Scribd
Upload
7 views

Modelling Nonlinear Seas Powerpoint Swan

The document discusses the challenges and achievements in modeling nonlinear seas, focusing on wave generation and the complexities involved, such as incorporating nonlinearity and directionality. It highlights the importance of understanding nonlinear wave interactions, crest height distributions, and the role of wave breaking in design considerations. The ultimate goal is to develop an empirical crest height distribution model that accounts for nonlinear amplification and wave breaking, supported by experimental and theoretical data.

Uploaded by

Gerry Ryan
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
7 views

Modelling Nonlinear Seas Powerpoint Swan

The document discusses the challenges and achievements in modeling nonlinear seas, focusing on wave generation and the complexities involved, such as incorporating nonlinearity and directionality. It highlights the importance of understanding nonlinear wave interactions, crest height distributions, and the role of wave breaking in design considerations. The ultimate goal is to develop an empirical crest height distribution model that accounts for nonlinear amplification and wave breaking, supported by experimental and theoretical data.

Uploaded by

Gerry Ryan
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 49

ITTC, 2014

Copenhagen, Denmark

Modelling Nonlinear Seas


Challenges and achievements

Chris Swan

Fluid Mechanics Section


Department of Civil & Environmental Engineering
Imperial College London
Background: wave generation
Increasingly complex requirements :

Regular (Stokes) waves

Long-crested random waves

Short-crested random waves


Fully incorporating:
- unsteadiness
For some applications - nonlinearity
- directionality
Nonlinear transient wave groups
(representing the most-probable shape of the largest waves
- in a fully nonlinear sense)
Background: industrial trends
• Increasing adoption of reliability criteria
• Based upon crest height / wave load statistics
• Often requiring 10-4 exceedence probabilities

Implies long random simulations

• Concern over abnormally large (Freak / Rogue) waves


• Occur in the tail of the distribution
• Is there some new physics governing these events?

Implies even longer random simulations

• Incorporation of wave breaking


• Kinematics
• Loads (possible step change: slamming)
Background: Scientific understanding
Nonlinear wave-wave interactions:
• Free waves / bound waves, O(a2), well established
• Resonant interactions, O(a3), allow spectral change
• Slow modulation (wave growth)
• Rapid evolution (characteristic of extreme events?)

Consequences:
• Spectral shape may vary (rapidly) in space & time
• Design spectra:
• May apply on average
• But not necessarily local to an extreme wave event
• Spectral broadening larger maximum crests
• Nonlinear evolution change in directional spread
• Directionality is key to wave breaking
breaking is fundamental to design
Wave Basin at Imperial College London

Page55
Page
Calibration of test facilities
(a) Basin calibration:
• Iterative approach to achieve:
• Desired frequency spectrum
• Desired directional spread
• What happens to the spectral evolution?
• Have important nonlinear effects been calibrated out?

(b) Paddle calibration


• Generation of underlying linear wave components
• Sea state will evolve as required
• Ideally based upon a theoretical transfer function
• Effective absorption essential (beach and paddles)

Methodology adopted at Imperial College


Methods of sea state generation
(a) Double summation method (DSM)
• All frequencies in all directions
• Sum over both frequency and direction
• Non-ergodic

(b) Single summation method (SSM)


• Any one frequency component generated in one direction
• Spectrum sub-divided into narrow (but finite) bands
• Within each band:
• Sequential components generated in sequential directions
• OK, but requires high resolution (calibration more difficult)

(c) Random directional method (RDM)


• Any one component in one direction
• Direction of propagation chosen randomly
• Based on normal distribution weighted by DSF
• Easy to incorporate random amplitudes
Generated data: frequency spectra
JONSWAP
- match to target & spatial uniformity

Hs=10m, Tp=16s, σθ=15°


Generated data: directional spreading
• match to target
• linear sea state

σθ=30°

σθ=15°
Crest height distributions
Individual wave components:
• random phases (0 → 2π)
• random amplitudes (Rayleigh distributed)
• direction of propagation, random with weighting based upon DSF

20 x 3-hour seeds Spatial uniformity


Crest height distributions
Laboratory data (Tp=16s, σθ=15°)

Hs=10m, ½Hskp=0.081

Hs=3m, ½Hskp=0.024
Crest height distributions
Laboratory data (Tp=16s, σθ=15°)

Hs=15.0m, ½Hskp=0.122

Hs=12.5m, ½Hskp=0.102
Crest height distributions
Laboratory data (Tp=16s, σθ=15°)

Hs=20.0m, ½Hskp=0.163

Hs=17.5m, ½Hskp=0.142
Crest height statistics
Evolution of distribution with progressively more seeds:
Comparisons to field data
• Analysis of available field data (>5x105 20min records)
• Undertaken within the CresT JIP

Field data: Hs>12m Laboratory data Hs=12.5m


Comparisons with numerical calculations
• Applied to non-breaking wave events
• Numerical calculations based upon focused wave groups
• Undertaken using a fully nonlinear BEM solution

Hs=10m, ½Hskp=0.081 Hs=15m, ½Hskp=0.122


Crest amplifications: Physical explanation
• Local and rapid spectral change
• Movement of energy to the higher frequencies
• Due to third-order resonant interactions
Long-term goal
• An empirical crest height distribution
• Incorporating: - nonlinear amplification
- wave breaking
• Based upon experimental & theoretical input

Hs=15m, ½Hskp=0.122, σθ=15° Hs=17.5m, ½Hskp=0.142, σθ=15°


- limited breaking - significant breaking
Nonlinear amplification and breaking
Critically dependent on steepness & directional spread

Tp=16s, Hs=20.0m, ½Hskp=0.163

Tp=16s, Hs=15.0m, ½Hskp=0.122


Directional analysis: input data

Input data:

η surface elevation
ηx, ηy surface gradients
u, v, w velocity components

Preferred approach: Input data: η, u, v


Methodology: EMEP
Directionality: nonlinear changes
Comparisons to laboratory data (Hs=10m, ½Hskp=0.081)
● σθ=15° ● calculated using the EMEP
● input data: η,u,v ● sea state generated using RDM

Directional spreading function, DSF

σθ vs. f/fp
Directionality: nonlinear changes
Comparisons to laboratory data (Hs=15.0m, ½Hskp=0.122)
● σθ=15° ● calculated using the EMEP
● input data: η,u,v ● sea state generated using RDM

Directional spreading function, DSF

σθ vs. f/fp
Directionality: nonlinear changes
Comparisons to laboratory data (Hs=20.0m, ½Hskp=0.163)
● σθ=15° ● calculated using the EMEP
● input data: η,u,v ● sea state generated using RDM

Directional spreading function, DSF

σθ vs. f/fp
Directionality, alternative quantification
Based upon the velocity reduction factor (VRF)
𝟏
𝟏 𝟏 𝟐 (𝒓𝒎𝒔 𝒖𝒙 𝒊𝒏 𝒅𝒊𝒓𝒆𝒄𝒕𝒊𝒐𝒏𝒂𝒍 𝒔𝒆𝒂)
VRF= + 𝒆𝒙𝒑 −𝟐𝝈𝟐 or VRF= (𝒓𝒎𝒔 𝒖𝒙 𝒊𝒏 𝒖𝒏𝒊−𝒅𝒊𝒓𝒆𝒄𝒕𝒊𝒐𝒏𝒂𝒍 𝒔𝒆𝒂)
𝟐 𝟐
(Tucker & Pitt, 2001)

Based upon:
● the analysis of an
entire wave record

+ the average of a wave-by-


wave analysis
Velocity Reduction Factor (VRF)
• comparisons to laboratory data
• VRF averaged over 20 x 3-hour seeds for each sea state
• changes with Hs

- - - linear input
● velocity ratio
+ earlier EMEP

Conclusion: steepest seas are more uni-directional


Velocity Reduction Factor (VRF)
• Comparisons to laboratory data (Hs=10m, σθ=15°, ½Hskp=0.081)
• VRF calculated for individual waves
• Plotted in terms of the normalised crest elevation, ηc/ηcmax
Velocity Reduction Factor (VRF)
• Comparisons to laboratory data (Hs=15m, σθ=15°, ½Hskp=0.122)
• VRF calculated for individual waves
• Plotted in terms of the normalised crest elevation, ηc/ηcmax
Velocity Reduction Factor (VRF)
• Comparisons to laboratory data (Hs=20m, σθ=15°, ½Hskp=0.163)
• VRF calculated for individual waves
• Plotted in terms of the normalised crest elevation, ηc/ηcmax

Conclusion: largest individual waves more long-crested


Wave breaking: Is it important in the field?
Laboratory data: 10-4 deep water design wave
10-4 design sea state (North Sea): WID event
Evidence of an over-turning wave
Wave breaking
Should be viewed as a process, not a single deterministic event
• constant spectral shape
• constant directional spread
• increased energy levels
Wave breaking: the role of directionality
σθ=0° σθ=15° σθ=30°
The occurrence of wave breaking
• Visual observations allow breaking waves to be identified
• Where breaking is dominant (on average) data is given in red
• With increasing steepness, the tail of the distribution is
controlled by breaking, hence the reduction in crest heights
• Laboratory data relates to σθ=15°

Hs=15m, ½Hskp=0.122, σθ=15° Hs=17.5m, ½Hskp=0.142, σθ=15° Hs=20m, ½Hskp=0.163, σθ=15°


Kinematics measurements

• Laser Doppler Anemometry (LDA).


• Provides time-history at a single point
• u(t), v(t) & w(t)
• Multiple runs to build spatial profiles
• Highest accuracy (better than ±1%)
• Largest data rate (kHz)
• Required seeding density more easily achieved

• Very time consuming


Repeatability of wave records:
Wave Case 1 Wave Case 2

Wave Case 3
Example data records (LDA: Wave Case 2)
Intermittent velocity records: high in wave crest
Horizontal velocities, u(t): Wave Case 1
- various elevations, -20.0m<z<+16.1m
Comparisons to predicted velocities, u(z):
Wave Case 1
– solutions matched to ηmax
Horizontal velocities, u(t): Wave Case 2
- various elevations, -30.0m<z<18.5m
Comparisons to predicted velocities, u(z):
Wave Case 2
– solutions matched to ηmax
Horizontal velocities, u(t): Wave Case 3
- various elevations, -30.0m<z<20.4m
Comparisons to predicted velocities, u(z):
Wave Case 3
– solutions matched to ηmax
Crest kinematics, u(z) for z>0
Concluding Remarks #1
Crest height statistics:
• Very long random wave simulations undertaken
• Significant departures from existing O(a2) design solutions

• Emphasised the importance of:


• Nonlinear amplifications (beyond 2nd order)
• Wave breaking

• Critically dependent upon:


• Sea state steepness (½Hskp)
• Directional spread
• Effective water depth (kpd)

• Largest waves are more uni-directional

• Average shape differs from linear predictions


Rigorous control of generated wave components is essential
Concluding Remarks #2

Kinematics measurements:
• Detailed observations above SWL

• Highlight the inadequacy of the commonly applied solutions

• umax as z→ηc significantly increased


• Relevant for wave slamming & wave-in-deck loading
• Phase velocity is not an effective upper bound: umax>c

• umax for z<SWL reduced


• Relevant for sub-structure loads & sea bed pipelines
• Present designs may be overly conservative
Concluding Remarks #3
Future developments:
Formulation of an empirically based wave model:
• Wave profiles
• Wave crest statistics
• Wave kinematics
Inclusive of nonlinearity, directionality and wave breaking

The true benefits of physical model testing lies in:


• Its combination with theoretical / numerical models
• Not as a simple means of validation / calibration
• Rather:
• The provision of the underlying physics
• The identification of critical effects

Taken together they can provide solutions to some of the most


challenging wave & wave-loading problems

You might also like