLA 60m 01/01/2007 31/03/2009 60s 31 NIGERIA 50m 19/02/2008 21/03/2009 60s 73 Table 3: Total database units Although major progress has been achieved during the past few years with describing squall physical properties, there are still many uncertainties when it comes to calculating their effects on floating structures. It turns out (see e.g. [2]) that squalls are the 2 Copyright © 2011 by ASME Data preparation Two subsets have been built from the available data. One subset called North of Gulf of Guinea was based on records from offshore Nigeria; the other, called South of Gulf of Guinea, was based on records from offshore Congo+Angola. Following further data quality control and preparation, which is not detailed here but which includes sensor height correction to a common height of 10 meters, the resulting two subsets included respectively 90 and 115 squall events for a cumulated duration of approximately 2 and 6.5 years. An example time series is given in Figure 1 in which is represented in blue the one minute mean wind speed (u1MIN), in red the one minute mean direction (θ1MIN), in dot blue the maximum three second gust each minute (U3SEC:1MIN). units, where spread moored FPSO’s used to be aligned with primary Swell. The mooring pattern is illustrated in Figure 2. The aim of the present work is to analyze the effect of winds, thus wind forces coefficients are necessary for the study. Current force coefficients have also been added in the model in order to account for damping due to the relative motions between the floater and water. Wind force coefficients and current force coefficients have been deduced from those of installed units. They are shown for illustration in Figure 3 and Figure 4. 700.0 A1 A16 A2 A15 A3 A14 A4 A13 P7P8 P6 P5 S8 S7 S6 S5 N C F16 F15 F14 F13 F1 F2 F3 F4 Point E F12 F11 F10 F9 F5 F6 F7 F8 P4 P3 P2 P1 S4 S3 S1S2 A5 A12 A6 A11 A7 A10 A8 A9 Figure 2: Mooring pattern Wind Force Coefficient [kNs²/m²] Last but not least, extreme wind speeds were calculated in a compatible method as in WAG1. SQUALL RESPONSE BASE APPROACH For the Gulf of Guinea environment, the response of a typical FPSO has been analyzed. The following methodology has been applied: a) Building of a numerical model to represent the floater configuration, b) Computing the response of the model under each squall event of the data base and then extrapolating the response to extremes (values at return periods of interest for the evaluation of the system) , c) Comparison between RBD results and other methodologies. Numerical Model A typical 2 million barrel spread moored FPSO has been considered. The mooring system of the unit is composed of 16 lines (4 bundles of 4 lines). The heading of the vessel has been set to 22.5° which is a common value for West Africa Wind Moment Coefficient [kNms²/m²] Figure 1: Example time series of wind speed and direction during a squall event 0 30 60 90 120 150 180 Incidence w.r.t FPSO [Degrees] Longitudinal Force Transversal Force Moment Figure 3: Wind force coefficients Wind/current force coefficients are commonly derived from wind tunnel tests. During the wind tunnel tests, the wind loads on the above-water portion of the FPSO are determined in a boundary-layer wind profile. Commonly the wind profile is 3 Copyright © 2011 by ASME represented by a logarithmic profile (NPD or API profile), but from recent results (WAG1 JIP or WAG3 JIP) it has been found that squall vertical profile does not match with these representations and is more slab-like. Wind tunnel facility could generate such profile as they do when performing test for current loading however to our knowledge it has not been yet performed and related problems such as boundary layer thickness and blockage correction should be addressed. 0 30 60 90 120 150 Current Moment Coefficient [kNms²/m²] Current Force Coefficient [kNs²/m²] Another source of possible inaccuracy in the evaluation of wind loads, which not further investigated in the present study, is the use of tunnel mean force coefficients to calculate the time varying load from the instantaneous wind speed. 180 domain simulation under environment, including first and second order effects. But in the present case, as waves are not considered, hydrodynamic loads are reduced to added mass. From the line point of view, the quasi-dynamic analysis takes into account: • the displacement of the upper end of the mooring line due to the floater low frequency and wave frequency motions, • the weight and the buoyancy of the mooring line components, • the elasticity of the mooring line components. The quasi dynamic analysis includes all the floater dynamic effects such as inertia, low frequency wind, except the dynamic of the line itself. Response based design (RBD) For each squall event of the database a time domain calculation has been carried out to extract the maximum axial line tension and the maximum offset amidships. Results of maximum tension in the most loaded line versus maximum offset are shown in: • Figure 5 for a water depth of 800m. • Figure 6 for a water depth of 1300m. Each cross or circle in Figure 5 and Figure 6 represent the results of one squall from the data base. 3000 Longitudinal Force Transversal Force Maximum axial tension [kN] Incidence w.r.t FPSO [Degrees] Moment Figure 4: Current force coefficients The 4x4 spread mooring system consists of the following segments, from anchor to fairlead: • 600m of 132mm studless chain, • 102 mm Sheathed spiral strand wire rope (the length has been adjusted according to the water depth – see below), • 100m of 132mm studless chain. 1500 0 Two different water depths have been considered for the numerical modeling: 1300m and 800m. Due to that, two systems with quite different natural periods are obtained: • Around 400s in surge and sway for the water depth of 1300m • Around 200s in surge and sway for the water depth of 800m These natural periods are in accordance with already installed FPSO’s of similar characteristics. The response of the squall will be obtained by a quasi-dynamic time domain calculation. The quasi-dynamic analysis is a time 25 Maximum offset [m] NORTH ZONE SOUTH ZONE Figure 5: Water depth 800m 4 Copyright © 2011 by ASME Water depth=1300m 3000 Maximum axial tension [kN] RP (years) 100 50 25 10 1 Offset (m) Water depth=800m 80 Maximum offset [m] NORTH ZONE SOUTH ZONE RP (years) Offset (m) 95% lower value (m) 95% upper value (m) 100 24 19 28 50 22 17 26 25 20 16 23 10 17 14 20 Figure 6: Water depth 1300m Then, results from the direct simulation have been extrapolated to obtain the extreme values at return periods of 100, 50, 25, 10 and 1 year (denoted RP100, RP50, RP25, RP10, and RP1 in figures below). 1 For the extrapolation of offsets, the following hypotheses have been made: • a threshold has been defined at a return period of 0.1 year, • a fit to a Gumbel function has been carried out. 11 9 12 Table 5: Extreme values for SOUTH Zone Water depth=1300m RP (years) Offset (m) 95% lower value (m) 95% upper value (m) 100 122 101 138 50 113 94 128 25 105 87 119 10 93 78 106 As the system is highly non linear for the tension parameter (the non linearity is less evident for offsets), the tension extrapolations will be deduced from the offset. Extreme values, and the limits of a 95% confidence (obtained by bootstrapping), are given in Table 4 to Table 7 for the offset parameter. It is observed that the responses (tension and offset) to the squalls are significantly influenced by the mechanical system: stiffness and natural periods, thus water depth. Although it is not obvious to extract a trend for all the systems, it appears that the NORTH zone is more severe than SOUTH zone, which is a consistent result with the physics of squalls. 95% upper value (m) 71 54 85 66 51 79 61 47 73 54 43 64 38 32 43 Table 4: Extreme values for SOUTH Zone 1500 0 95% lower value (m) 1 65 55 73 Table 6: Extreme values for NORTH Zone Water depth=800m RP (years) 100 50 25 10 1 5 Offset (m) 95% lower value (m) 95% upper value (m) 51 44 55 47 41 51 43 38 46 38 33 41 25 22 27 Table 7: Extreme values for NORTH Zone Copyright © 2011 by ASME Comparison between statistical extreme values from RBD and other approaches During the design of the first floating units in West Africa the squalls were not a well documented phenomenon. The squall events were not discriminated from the normal wind, and the floating units were designed using a constant wind or a spectral wind (typically a NPD spectrum). Nowadays the common practice is to rescale the squalls time series records so that the maximum wind speed matches the value at a given return period. The response is then taken as the maximum response over all the squall time series, or as the MPM- or the expected value- of the distribution of response. In the following those approaches will be compared to the statistical extremes derived from the direct calculations. Thus, the first step is to rescale each squall of the data base and to calculate the induced response (tension and offset). The rescaling has been done with target 1 minute maximum wind velocity between 15 and 40m/s. An example of the results obtained from the rescaling is presented in Figure 7 for: • The axial tension parameter • A water depth of 1300m • SOUTH zone Each line in Figure 7 represents the results for one rescaled squall. Thus, the values obtained at a given return period (100 years, 50 years, 25 years and 1 year) by the RBD approach have been compared to values obtained by: • rescaling • constant wind • spectral wind for a one minute wind speed with the same return period. For illustration, the results obtained for the offset parameter at the NORTH zone in Figure 8 (water depth 800m) and Figure 9 (water depth of 1300m). Due to the rather limited number of squall events it seemed advisable, for the RBD results, to consider the upper value of the 95% confidence interval. Then, it can be observed that the maximum response obtained by a rescaling procedure is always on the conservative side (i.e. higher than the 95% confidence interval), whereas, at least for the higher return periods in all cases, both the other rescaling approaches and the sustained wind approaches are on the nonconservative side. 70 60 OFFSET [m] Maximum axial tensionin most loaded line [kN] 50 40 30 20 10 0 1 minute wind speed at an elevation of 10m [m/s] RP100 Figure 7: Results rescaling – Tension – Water depth =1300m – SOUTH ZONE RP50 RP25 RP10 RP1 MAXIMUM OVER ALL SQUALLS RBD 95% CONFIDENCE INTERVAL RBD SPECTRAL WIND EXPECTED MPM CONSTANT WIND From the set of results at any given rescaling wind speed, the following information could be extracted: • The maximum response over all the database. • The most probable value (“MPM”) • The “EXPECTED” value (mean value of the distribution). Figure 8: NORTH ZONE - Water depth 800m In order to complete the comparison the response of the floater under constant wind and spectral wind has been calculated too. 6 Copyright © 2011 by ASME (the wind participation factor is defined as the ratio between the wind force for a unit wind speed and the generalized mass of the oscillator). Time response for a typical squall and maximum value are shown on Figure 10. 160 140 120 OFFSET [m] 100 80 60 40 20 0 RP100 RP50 RP25 RP10 RP1 MAXIMUM OVER ALL SQUALLS RBD 95% CONFIDENCE INTERVAL RBD SPECTRAL WIND EXPECTED MPM CONSTANT WIND Figure 9: NORTH ZONE - Water depth 1300m SQUALL RESPONSE OSCILLATOR SPECTRUM OF LINEAR Figure 10: Maximum offset and associated direction (red circle) and normalized linear X – Y response (green line) for a typical squall and for given values of the natural period and the damping percentage (T = 300 s, ξ = 6%) Squall Response Index In order to quantify the squall severity and by analogy with the pseudo-spectral acceleration in use in earthquake engineering, a Squall Response Index has been defined as the average value of the pseudo-acceleration associated with the maximum normalized offset Rk (T , ξ ) over the period and damping intervals: In earthquake engineering, the concept of the response spectrum has been used for many years as an effective way to deal with random transient loadings induced by earthquakes and to evaluate maximum seismic responses. By analogy with the earthquake response spectra and for the Gulf of Guinea environment, the squall time histories have been transformed into Squall Response Spectra by a three-step approach: a) Computation of the maximum offset values and their associated directions as responses of parametric single degree of freedom linear oscillators, b) Evaluation of the squall severity by the Squall Response Index as average value of the pseudo-acceleration, c) Determination of the Squall Response Spectra by fitting extreme-value distributions to the parametric values of extreme response for a given return period. Parametric Linear Response For all squalls, {S k , k = 1..n}, defined by their speed 2 2π ∫ ∫ Rk (T , ξ ). T dTdξ For a given squall {S k } , the value of the Squall Response Index 1 SRI k = ∆T .∆ξ SRI k (see Figure 11) is compared with the maximum normalized offset response Rk (T , ξ ) as parametric function of natural period T and damping percentage ξ. The frequency content of the squall time signal is clearly apparent on Figure 11 as it produces greater response of the parametric linear oscillator. The Squall Response Index has been computed for all squalls recorded in the South and North zones of the Gulf of Guinea. The results for the fifty most important squalls in the North and South zones, are plotted in Figure 12 by decreasing severity, showing a great dispersion and a ratio of 1 to 5. Vk (t ) and direction Dk (t ) time histories and for all natural periods T and damping percentages ξ defined within given intervals, Txξ ∈ Tmin , Tmax x ξ min , ξ max , the maximum [ offset ][ ] Rk (T , ξ ) and the associated direction Dk (T , ξ ) have been computed as the maxima values of the time response for a normalized linear oscillator characterized by its natural period T, its damping percentage ξ and a unit wind participation factor 7 Copyright © 2011 by ASME Significant differences have been observed for the Squall Response Spectra for the North and South zones of the Gulf of Guinea in terms of the 100-year extreme offset for parametric values of the natural period (see Figure 13 and Figure 14), particularly within the natural period interval between 200 and 400 s. This observation could possibly be the result of insufficient squall data for the North zone. Figure 11: Parametric maximum offset depending upon natural period T and three different values of the damping percentage ξ (red = 4%, green = 6%, blue = 8%) and Squall Response Index (horizontal black line, for a typical squall Figure 13: NORTH ZONE / 100-year extreme value of the maximum normalized offset R as function of period T and for three different values of damping ξ (red = 4%, green = 6%, blue = 8%) Figure 12: Distribution of the Squall Response Index for the fifty most important squalls in the North and South zones of the Gulf of Guinea Squall Response Spectrum For any given FPSO modal data specified by its natural period T, its modal damping percentage ξ, its generalized mass M and wind participation factor w, the Squall Response Spectra could be defined as the combination of the fitted probability distributions for the 100-year extreme offset and its associated direction : • Extreme Value distribution for the parameterized maximum offset R (T , ξ ) fitted to the offset values Rk (T , ξ ) with a 0.1 threshold for the return period, • Circular von Mises distribution up to the 4th order for the parameterized direction D (T , ξ ) fitted to the direction values values Dk (T , ξ ) associated with the maximum offset Rk (T , ξ ) Figure1 14: SOUTH ZONE / 100-year extreme value of the maximum normalized offset R as function of period T and for three different values of damping ξ (red = 4%, green = 6%, blue = 8%) The offset direction distributions for the North and South zones of the Gulf of Guinea have also significant differences in terms of most probable directions for squall responses as exhibited on Figure 15 and Figure 16 (polar plots of circular direction distributions in geographic coordinates). However, these distributions exhibit a greater stability in terms of sensitivity to natural period and damping coefficient. 8 Copyright © 2011 by ASME conditions. This methodology seems to be promising for spread moored units but requires further work. It could facilitate access to reliable results during early design phase. Figure 15: NORTH ZONE / von Mises circular distribution of the offset direction for T=300s and for three different values of damping ξ (red = 4%, green = 6%, blue = 8%) For detailed design of spread moored units, the common practice nowadays is to rescale the squalls and to consider either the maximum response obtained from the worst squall events or another value extracted from the set of responses. It seems that considering for design the MPM or the expected value of the distribution of response of the rescaled squalls is not appropriate, as response obtained in such way is most often lower than the extrapolation from RBD models. On the other hand, it has been observed that applying the rescaling methodology is a conservative approach if the maximum of response is considered. The resulting conservatism, when compared to the RBD extrapolations, is acceptable and the results obtained are quite close. Thus it seems reasonable to continue to design floating units with the maximum response of the rescaled squalls, although it could be optimized if required. The present studies have been made with a spread moored system in deep water. The case of a weathervaning system or the case of a coastal unit (shallow water) could lead to different results. Additionally, despite the increased number of squall events with respect to present metocean specification it should be highlighted that this number is still too low to derive reliable extreme values. Thus it seems advisable in the present study to consider the upper bound of the 95% confidence interval of the extrapolations. For further research an increased database is needed in order to reduce the uncertainty in selecting the extrapolation distribution and probably the confidence interval. Figure 16: SOUTH ZONE / von Mises circular distribution of the offset direction for T=300s and for three different values of damping ξ (red = 4%, green = 6%, blue = 8%) The Squall Response Spectra can then be used to determine an approximation of the extreme squall response of a FPSO represented by its surge and sway modes using a quadratic modal combination technique but the detailed methodology is not presented here and will be the subject of a future dedicated paper about the Squall Spectral Response. CONCLUSIONS A response-based approach and a spectral approach have been developed to assess the response of spread moored FPSO’s to squalls in West Africa. Using the squall response spectra, the response to squalls is calculated once for all, independently and quickly – for any system to obtain an order of magnitude of response to extreme squalls, and its significance with respect to other design Last, the way of accounting for the non-linearity of response in statistical extrapolation and the validity of the wind coefficients for this kind of application requires further work too. NOMENCLATURE RBD: Response-Based Design RPX: X years return period MPM: Most Probable Maximum. ACKNOWLEDGMENTS The authors wish to thank Bureau Veritas, UTC and Total for permission to publish this paper. Special thanks are addressed to Total Upstream Nigeria, Total E&P Angola and Total E&P Congo for providing access to the in-situ squall wind records. The views expressed are those of the authors, and do not necessarily reflect those of Bureau Veritas, UTC and Total. 9 Copyright © 2011 by ASME REFERENCES [1] Fugro GEOS, 2004. West Africa Gust Joint Industry Project. Phase 1 Final Report. C56110/3219/R1. Confidential. [2] Legerstee, F., François, M., Morandini, C. and Le Guennec, S. (2006). Squall: Nightmare for Designers of Deepwater West African Mooring Systems. Proceedings of the 25th International Conference on Offshore Mechanics and Arctic Engineering, Hamburg, Germany, 4-9 June 2006. OMAE200692328 [3] Jeans, G., Redford, S., Bellamy, I. and Mundy, R., (2008). The WAG Squall Measurement System. Proceedings of the 27th International Conference on Offshore Mechanics and Arctic Engineering, Estoril, Portugal, 15-20 June 2008. OMAE2008-57337. [4] Jeans, G., and Johnson, R., (2011). The WAG platform structural effects study. Proceedings of the 30th International Conference on Offshore Mechanics and Arctic Engineering, Rotterdam, The Netherlands, June 19-24, 2011. OMAE201150138 10 Copyright © 2011 by ASME