Presentation on theme: "Design assessment of dynamic amplification What is measured?"— Presentation transcript:
1 Design assessment of dynamic amplification What is measured? Classification: Statoil internal Status: DraftKvitebjørn JacketDesign assessment of dynamic amplification What is measured?Sverre Haver, Statoil January 2008
4 Top side massSecond order surface process is simulatedLinear springs model platform-soil system
5 First step – eigenvalue analysis Simulated sea states
6 Method used for calculating global characteristic loads Design wave method: There is a well proven metodology available when design wave approach is used for quasi-static problems (See N-003). Kvitebjørn is too sensitive to dynamics for thrusting solely on design wave method and a simple estimate for the dynamic amplification.Time domain solution of equation of motion: Method in principle very adequate for the Kvitebjørn case. However, there is no detailed recipee for how to do such an analyis.Choice of method for predicting design characteristic loads: Design wave method to determine the 10-2 – probability quasistatic response: + Time domain analysis for obtaing a equivalent dynamic amplification factor.
7 Quasistatic and dynamic simulation for hs=14.9m and tp=16s. A considerable resonant response is observed.Gumbel model fitted to sample
9 The dynamic amplification factor is an equivalent factor The dynamic amplification factor is an equivalent factor. It is the ratio between the 95% 3-hour maximum dynamic response and the 95% 3-hour quasistatic response. The extremes do not necessarily coincide in time.
16 Dimensjon i bunn: 50m x 50mDimensjon i topp: 30m x 22,5mTotal lengde: 177,9mVekt: tLegg nedre del:Ø T = 100/60Legg øvre del: Ø T = 100/70Stag nedre del: Ø1200/Ø1300 T=25/30Stag øvre del: Ø900 T = 65
17 Connection Pin Ø2900 x 95/100 A2: 10,4 m B2: 8,9 m A1 og B1: 7.9 m Weld beads c/c 200 mm
18 Dokking av øvre del (JUS) ned i nedre del(JBS) InnstallasjonDokking av øvre del (JUS) ned i nedre del(JBS)Nivellering og evt. jekkingAktivisering av grippereGrouting av hulrom på 165 mm med spesial grout med trykkstyrke på 115 Mpa. Design basert på 80 Mpa. Kraftfordeling60% av trykk kreftene overføres på spiss motstand,og 40% på skjærStrekk krefter overføres på friksjon (skjær)
22 Kvitebjørn platform ble ferdig innstalert 20.05.03
23 Dynamic behaviour of Kvitebjørn jacket structure Classification: Statoil internal Status: DraftDynamic behaviour of Kvitebjørn jacket structureNumerical predictions versus full-scale measurementsDaniel Karunakaran, Subsea7, Stavanger, NorwaySverre Haver, Statoil, Stavanger, Norway
24 Problem Slender steel structure in rather deep water (190m). Utilizing values adopted in design, the largest natural period was 5s.Hydrodynamic loading is non-linear, i.e. for an ocean wave with period 15s load fluctuations will also be experienced for 7.5s, 5.0s, 3.75s, …The wave period of the annual probability design wave is around 15s.At design ”DAF” = xmaxdyn/xmaxstat was estimated to be 1.4 – 1.6.Colour points sensors collecting full scale data. show positions of various
26 Standard deviation – measured versus predicted Utilizing design assumptions regarding topside weight and soil-structure stiffnessConclusion: Predictions are well on the conservative side for the extreme sea states. Most important reason: Less topside weight(?) and stiffer soil-structure interaction.
27 Natural periods – design figures versus observations first winter (Tuned model prepared in 2004/2005.)For tuned model the following actions were taken: 1) Topside weight reduced from 23000tons to 18000tons. (Comment 2008: Not correct!) 2) Soil-structure stiffness increased by a factor (Comment 2008: Too large factor.)
28 Deck displacement January 1 2004, 12:00 – 13:00 hs=12.3m and tp=13.8s Quasistatic (mm) Dynamic (mm)Measured Simulated Measured SimulatedX- AY- AX- BY- B
29 Dynamic leg forces, January 1 2004, 12:00 – 13:00
30 Future challenges: Missed events Observations:Measured waves are not very large, but measured responses are rather large.Simulations using measured wave train do not identify this event.Reasons? A) Transformed measured wave train at leg positions are not proper. B) An important load mechanism is not captured by the simulations.
31 ConclusionsNatural periods estimated from measurements are considerably smaller than those predicted in the design phase, indicating higher soil stiffness and lower deck weight.Bearing in mind all uncertainties associated with the measured values, the numerical simulations using the tuned model compare reasonably well with measurements. Measured total response is typically on the conservative side. (Comment 2008: We are presently working on establishing a more correct tuned model)Simulated quasi-static leg forces compare better with measurements than the resonant response. Reason seems to be to large exitation at the natural frequency in the simulations.There are some few large measured response events which are not reproduced by the present simulation sceme.