1 MARINTEK Statistical Properties of Wave Kinematics and Related Forces Carl Trygve Stansberg MARINTEK/CeSOS , Trondheim, Norway CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling, Trondheim, Norway, 19 December 2005 Test1826 -IrrW -721s k 0 A 0 =0.395 -0,5 0 0,5 0 0,5 1 1,5 U x/U ref 0 k 0 z Linear S econd-order G rue's m ethod W heeler(from linear) W heeler(from m easur) LD V experim ent
21
Embed
MARINTEK 1 Statistical Properties of Wave Kinematics and Related Forces Carl Trygve Stansberg MARINTEK/CeSOS, Trondheim, Norway CeSOS Workshop on Research.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1MARINTEK
Statistical Properties of Wave Kinematics and Related Forces
Carl Trygve StansbergMARINTEK/CeSOS , Trondheim, Norway
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005
- Full-duration 3-hours storm record (1 realisation only, with 1000 wave cycles)
- Simulate linear and second-order
- Study probability distribution of crest heights Ac & velocity peaks Uc
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005
6MARINTEK
Sample time series, linear and second-order elevation
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005
7MARINTEK
Sample time series, linear and second-order free-surface velocity
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005
8MARINTEK
Sample time series, linear and second-order velocityat z=0
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005
9MARINTEK
Probability distributions from simulations, linear and second-order crest heights
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005
10MARINTEK
Probability distributions from simulations, linear and second-order velocity peaks
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005
11MARINTEK
Simplified second-order formula for maximum crest height:
E[Amax] = AR (1 + ½ kp AR )
(from Stansberg, (1998), based on Kriebel & Dawson (1993), Tayfun (1980))
whereAR [ (2 ln (M)) + 0.577/ (2 ln (M))]
kp = wave number = (2fp)2/g
This follows from modified Rayleigh distribution model for short-time statistics of nonlinear peaks a’:
P[A < a’] = 1 -exp [-a2/22]
where a = a’(1- ½ kpa) are the linear crests
This formulation is based on second-order regular wave theory.
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005
12MARINTEK
Simplified second-order formula for maximum crest height (cont.):
The simple formula compares surprisingly well with full second-order simulations, see e.g. Stansberg (1998):
although the negative difference-frequency effects are neglected.
Likely reason: Use of the spectral peak frequency fp probably leads to too long wave periods for the highest crests:
Forristall (2000) suggests fAmax = 1.05fp for the highest crests.
We think it should be even shorter, because in a random simulation it is locally shorter at a high peak than over the whole cyclus (found from Hilbert transform analysis of linear records). Thus we have found fAmax 1.15fp
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005
13MARINTEK
Similarly, we use the same approach in suggesting a simplified second-order formula for high velocity peaks Uc at the free surface
Assume regular waves:
utot(t,z) = u0(t,z) +(∂u0/∂z│z=0)∙z
which can be written, under the crest z=Ac :
uC = u0(1 + kp Ac )
and the peak value distribution function becomes:
P[u < uC ] = 1 - exp [- u02/2u
2] = 1 - exp [- uc2(1- kp Ac )2/2u
2]
(As for the crest heights before, we choose to use kp here, to compensate for the neglecting of difference-frequency terms).
Notice: Nonlinear term is twice as important as for crests
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005
14MARINTEK
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005
Check of simplified crest height distribution(thin line), vs. simulations
15MARINTEK
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005
Check of simplified velocity peak distribution(thin line), vs. simulations
16MARINTEK
Effect on velocity-determined wave forces(introductory study)
Type 1: Wave slamming
F ½ Cs A Uc2
(slamming coefficient Cs 3 – 6, depending on relative angle)
Type 2: Drag forces
F = ½ D CD u |u| (and then integrated up to the free surface)
Here we limit our study to look at properties of the peaks of free-surface velocity squared – indicates the statistical properties of local forces
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005
17MARINTEK
Sample record of u2, linear and second-order kinematics
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005
18MARINTEK
Total 3-hour record of u2, with linear and second-order kinematics (u > 6.5m/s)
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005
Number of “events” increased by several 100%!
19MARINTEK
Probability distributions of u2, linear and second-order kinematics, including comparison to exponential model
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005
20MARINTEK
Probability distribution of u2, with second-order kinematics,
compared to simplified distribution based on:
P[Y < y’] = 1 - exp[-y2/y2]
where
y = u02 (from linear velocity)
y’ = y(1 + kpAc)2 (from nonlinear velocity)
(simplified distribution shown with dashed line)
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005
21MARINTEK
Summary with conclusions
- Statistical properties of particle velocity peaks in steep random waves have been investigated.
- Second-order contributions lead to 30% increased maximum free-surface velocities (while crest heights are increased 15%)
- A simple distribution model, similar to Kriebel & Dawson’s for crests, compares well with the simulated results
- Preliminary studies of effects on resulting wave forces have been made by considering velocity squared.
- The results show considerable contributions from second-order kinematics on forces – almost 100% increase. A simplified, modified exponential distribution model compares well with simulations
- Further work recommended on statistical properties of integrated drag forces, and on related moments around z=0
CeSOS Workshop on Research Challenges in Probabilistic Load and Response Modelling,Trondheim, Norway, 19 December 2005