<?xml version='1.0' encoding='utf-8' ?>
<!DOCTYPE html>
<html xmlns="http://www.w3.org/1999/xhtml">
	<head>
		<meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
		<link type="text/css" rel="stylesheet" href="doxygen.css"/>
		<link type="text/css" rel="stylesheet" href="tabs.css"/>
	</head>
	<body>
	<div>
<!-- Generated by Doxygen 1.8.5 -->
</div><!-- top -->
<div class="header">
  <div class="headertitle">
<div class="title">Load Models </div>  </div>
</div><!--header-->
<div class="contents">
<div class="toc"><h3>Table of Contents</h3>
<ul><li class="level1"><a href="#tm_load_overview">Overview of Load Effects</a></li>
<li class="level1"><a href="#tm_load_hydrostatic">Hydrostatic Pressure Effects</a></li>
<li class="level1"><a href="#tm_load_wave">Wave Potential</a><ul><li class="level2"><a href="#tm_load_wave_regular">Regular Waves</a></li>
<li class="level2"><a href="#tm_load_wave_irregular">Irregular Waves</a></li>
<li class="level2"><a href="#tm_load_wave_kinematics">Wave Kinematics Including Diffraction Effects</a></li>
<li class="level2"><a href="#tm_load_wave_second">Second Order Stokes' Irregular Waves</a><ul><li class="level3"><a href="#tm_load_wave_second_stokes">Stokes' theory</a><ul><li class="level4"><a href="#tm_load_wave_second_stokes_potential">The wave potential</a></li>
<li class="level4"><a href="#tm_load_wave_second_stokes_boundary">Boundary conditions</a></li>
<li class="level4"><a href="#tm_load_wave_second_stokes_series">Series expansion of the surface elevation</a></li>
<li class="level4"><a href="#tm_load_wave_second_stokes_first">First order solution</a></li>
<li class="level4"><a href="#tm_load_wave_second_stokes_second">Second order solution</a></li>
<li class="level4"><a href="#tm_load_wave_second_stokes_sum">The sum-frequency and difference-frequency transfer functions</a></li>
<li class="level4"><a href="#tm_load_wave_second_stokes_diagonal">The diagonal of the difference-frequency kernels</a></li>
<li class="level4"><a href="#tm_load_wave_second_stokes_high">High frequency contributions from the sum-frequency transfer functions</a></li>
</ul>
</li>
<li class="level3"><a href="#tm_load_wave_second_spectrum">Spectrum</a><ul><li class="level4"><a href="#tm_load_wave_second_spectrum_linear">The linear wave and the filtering procedure</a></li>
<li class="level4"><a href="#tm_load_wave_second_spectrum_separability">Separability of the directional spectrum</a></li>
</ul>
</li>
<li class="level3"><a href="#tm_load_wave_second_kinematics">Kinematics</a><ul><li class="level4"><a href="#tm_load_wave_second_kinematics_convection">Convection terms</a></li>
<li class="level4"><a href="#tm_load_wave_second_kinematics_near">Kinematics near to and above the mean surface level</a></li>
<li class="level4"><a href="#tm_load_wave_second_kinematics_above">Kinematics above the sea surface</a></li>
<li class="level4"><a href="#tm_load_wave_second_kinematics_below">Kinematics on the surface and below the mean surface level</a></li>
</ul>
</li>
</ul>
</li>
</ul>
</li>
<li class="level1"><a href="#tm_load_current">Current Description</a></li>
<li class="level1"><a href="#tm_load_hydrodynamic_models">Hydrodynamic Load Models for Submerged Elements</a><ul><li class="level2"><a href="#tm_genmori">Generalized Morison's Equation</a></li>
<li class="level2"><a href="#tm_mfloads">MacCamy-Fuchs with Quadratic Drag</a></li>
</ul>
</li>
<li class="level1"><a href="#tm_load_forced">Forced Vessel Motion</a><ul><li class="level2"><a href="#tm_load_forced_hf">HF Vessel Motion Model</a></li>
<li class="level2"><a href="#tm_load_forced_lf">LF Vessel Motion Model</a></li>
<li class="level2"><a href="#tm_load_forced_frequency">Frequency-domain Vessel Motion Analysis</a></li>
</ul>
</li>
<li class="level1"><a href="#tm_load_effects">Effects from Internal Fluid Flow</a></li>
<li class="level1"><a href="#tm_load_hydrodynamic_on">Hydrodynamic Load on Partly Submerged, Floating Elements</a><ul><li class="level2"><a href="#tm_load_hydrodynamic_on_potential">Potential Flow Contribution</a></li>
<li class="level2"><a href="#tm_load_hydrodynamic_on_simplified">Simplified Approach</a></li>
<li class="level2"><a href="#tm_load_hydrodynamic_on_viscous">Viscous Forces</a></li>
<li class="level2"><a href="#tm_load_hydrodynamic_on_moment">Moment around Local x-axis</a></li>
<li class="level2"><a href="#tm_load_hydrodynamic_on_discussion">Discussion</a></li>
<li class="level2"><a href="#tm_load_hydrodynamic_on_implementation">Implementation in Finite Element Formulation</a></li>
</ul>
</li>
<li class="level1"><a href="#aerodynamicLoads">Aerodynamic loads</a><ul><li class="level2"><a href="#AeroElements">Lift forces, drag forces, and torsional moments on elements</a><ul><li class="level3"><a href="#MorisonAeroDrag">Morison-type aerodynamic drag loads (CHTYPE=MORI)</a></li>
<li class="level3"><a href="#liftDragAirfoils">Lift forces, drag forces, and torsional moments on elements (CHTYPE=AIRF)</a></li>
</ul>
</li>
<li class="level2"><a href="#BEM">Blade element/momentum (BEM) theory for wind turbines</a><ul><li class="level3"><a href="#outlineOfBEM">Outline of BEM calculation</a></li>
<li class="level3"><a href="#BEMCorrectionFactors">Correction factors</a><ul><li class="level4"><a href="#DynamicWake">Dynamic wake</a></li>
<li class="level4"><a href="#DynamicStall">Dynamic stall</a></li>
<li class="level4"><a href="#SkewedInflow">Skewed inflow</a></li>
<li class="level4"><a href="#TowerInfluence">Upwind tower influence</a></li>
<li class="level4"><a href="#DWTowerInfluence">Downwind tower influence</a></li>
</ul>
</li>
<li class="level3"><a href="#BEMLimitations">BEM Limitations</a></li>
</ul>
</li>
</ul>
</li>
<li class="level1"><a href="#tm_load_references">References</a></li>
</ul>
</div>
<div class="textblock"><p><a class="anchor" id="load_models"></a></p>
<h1><a class="anchor" id="tm_load_overview"></a>
Overview of Load Effects</h1>
<p>The following physical effects contribute to system loads:</p>
<ul>
<li>weight and inertia, governed by line mass, including any pipe contents and external wrapping</li>
<li>hydrostatic forces, dependent on pressure gradients</li>
<li>hydrodynamic forces, dependent on wave, current and structure motions</li>
<li>forced motion of line, dependent on vessel motions</li>
<li>aerodynamic forces, dependent on wind and structure motions</li>
</ul>
<p>Since the analysis is based on a beam model with pressure-independent cross section properties, the hydrostatic pressure level does not affect the static or dynamic equilibrium condition, hence the pressure gradients are sufficient.</p>
<p>All loads are applied either as distributed line loads (force/length) or as point loads.</p>
<p>The following sections give an outline of essential features of the implemented load models.</p>
<p>With regard to surface-piercing structures, two classes of line components are used:</p>
<ol type="1">
<li>Hydrostatic and hydrodynamic loads are applied on a length equal to the submerged length of the component axis, i.e. the volume forces are concentrated in the pipe axis. Referring to the user manual this applies to components CRS0, CRS1 and CRS2. Hydrodynamic loads models are described in section <a class="el" href="a00014.html#tm_load_hydrodynamic_models">Hydrodynamic Load Models for Submerged Elements</a>.</li>
<li>Hydrostatic and hydrodynamic forces are distributed according to the actual diameter, or shape of the buoyancy volume. Thus, the cross section can be partly submerged, and structures such as floating hoses and fish cages can be modelled. Referring to the user manual this applies to components CRS3, CRS4 and CRS5. Hydrodynamic load models are described in section <a class="el" href="a00014.html#tm_load_hydrodynamic_on">Hydrodynamic Load on Partly Submerged, Floating Elements</a>.</li>
</ol>
<p>For submerged structures, and also for surface-piercing normal pipes and cables, the first class is used, and generally this cross section is referred to in the subsequent discussion. The second class is referred to as a "partly submerged" cross section.</p>
<h1><a class="anchor" id="tm_load_hydrostatic"></a>
Hydrostatic Pressure Effects</h1>
<p>Hydrostatic pressure effects on marine risers have been discussed by several authors in the open literature, e.g. Morgan and Peret (1974), Sparks (1984) and Mathisen and Bergan (1986). This discussion normally involves the following notions that need to be defined.</p>
<ul>
<li>effective tension: force in the pipe that affects stability. This force is relevant for governing the shape of cables and pipes, including buckling analysis, and is used for calculation of geometric stiffness in the finite element method.</li>
<li>axial stress resultant: force found by integrating normal stresses over the cross section. In the presence of external hydrostatic pressure, this force is different from the effective tension.</li>
</ul>
<p>In principle, two different approaches are possible when analysing a pipe or other slender structure subjected to external and internal hydrostatic pressure.</p>
<ol type="1">
<li><p class="startli">Volume force model</p>
<p class="startli">Use of vertical, conservative forces to represent hydrostatic effects. These forces will be in equilibrium with the effective tension, which means that axial stresses need not necessarily be calculated during an iteration for equilibrium. The theoretical foundation for this way of modelling hydrostatic forces is given by Sparks (1984).</p>
</li>
<li><p class="startli">Pressure force model</p>
<p class="startli">Calculation of non-conservative hydrostatic forces by considering pressure on the deformed pipe geometry. These forces are in equilibrium with the axial stress resultants. Effective tension needed for calculation of the stiffness matrix must, however, be found by introducing artificial pressure forces in axial direction. This method is described by Mathisen and Bergan (1986).</p>
</li>
</ol>
<p>Both methods are in principle correct, and all nonlinear geometric and material effects can be modelled within both concepts. The choice of method should therefore be made after considering use of computer time and complexity in formulation of the alternatives. For a structure where hydrostatic pressure does not affect the volume or cross section properties the first approach is correct, and it is computationally much simpler. The conclusion is obvious: in the present program hydrostatic effects will be modelled by conservative, vertical forces. No hydrostatic force variation for caused by deformations will be accounted, except for possible increase or decrease of submerged volume and wetted pipe surface at the sea surface.</p>
<p>Hydrostatic pressure is therefore treated in terms of effective weight and effective tension, defined as:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ w = m_{p}g-A_{e}\rho g + A_{y} \rho_{i} g \tag{7.1} \]" src="form_602.png"/>
</p>
<p><a class="anchor" id="tm_eq_7_2"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ T = T_{p} + A_{e} P_{e} - A_{i} P_{i} - \rho_{i} A_{i} v^2_{i} \tag{7.2} \]" src="form_603.png"/>
</p>
<p>where</p>
<ul>
<li><img class="formulaInl" alt="$ T \,\qquad $" src="form_604.png"/> : effective tension</li>
<li><img class="formulaInl" alt="$ T_{p} \!\qquad $" src="form_605.png"/> : tension in pipe wall, i.e. resulting force from normal stresses</li>
<li><img class="formulaInl" alt="$ A_{e}, A_{i} $" src="form_606.png"/> : external/internal cross sectional</li>
<li><img class="formulaInl" alt="$ P_{e}, P_{i} \: $" src="form_607.png"/> : external/internal hydrostatic pressure</li>
<li><img class="formulaInl" alt="$ w \:\qquad $" src="form_608.png"/> : effective weight per unit length, i.e. submerged weight of pipe including content</li>
<li><img class="formulaInl" alt="$ m_{p} \!\!\qquad $" src="form_609.png"/> : mass of pipe per unit length</li>
<li><img class="formulaInl" alt="$ \rho_{i} \,\qquad $" src="form_610.png"/> : density of internal fluid</li>
<li><img class="formulaInl" alt="$ \rho \;\, \qquad $" src="form_611.png"/> : water density</li>
<li><img class="formulaInl" alt="$ g \;\, \qquad $" src="form_612.png"/> : acceleration of gravity</li>
<li><img class="formulaInl" alt="$ v_{i} \;\!\qquad $" src="form_613.png"/> : velocity of internal fluid flow</li>
</ul>
<p>The basic axial force output from static and dynamic analysis is the effective tension. The axial pipe wall force can, if required, be calculated by post processing according to <a class="el" href="a00014.html#tm_eq_7_2">Eq. (7.2)</a>. The last term in <a class="el" href="a00014.html#tm_eq_7_2">Eq. (7.2)</a>, which results from a constant velocity internal fluid flow, is further discussed in <a class="el" href="a00014.html#tm_load_effects">Effects from Internal Fluid Flow</a>.</p>
<h1><a class="anchor" id="tm_load_wave"></a>
Wave Potential</h1>
<h2><a class="anchor" id="tm_load_wave_regular"></a>
Regular Waves</h2>
<p>Regular waves are modelled by two alternative methods</p>
<ul>
<li>Airy linear wave theory</li>
<li>Stokes' 5th order wave theory</li>
</ul>
<p>Stokes' theory is outlined by Skjelbreia and Hendrickson (1960). Both theories describe regular, long crested waves propagating in an arbitrary direction. The main difference between the theories is that the boundary condition pressure ( <img class="formulaInl" alt="$ p = 0.0 $" src="form_614.png"/> ) is valid at mean water level for the linear theory, but is approximately correct at the wave surface for Stokes' theory. This implies that the linear wave theory is strictly valid for infinitesimal wave amplitudes only, while Stokes' theory describes a wave of finite height. Wave-induced velocities and acceleration in a wave crest can therefore consistently be described by Stokes' theory only. The wave profile defined by the two theories for waves of identical lengths and heights are shown in figure <a class="el" href="a00014.html#Comparison_of_wave_profiles">Comparison of wave profiles</a>.</p>
<p>The wave potential <img class="formulaInl" alt="$ \displaystyle \phi_{0} $" src="form_615.png"/> for a regular wave according to Airy's theory can be expressed as follows:</p>
<p><a class="anchor" id="tm_eq_7_3"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \phi_0 = \frac{\zeta_{a}g}{\omega}C_1 \cos(-\omega t + kX\cos{\beta} + kY\sin{\beta} + \psi_{\zeta}) \tag{7.3} \]" src="form_616.png"/>
</p>
<p>where</p>
<ul>
<li><img class="formulaInl" alt="$ \zeta_{a} $" src="form_617.png"/> : wave amplitude</li>
<li><img class="formulaInl" alt="$ g \;\, $" src="form_618.png"/> : acceleration of gravity</li>
<li><img class="formulaInl" alt="$ k \;\, $" src="form_619.png"/> : wave number</li>
<li><img class="formulaInl" alt="$ \beta \;\, $" src="form_620.png"/> : direction of wave propagation. ( <img class="formulaInl" alt="$ \beta = 0 $" src="form_621.png"/> corresponds to wave propagation along the positive X-axis)</li>
<li><img class="formulaInl" alt="$ \psi_{\zeta} $" src="form_622.png"/> : a phase angle lag</li>
</ul>
<p><a class="anchor" id="Comparison_of_wave_profiles"></a></p>
<div class="image">
<img src="RIFLEX_TheoryManual_42_v0_708.svg" alt="RIFLEX_TheoryManual_42_v0_708.svg"/>
<div class="caption">
Comparison of wave profiles</div></div>
<p><img class="formulaInl" alt="$ C_\text{1} $" src="form_623.png"/> is given by:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ C_{1} = \frac{\textrm{cosh}\left( k(Z + D) \right) }{\textrm{cosh}(kD)} \tag{7.4} \]" src="form_624.png"/>
</p>
<p>where <img class="formulaInl" alt="$ D $" src="form_625.png"/> is the water depth.</p>
<p>In deep water <img class="formulaInl" alt="$ C_{1} $" src="form_626.png"/> can be approximated by</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ C_{1} = \textrm{e}^{kZ} \tag{7.5} \]" src="form_627.png"/>
</p>
<p>We then obtain the following relations for the particle velocities and acceleration in the undisturbed wave field:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{aligned} v_{x} &amp;= -\zeta_{a} \omega \cos(\beta) C_{2} \sin(\psi) \\ v_{y} &amp;= -\zeta_{a} \omega \sin(\beta) C_{2} \sin(\psi) \\ v_{z} &amp;= -\zeta_{a} \omega C_{3} \cos(\psi) \qquad \\ a_{x} &amp;= \zeta_{a} \omega^2\cos(\beta) C_{2} \cos(\psi) \\ a_{y} &amp;= \zeta_{a} \omega^2 \sin(\beta) C_{2} \cos(\psi) \\ a_{z} &amp;= -\zeta_{a} \omega^2 C_{3} \sin(\psi) \end{aligned} \tag{7.6} \]" src="form_628.png"/>
</p>
<p>where</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \psi = -\omega t + kX\cos(\beta) + kY\sin(\beta) + \psi_{\zeta} \tag{7.7} \]" src="form_629.png"/>
</p>
<p>Using the deep water approximation we have:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ C_{1} = C_{2} = C_{3} = \textrm{e}^{kZ} \tag{7.8} \]" src="form_630.png"/>
</p>
<p>Taking into account finite water depth we have:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{aligned} C_{1}&amp; = \frac{\textrm{cosh}(k(Z+D))}{\textrm{cosh}(kD)}\\ C_{2}&amp; = \frac{\textrm{cosh}(k(Z+D))}{\textrm{sinh}(kD)}\\ C_{3}&amp; = \frac{\textrm{sinh}(k(Z+D))}{\textrm{sinh}(kD)}\\ \end{aligned} \tag{7.9} \]" src="form_631.png"/>
</p>
<p>The surface elevation is given by:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \zeta = -\zeta_{a} \sin\psi = \zeta_{a} \sin (\omega t - kX\cos(\beta) - kY\sin(\beta) + \phi ) \tag{7.10} \]" src="form_632.png"/>
</p>
<p>where <img class="formulaInl" alt="$ \phi = -\psi $" src="form_633.png"/> . The forward phase shift, <img class="formulaInl" alt="$ \phi $" src="form_634.png"/> , is denoted "phase angle" in the program. Similarly the linearized dynamic pressure is given by:</p>
<p><a class="anchor" id="tm_eq_7_11"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ p_{d} = \rho g\zeta_{a}C_{1}\sin(\psi) \tag{7.11} \]" src="form_635.png"/>
</p>
<p>The wave potential described here is, as previously stated, valid to mean water level only. Prediction of velocities and acceleration in the wave crest is, however, important. The present program has four different methods for handling this problem for linear waves. These are illustrated in <a class="el" href="a00014.html#Methods_for_use_of_wave_potential_close_to_surface">Methods for use of wave potential close to surface</a>.</p>
<p><a class="anchor" id="Methods_for_use_of_wave_potential_close_to_surface"></a></p>
<div class="image">
<img src="RIFLEX_TheoryManual_42_v0_729.svg" alt="RIFLEX_TheoryManual_42_v0_729.svg"/>
<div class="caption">
Methods for use of wave potential close to surface.</div></div>
<p>Method 1 is the consistent use of linear wave theory assuming that the water level always remains at mean water level, referring to the figure <a class="el" href="a00014.html#Methods_for_use_of_wave_potential_close_to_surface">Methods for use of wave potential close to surface</a>.</p>
<p>This method is included in order to have a reference point, particularly for comparing results with frequency domain solutions.</p>
<p>Method 2 is based on the assumption that the potential is correct at some distance from the surface, and that calculated surface values are correct at the instantaneous surface. Consequently, the potential profile is deformed by stretching and compression according to the following formula (Wheeler 1970):</p>
<p><a class="anchor" id="tm_eq_7_12"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ Z' = D\frac{Z-\zeta}{D+\zeta} \tag{7.12} \]" src="form_636.png"/>
</p>
<p><img class="formulaInl" alt="$ Z' \quad $" src="form_637.png"/> - modified Z-coordinate to be introduced in place of Z in equations (<a class="el" href="a00014.html#tm_eq_7_3">7.3</a> ) - (<a class="el" href="a00014.html#tm_eq_7_11">7.11</a>) <br/>
 <img class="formulaInl" alt="$ \zeta^\textrm{ }\quad\,\, $" src="form_638.png"/> - actual surface elevation <br/>
 <img class="formulaInl" alt="$ D^\textrm{ }\quad $" src="form_639.png"/> - water depth <br/>
</p>
<p>Method 3 is a parallel move of potential, assuming that the potential itself is correctly described by linear theory, but valid from the instantaneous sea surface. Note that this strategy does not include a re-definition of the potential from variation of effective water depth.</p>
<p>Method 4 allows the user to apply the wave kinematics up to the instantaneous free surface without stretching: for Z levels above mean water level, results from Z = 0 are used.</p>
<p>Using regular wave models, wave-induced velocities and acceleration are calculated at every node at every time step during the time integration. No storage of such data is carried out. During this process, updating of node position can easily be considered simply by calculating the wave potential at the instantaneous structural position determined by the static position plus dynamic displacement.</p>
<p>The program offers this possibility in addition to the traditional method using the static node positions as reference for calculation of wave kinematics. These alternatives are illustrated in <a class="el" href="a00014.html#Static_and_dynamic_node_positions">Static and dynamic node positions</a>.</p>
<p>If a major part of the structural motion is due to linear, wave-induced support vessel motion, it is recommended to use wave kinematics at the static position. Otherwise, the forced terminal point motions and the wave kinematics will be inconsistent.</p>
<p><a class="anchor" id="Static_and_dynamic_node_positions"></a></p>
<div class="image">
<img src="RIFLEX_TheoryManual_42_v0_734.svg" alt="RIFLEX_TheoryManual_42_v0_734.svg"/>
<div class="caption">
Static and dynamic node positions.</div></div>
<h2><a class="anchor" id="tm_load_wave_irregular"></a>
Irregular Waves</h2>
<p>The present section describes the modelling of an undisturbed Airy wave field which is normally used as the basis for wave loading on slender structures. For modelling of disturbed wave kinematics near the support vessel, see section <a class="el" href="a00014.html#Wave_Kinematics_Including_Diffraction_Effects">Wave Kinematics Including Diffraction Effects</a>.</p>
<p>An irregular sea state is described as a sum of two wave spectra: a wind sea contribution and a swell contribution:</p>
<p><a class="anchor" id="tm_eq_7_13"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ S_{\zeta, \textrm{TOT}}(\beta ,\omega ) = S_{\zeta , {1}}(\omega )\phi_{1}(\beta - \beta_{1}) + S_{\zeta , {2}}(\omega) \phi_{2} (\beta - \beta_{2}) \tag{7.13} \]" src="form_640.png"/>
</p>
<p>in which <img class="formulaInl" alt="$ S_{\zeta , {1}} $" src="form_641.png"/> and <img class="formulaInl" alt="$ S_{\zeta , {2}} $" src="form_642.png"/> describe the frequency distribution of wind sea and swell, respectively. Several standard wave spectra are included, such as Pierson-Moscowitz, JONSWAP, etc., as well as a general numeric description which can be used for description of measured data or regular waves (see Input to INPMOD in the User Manual).</p>
<p><img class="formulaInl" alt="$ \phi_{1} $" src="form_643.png"/> and <img class="formulaInl" alt="$ \phi_{2} $" src="form_644.png"/> describe the directionality of the waves. Unidirectional waves and cosine spreading functions are included.</p>
<p><img class="formulaInl" alt="$ \beta $" src="form_463.png"/> is the direction of wave propagation.</p>
<p><a class="anchor" id="Definition_of_wave_directions"></a></p>
<div class="image">
<img src="RIFLEX_TheoryManual_42_v0_741.svg" alt="RIFLEX_TheoryManual_42_v0_741.svg"/>
<div class="caption">
Definition of wave directions.</div></div>
<p>The spectrum directionality parameters satisfy the relations:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{array}{l} \displaystyle \int_{-\frac{\pi}{2}}^\frac{\pi}{2}\!\phi_\textrm{j}(\beta) \textrm{d}\beta = 1.0\\ \\ \displaystyle \phi_\textrm{j} (\beta) = 0,\, \frac{\pi}{2} < \beta < \frac{3\pi}{2} \end{array}\tag{7.14} \]" src="form_645.png"/>
</p>
<p><br/>
</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \int_{0}^{\infty}\!{S_{\zeta , 1}}(\omega)\textrm{d}{\omega} + \int_{0}^{\infty}\!{S_{\zeta , 2}}(\omega)\textrm{d}{\omega} = \sigma_{\zeta}^2 \tag{7.15} \]" src="form_646.png"/>
</p>
<p>In which <img class="formulaInl" alt="$ \sigma_{\zeta}^2 $" src="form_647.png"/> is the variance of the surface elevation.</p>
<p>In order to generate time series of surface elevation, water particle velocities and acceleration, the short-crested irregular sea is discretized into a set of harmonic components. Thus, in complex notation the surface elevation is expressed by:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{array}{l} \displaystyle Z_\zeta = \sum_{\textrm{j} = 1}^{N_\beta}\, \sum_{\textrm{k} = 1}^{N_\infty}\, Z_\textrm{jk} = \sum_{\textrm{j} =1}^{N_\beta}\, \sum_{\textrm{k} =1}^{N_\infty}\, A_\textrm{jk}^{e^{i(\omega_\textrm{k} t + \phi_\textrm{jk}^p + \phi_\textrm{jk} )}} \\ \\ \displaystyle A_\textrm{jk} = |Z_\textrm{jk}| = \sqrt{2S_\zeta (\beta_\textrm{j}, \omega_\textrm{k}) \Delta\beta\Delta\omega} \\ \\ \arg(Z_\textrm{jk} ) = \omega_\textrm{k}t + \phi_\textrm{jk}^p + \phi_\textrm{jk} \end{array} \tag{7.16} \]" src="form_648.png"/>
</p>
<p>The random phase angles, <img class="formulaInl" alt="$ \phi_\textrm{jk} $" src="form_649.png"/> , are sampled from a uniform distribution over <img class="formulaInl" alt="$ [ -\pi, \pi] $" src="form_650.png"/> . The position-dependent phase angle is:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \phi_\textrm{jk}^p = - k _\textrm{k}X\cos(\beta_\textrm{j}) - k_\textrm{k} Y \sin(\beta_\textrm{j}) \tag{7.17} \]" src="form_651.png"/>
</p>
<p>The surface elevation can be expressed as:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{array}{l} \displaystyle \zeta (t) = \textrm{Im}(Z_\zeta) = \textrm{Re}(Z_\zeta \cdot \textrm{e}^{-i\pi/2})\\ \displaystyle = \sum_{\textrm{j} =1}^{N_\beta}\, \sum_{\textrm{k} = 1}^{N_\infty}\, A_\textrm{jk} \sin (\omega_\textrm{k}t + \phi_\textrm{jk}^p + \phi_ \textrm{jk} ) \end{array} \tag{7.18} \]" src="form_652.png"/>
</p>
<p>The velocity and acceleration components are derived from the surface elevation components.</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{array}{l} \dot{Z}_\textrm{jk} = iw_\textrm{k} Z_\textrm{jk} \\ \\ \ddot{Z}_\textrm{jk} = -w^2_\textrm{k} Z_\textrm{jk} \end{array} \tag{7.19} \]" src="form_653.png"/>
</p>
<p>Adding the harmonic components to obtain time series is performed by a Fourier Transformation algorithm. This approach implies a set of relations between time increment, <img class="formulaInl" alt="$ \Delta t $" src="form_654.png"/> , frequency increment, <img class="formulaInl" alt="$ \Delta \omega $" src="form_655.png"/> , number of time steps, <img class="formulaInl" alt="$ N_t $" src="form_409.png"/> and number of components, <img class="formulaInl" alt="$ N_\omega $" src="form_656.png"/>.</p>
<p><a class="anchor" id="tm_eq_7_20"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \Delta\omega = 2\pi/(N_{t} \Delta t) \tag{7.20} \]" src="form_657.png"/>
</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ N_\omega = (N_t/2) +1 \tag{7.21} \]" src="form_658.png"/>
</p>
<p>Thus the duration of the time series is limited to:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ T = N_t \Delta t \cong 2N_\omega \Delta t\tag{7.22} \]" src="form_659.png"/>
</p>
<p>The number of non-zero components, <img class="formulaInl" alt="$ n_\omega $" src="form_660.png"/> , is, however, usually much less than <img class="formulaInl" alt="$ N_\omega $" src="form_656.png"/> due to the limited range of wave spectra which is of interest. Typically only wave periods larger than <img class="formulaInl" alt="$ t_\textrm{lim} \cong 2-3$" src="form_661.png"/>s. are of interest. With a sampling interval, <img class="formulaInl" alt="$ \Delta t = 0.5 $" src="form_662.png"/>s, the number of non-zero components is:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ n_\omega = N_\omega \Delta t / t_\textrm{lim} \cong 0.25 N_\omega \tag{7.23} \]" src="form_663.png"/>
</p>
<p>The frequency increment and the number of components are chosen so that the time increment and the number of time steps coincide with the wave frequency vessel motion simulation parameters, <img class="formulaInl" alt="$ \Delta t_{t\textrm{HF}} $" src="form_664.png"/> and <img class="formulaInl" alt="$ N_{t\textrm{HF}} $" src="form_665.png"/> , see section <a class="el" href="a00014.html#HF_Vessel_Motion_Model">HF Vessel Motion Model</a>.</p>
<p>In order to increase the duration of the simulation period, several sequences, each of duration <img class="formulaInl" alt="$ T $" src="form_666.png"/> , can be generated. For each new sequence a new set of random phase angles, <img class="formulaInl" alt="$ \psi_{\zeta \, \textrm{jk}} $" src="form_667.png"/> , is sampled.</p>
<p>Velocities, acceleration and surface elevation are calculated for predefined positions using an FFT algorithm. Two different strategies are possible if changed node positions should be considered in a dynamic analysis involving irregular seas.</p>
<ul>
<li>Calculation of velocities and acceleration by FFT in a four dimensional grid and interpolation during dynamic analysis to obtain values at wanted positions and time steps.</li>
<li>Calculation of velocities and acceleration by using Fourier series and direct calculation of all harmonic components with traditional sine and cosine functions for every wave component at wanted positions and time increments.</li>
</ul>
<p>The first strategy has been implemented.</p>
<p>In the present program, velocities and acceleration are calculated for pre-defined points that coincide with nodes in the finite element model at static equilibrium position, but not necessarily covering all nodes. This is feasible since potential profiles are smooth, which makes interpolation of intermediate values an attractive alternative to calculation and storage for every node. In addition, this interpolation is necessary in any case if stretching of the potential profiles is performed in order to describe water particle motions close to the sea surface.</p>
<p>Three options are available with respect to wave particle motion modelling, and, hence, wave force modelling.</p>
<ol type="1">
<li>Calculate forces to still water level, applied on the part of the riser that is submerged in static equilibrium position.</li>
<li>Calculate forces to still water level, update line submergence according to vertical motion, see method 1 in figure <a class="el" href="a00014.html#Methods_for_use_of_wave_potential_close_to_surface">Methods for use of wave potential close to surface</a>.</li>
<li>Calculate forces to actual, instantaneous water level by stretching/shrinking the z-coordinate, see method 2 in figure <a class="el" href="a00014.html#Methods_for_use_of_wave_potential_close_to_surface">Methods for use of wave potential close to surface</a> (Wheeler's method).</li>
<li>Calculate forces to actual, instantaneous water level by shifting the z-axis upwards or downwards, following the actual wave surface, see method 3 in figure <a class="el" href="a00014.html#Methods_for_use_of_wave_potential_close_to_surface">Methods for use of wave potential close to surface</a>.</li>
</ol>
<p>Since the wave-induced loads are most important for the upper part, and even there may be of minor importance for slender lines, such as anchor lines, options for calculating wave kinematics only down to a specified z-coordinate, and even omitting the wave kinematics are included.</p>
<p><a class="anchor" id="Wave_Kinematics_Including_Diffraction_Effects"></a></p>
<h2><a class="anchor" id="tm_load_wave_kinematics"></a>
Wave Kinematics Including Diffraction Effects</h2>
<p>Diffraction effects are described in terms of wave kinematics transfer functions relating wave particle velocities in x, y and z-directions to the undisturbed surface elevation.</p>
<p>The transfer functions for calculation of distributed wave velocities and accelerations <img class="formulaInl" alt="$ u_\textrm{j}^d $" src="form_668.png"/> and <img class="formulaInl" alt="$ \dot{u}_\textrm{j}^d $" src="form_669.png"/> , respectively, are defined as:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ H_ {u_\textrm{j}^d}(\beta,\omega) = \frac{u_\textrm{j}^d (\beta, \omega)}{\zeta_a (\beta,\omega)} \tag{7.24} \]" src="form_670.png"/>
</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ H_ {\dot{u}_\textrm{j}^d}(\beta,\omega) = \frac{\dot{u}_\textrm{j}^d (\beta, \omega)}{\zeta_a (\beta,\omega)} \tag{7.25} \]" src="form_671.png"/>
</p>
<p>where <img class="formulaInl" alt="$ \textrm{j} = 1,2,3 $" src="form_672.png"/> signifies <img class="formulaInl" alt="$ x $" src="form_163.png"/> , <img class="formulaInl" alt="$ y$" src="form_673.png"/> and <img class="formulaInl" alt="$ z$" src="form_674.png"/> directions, respectively.</p>
<p>Consistent vessel motion transfer functions and wave kinematics transfer functions for given locations in the vessel coordinate system are available from sink/source computer programs. The sink/source computations are normally carried out in the following major operations:</p>
<ol type="1">
<li>Solution of global system of equations to determine unknown source intensities for all frequencies and wave directions considered.</li>
<li>Computations of vessel motion transfer functions based on results from 1).</li>
<li>Calculation of wave kinematics transfer functions based on results from 1).</li>
</ol>
<p>In slender structure applications it will normally be sufficient to consider wave kinematics transfer functions for a few locations along the structure near the vessel (3-5 locations will tentatively be sufficient). For a limited number of wave kinematics transfer functions, the computation time for step 3) is small when compared to the computational efforts required in step 1).</p>
<p>In order to take advantage of the possibility of splitting the transfer function computations as described above, the following strategy is applied:</p>
<ol type="1">
<li>Vessel motion transfer functions are given as input to INPMOD (i.e. results from step 1) and 2) in transfer function calculations).</li>
<li>The static configuration is computed by running STAMOD. Wave kinematics transfer functions for specified locations along the static configuration can then be efficiently computed (i.e. step 3 in transfer function calculation) prior to dynamic analysis.</li>
<li>Wave kinematics transfer functions are given directly as input to DYNMOD on a separate input file. The locations of transfer functions are specified by z-coordinates in vessel coordinate system. Transfer functions at intermediate nodes in the diffraction area are found by interpolation along the structure, see figure <a class="el" href="a00014.html#transfer_functions_given_at_specified_location">Transfer functions given at specified locations along static riser configuration</a>.</li>
<li>Wave kinematics time series are generated based on velocity and acceleration spectra computed from interpolated transfer functions in the diffraction area using the FFT technique. Undisturbed wave kinematics are applied outside the diffraction area. Consistent phase angles are applied in the FFT procedure to obtain consistent vessel motions and undisturbed/disturbed wave kinematics.</li>
</ol>
<p>By this strategy, only step 3) in the calculation of transfer functions must be carried out for each new static configuration which will give a computationally efficient procedure.</p>
<p><a class="anchor" id="transfer_functions_given_at_specified_location"></a></p>
<div class="image">
<img src="RIFLEX_TheoryManual_42_v0_777.svg" alt="RIFLEX_TheoryManual_42_v0_777.svg"/>
<div class="caption">
Transfer functions given at specified locations along static riser configuration</div></div>
<p>The described approach can be extended to a complete 3D description by generation of wave kinematics transfer functions in a grid covering the dynamic motion range of the structure as indicated in figure <a class="el" href="a00014.html#grid_of_transfer_functions">Grid of transfer functions</a>. It should, however, be noted that significantly more input is required and that a more complex data structure and interpolation procedure is necessary (not implemented in present version).</p>
<p><a class="anchor" id="grid_of_transfer_functions"></a></p>
<div class="image">
<img src="RIFLEX_TheoryManual_42_v0_778.svg" alt="RIFLEX_TheoryManual_42_v0_778.svg"/>
<div class="caption">
Grid of transfer functions (not implemented in the present version)</div></div>
<h2><a class="anchor" id="tm_load_wave_second"></a>
Second Order Stokes' Irregular Waves</h2>
<h3><a class="anchor" id="tm_load_wave_second_stokes"></a>
Stokes' theory</h3>
<p>In this section, Stokes' theory is presented. Following the governing set of differential equations, it is shown how a Taylor expansion of the velocity potential around the mean surface level leads to a series expansion in terms of wave steepness. Analytical solutions of the velocity potential and wave elevation are given up to and including second order. This is followed by discussions of issues of importance to the application of Stokes' theory for irregular waves with a target spectrum.</p>
<h4><a class="anchor" id="tm_load_wave_second_stokes_potential"></a>
The wave potential</h4>
<p>Stokes' theory of ocean waves assumes irrotational flow of an incompressible fluid, described by a scalar velocity potential <img class="formulaInl" alt="$ \Phi(\vec{x},t) $" src="form_675.png"/> which satisfies</p>
<p><a class="anchor" id="tm_eq_7_26"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \nabla ^2 \Phi = 0 , \tag{7.26} \]" src="form_676.png"/>
</p>
<p>with a velocity field <img class="formulaInl" alt="$ \vec{u} $" src="form_677.png"/> given by</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \vec{u} = \nabla \Phi . \tag {7.27} \]" src="form_678.png"/>
</p>
<h4><a class="anchor" id="tm_load_wave_second_stokes_boundary"></a>
Boundary conditions</h4>
<p>Assuming an impermeable sea-floor of constant mean depth <img class="formulaInl" alt="$ h $" src="form_679.png"/>, we have <img class="formulaInl" alt="$ u_z(z=-h)=0 $" src="form_680.png"/>, which we write in terms of the wave potential as</p>
<p><a class="anchor" id="tm_eq_7_28"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \partial_z \Phi = 0 \hspace{5mm} \mathrm{at} \hspace{2mm} z = -h . \tag {7.28} \]" src="form_681.png"/>
</p>
<p>The free surface of the wave is given by <img class="formulaInl" alt="$ z = \eta(x,y,t) $" src="form_682.png"/>. For sea states with long-crested waves the <img class="formulaInl" alt="$ y $" src="form_683.png"/> coordinate is irrelevant, while for short-crested waves it is sometimes useful to transform between a cartesian and a polar coordinate system, such that <img class="formulaInl" alt="$ z = \eta(r,\theta,t) $" src="form_684.png"/>. Now, <img class="formulaInl" alt="$ r $" src="form_685.png"/> plays the role which <img class="formulaInl" alt="$ x $" src="form_163.png"/> plays in long-crested waves, and <img class="formulaInl" alt="$ \eta $" src="form_686.png"/> results from the superposition of several <img class="formulaInl" alt="$ \eta_{\mathrm i}(r,t) \equiv \eta(r,\theta_{\mathrm i},t) $" src="form_687.png"/>.</p>
<p>Two boundary conditions are imposed at the wave surface. The kinematic boundary condition</p>
<p><a class="anchor" id="tm_eq_7_29"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \partial_t \eta + \partial_x \Phi \partial_x \eta + \partial_y \Phi \partial_y \eta - \partial_z \Phi = 0 \hspace{5mm} \mathrm{at} \hspace{2mm} z = \eta , \tag {7.29} \]" src="form_688.png"/>
</p>
<p>which states that a particle at the surface moves together with the surface, and the dynamic boundary condition</p>
<p><a class="anchor" id="tm_eq_7_30"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \partial_t \Phi + \frac{1}{2} \left| \vec{u} \right| ^2 + g \eta = 0 \hspace{5mm} \mathrm{at} \hspace{2mm} z = \eta . \tag {7.30} \]" src="form_689.png"/>
</p>
<p>which follows from assuming constant pressure along the surface. Here, <img class="formulaInl" alt="$ g $" src="form_273.png"/> is the acceleration of gravity.</p>
<p>Together with <a class="el" href="a00014.html#tm_eq_7_26">Eq. (7.26)</a>, which applies within the bulk of the ocean, <a class="el" href="a00014.html#tm_eq_7_28">Eq. (7.28)</a>, <a class="el" href="a00014.html#tm_eq_7_29">Eq. (7.29)</a> and <a class="el" href="a00014.html#tm_eq_7_30">Eq. (7.30)</a> completely specify the set of differential equations from which <img class="formulaInl" alt="$ \Phi $" src="form_690.png"/> is obtained.</p>
<h4><a class="anchor" id="tm_load_wave_second_stokes_series"></a>
Series expansion of the surface elevation</h4>
<p>In order to obtain analytic results for <img class="formulaInl" alt="$ \Phi $" src="form_690.png"/>, a perturbative approach is taken.</p>
<p>The velocity potential at the wave surface is extrapolated onto the mean surface level <img class="formulaInl" alt="$ z=0 $" src="form_691.png"/> using a Taylor expansion, with terms of higher order in the expansion expected to contribute at higher order in the ordering parameter <img class="formulaInl" alt="$ \epsilon $" src="form_692.png"/>.</p>
<p>The expansion is, to order <img class="formulaInl" alt="$ \mathcal{O}(\epsilon^2) $" src="form_693.png"/>,</p>
<p><a class="anchor" id="tm_eq_7_31"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \Phi \rvert_{z>0} = \Phi \rvert_{z=0} + z \left\{ \partial_z \Phi \right\}_{z=0} + \frac{1}{2} z^2 \left\{ \partial_z^2 \Phi \right\}_{z=0} + \dots . \tag {7.31} \]" src="form_694.png"/>
</p>
<p>Likewise, the surface elevation is written as a series expansion,</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \eta = \eta^{(1)} + \eta^{(2)} + \dots , \tag {7.32} \]" src="form_695.png"/>
</p>
<p>where it is usually assumed that <img class="formulaInl" alt="$ \eta^{(1)} $" src="form_696.png"/> is a <em>linear</em> wave. This is a construction which allows us to study the non-linear effects which appear at second order and beyond.</p>
<p>The surface elevation is a function of <img class="formulaInl" alt="$ x $" src="form_163.png"/>, <img class="formulaInl" alt="$ y $" src="form_683.png"/> and <img class="formulaInl" alt="$ t $" src="form_10.png"/>, or equivalently and for many purposes more conveniently, a function of <img class="formulaInl" alt="$ r $" src="form_685.png"/>, <img class="formulaInl" alt="$ \theta $" src="form_246.png"/> and <img class="formulaInl" alt="$ t $" src="form_10.png"/>. Staying with the former set of coordinates, the surface elevation in Fourier space is a function of wavenumber <img class="formulaInl" alt="$ \vec{k}=(k_x,k_y) $" src="form_697.png"/> and angular frequency <img class="formulaInl" alt="$ \omega $" src="form_698.png"/>. They are related through the dispersion relation for surface waves at finite water-depth,</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \omega^2 = g|\vec{k}| \tanh |\vec{k}|h , \tag {7.33} \]" src="form_699.png"/>
</p>
<p>which is complete to second order. At third order, the dispersion relation includes an additional term.</p>
<p>The first order contribution <img class="formulaInl" alt="$ \eta^{(1)} $" src="form_696.png"/> is provided to <code>WaveKin</code> by its Fourier components <img class="formulaInl" alt="$ a_{\mathrm i} $" src="form_700.png"/>,</p>
<p><a class="anchor" id="tm_eq_7_34"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \eta^{(1)} = \sum_{{\mathrm i} = -N}^{N} \sum_{{\mathrm k} = 1}^{M} \frac{1}{2} a_{\mathrm{ik}} e^{i \psi_{\mathrm{ik}}} , \tag {7.34} \]" src="form_701.png"/>
</p>
<p>where the sums run over the number of positive frequencies <img class="formulaInl" alt="$ N $" src="form_434.png"/> and the number of directions <img class="formulaInl" alt="$ M $" src="form_702.png"/> in the discretized directional spectrum, <img class="formulaInl" alt="$ \psi_{\mathrm{ik}} = \omega_{\mathrm i} t - \vec{k_{\mathrm{ik}}} \cdot \vec{x} - \varepsilon_{\mathrm{ik}} $" src="form_703.png"/> is the phase function, with angular frequency <img class="formulaInl" alt="$ \omega_{\mathrm i} = {\mathrm i} \Delta \omega $" src="form_704.png"/>. The phase is <img class="formulaInl" alt="$ \varepsilon_{\mathrm{ik}} $" src="form_705.png"/> and the wavenumber <img class="formulaInl" alt="$ \vec{k_{\mathrm{ik}}} $" src="form_706.png"/> is obtained from <img class="formulaInl" alt="$ \omega_{\mathrm i} $" src="form_707.png"/> through use of the dispersion relation and wave direction <img class="formulaInl" alt="$ \theta_{\mathrm{ik}} $" src="form_708.png"/>. The two subscripts denote a discretization of the Fourier components of a directional spectrum. In the following, the second subscript <img class="formulaInl" alt="$ \mathrm{k} $" src="form_709.png"/> is dropped in favor of a compact notation. It should be implicitly understood that for directional spectra, all sums over Fourier components also imply a sum over directions.</p>
<p>For a simulated wave from a given spectrum, the random phase will typically be drawn from a uniform probability distribution. For a directional spectrum this involves <img class="formulaInl" alt="$ MN $" src="form_710.png"/> independent and identically distributed random numbers. See <a class="el" href="a00014.html#tm_load_wave_second_spectrum">Spectrum</a> for a further discussion on this.</p>
<p>The sum of <a class="el" href="a00014.html#tm_eq_7_34">Eq. (7.34)</a> can be re-written to run over positive frequencies only. Because <img class="formulaInl" alt="$ \eta(x,y,t) $" src="form_711.png"/> is real-valued, <img class="formulaInl" alt="$ a_{-\mathrm i} = \bar{a}_{\mathrm i} $" src="form_712.png"/>. This gives</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \eta^{(1)} = \sum_{{\mathrm i} = 1}^{N} a_{\mathrm{i}} \cos \psi_{\mathrm{i}} , \tag {7.35} \]" src="form_713.png"/>
</p>
<p>where the sum over directions is now implicit.</p>
<p>Velocities and accelerations are also real-valued; <code>WaveKin</code> therefore works with one-sided spectra only. Time-series may have a non-zero mean value, therefore the one-sided spectra include the zero-frequency components. The zero-frequency component for surface elevation <img class="formulaInl" alt="$ a_0 $" src="form_714.png"/> is also included in the one-sided spectrum, though its value should be zero or near zero. See <a class="el" href="a00014.html#tm_load_wave_second_stokes_sum">The sum-frequency and difference-frequency transfer functions</a> for how symmetries allow summations of first and second order terms to run over positive frequencies only and <a class="el" href="a00014.html#tm_load_wave_second_stokes_diagonal">The diagonal of the difference-frequency kernels</a> for a further discussion on the zero-frequency components.</p>
<p>The expansion is in terms of an ordering parameter <img class="formulaInl" alt="$ \epsilon $" src="form_692.png"/>. It can be seen that higher orders in the expansions of the boundary conditions at the wave surface results in terms with higher powers of spatial derivatives, surface elevation, and the velocity potential. In Fourier space, spatial derivatives result in additional factors of <img class="formulaInl" alt="$ k $" src="form_392.png"/>, and we have constructed the first order surface elevation to be a sum of harmonics with amplitude <img class="formulaInl" alt="$ |a| $" src="form_715.png"/>.</p>
<p>The combination of factors of <img class="formulaInl" alt="$ k $" src="form_392.png"/> and <img class="formulaInl" alt="$ a $" src="form_716.png"/> provides a physical meaning to the ordering parameter. The wave steepness is proportional to <img class="formulaInl" alt="$ ak $" src="form_717.png"/>, for a harmonic wave elevation <img class="formulaInl" alt="$ \eta_k = a \sin kx $" src="form_718.png"/>. In order for the series expansion to converge, the linear wave should have a spectral content which satisfies <img class="formulaInl" alt="$ |a_{\mathrm i}| k_{\mathrm i} < \epsilon $" src="form_719.png"/> for all <img class="formulaInl" alt="$ \mathrm{i} $" src="form_720.png"/>.</p>
<p>In addition to the Taylor series expansion of the wave surface boundary conditions and the expansion of <img class="formulaInl" alt="$ \eta $" src="form_686.png"/>, the velocity potential is written as a series expansion</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \Phi = \Phi^{(1)} + \Phi^{(2)} + \dots . \tag {7.36} \]" src="form_721.png"/>
</p>
<p>A solution is obtained for each term of the expansion by ordering the terms in the wave surface boundary conditions, setting <img class="formulaInl" alt="$ z=\eta $" src="form_722.png"/> in <a class="el" href="a00014.html#tm_eq_7_31">Eq. (7.31)</a>. Terms at equal order in <img class="formulaInl" alt="$ \epsilon $" src="form_692.png"/> are solved for independently.</p>
<h4><a class="anchor" id="tm_load_wave_second_stokes_first"></a>
First order solution</h4>
<p>Solving for the first order velocity potential gives </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \Phi^{(1)}(\vec{x},t) = \sum_{{\mathrm i} = -N}^{N} \frac{i g a_{\mathrm i}}{2 \omega_{\mathrm i}} \frac{\cosh |\vec{k_{\mathrm i}}|(z+h)}{\cosh |\vec{k_{\mathrm i}}|h} e^{i \psi_{\mathrm i}} , \tag {7.37} \]" src="form_723.png"/>
</p>
 <p class="formulaDsp">
<img class="formulaDsp" alt="\[ \Phi^{(1)}(\vec{x},t) = - \sum_{{\mathrm i} = 1}^{N} \frac{g a_{\mathrm i}}{\omega_{\mathrm i}} \frac{\cosh |\vec{k_{\mathrm i}}|(z+h)}{\cosh |\vec{k_{\mathrm i}}|h} \sin \psi_{\mathrm i} . \tag {7.38} \]" src="form_724.png"/>
</p>
<h4><a class="anchor" id="tm_load_wave_second_stokes_second"></a>
Second order solution</h4>
<p>Following <a class="el" href="a00014.html#tm_load_references">Marthinsen and Winterstein (1992)</a> for long-crested waves and <a class="el" href="a00014.html#tm_load_references">Sharma and Dean (1981)</a> for short-crested waves, the velocity potential and second order contribution to surface elevation is given to second order by a set of rather complex expressions, organized in terms of sum-frequency and difference-frequency contributions. The expressions are analytic, and therefore support differentiation, allowing the analytic calculation of velocities and accelerations.</p>
<p>Solving for the second order surface elevation gives</p>
<p><a class="anchor" id="tm_eq_7_39"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \eta^{(2)} = \sum_{{\mathrm{i,j}} = -N}^{N} \frac{a_{\mathrm i} a_{\mathrm j}}{2} E_{\mathrm{i,j}} e^{i (\psi_{\mathrm i} + \psi_{\mathrm j})} . \tag {7.39} \]" src="form_725.png"/>
</p>
<p>Solving for the second order velocity potential gives</p>
<p><a class="anchor" id="tm_eq_7_40"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \Phi^{(2)}(\vec{x},t) = \sum_{{\mathrm{i,j}} = -N}^{N} \frac{i a_{\mathrm i} a_{\mathrm j}}{2} P_{\mathrm{i,j}} \frac{\cosh |\vec{k_{\mathrm i}}+\vec{k_{\mathrm j}}|(z+h)} {\cosh |\vec{k_{\mathrm i}}+\vec{k_{\mathrm j}}|h} e^{i (\psi_{\mathrm i} + \psi_{\mathrm j})} + \Theta^{(2)}(t) , \tag {7.40} \]" src="form_726.png"/>
</p>
<p>where the latter term is independent of spatial coordinates and therefore irrelevant for kinematics.</p>
<p>The second order kernels <img class="formulaInl" alt="$ E_{\mathrm{i,j}} $" src="form_727.png"/> and <img class="formulaInl" alt="$ P_{\mathrm{i,j}} $" src="form_728.png"/> are given by</p>
<p><a class="anchor" id="tm_eq_7_41"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ E_{\mathrm{i,j}} = \frac{\omega_{\mathrm i}+\omega_{\mathrm j}}{g} P_{\mathrm{i,j}} - \frac{g}{4} \frac{\vec{k_{\mathrm i}} \cdot \vec{k_{\mathrm j}}} {\omega_{\mathrm i} \omega_{\mathrm j}} - \frac{1}{4g} (\omega_{\mathrm i}^2 + \omega_{\mathrm j}^2 + \omega_{\mathrm i} \omega_{\mathrm j}) , \tag {7.41} \]" src="form_729.png"/>
</p>
<p>and</p>
<p><a class="anchor" id="tm_eq_7_42"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ P_{\mathrm{i,j}} = \frac { \frac{g^2}{2} \frac{\vec{k_{\mathrm i}} \cdot \vec{k_{\mathrm j}}} {\omega_{\mathrm i} \omega_{\mathrm j}} - \frac{1}{4} (\omega_{\mathrm i}^2 + \omega_{\mathrm j}^2 + \omega_{\mathrm i} \omega_{\mathrm j}) + \frac{g^2}{4} \frac{\omega_{\mathrm j} k_{\mathrm i}^2 + \omega_{\mathrm i} k_{\mathrm j}^2} {\omega_{\mathrm i} \omega_{\mathrm j} (\omega_{\mathrm i}+\omega_{\mathrm j})} } { \omega_{\mathrm i}+\omega_{\mathrm j} - g \frac{|\vec{k_{\mathrm i}}+\vec{k_{\mathrm j}}|} {\omega_{\mathrm i}+\omega_{\mathrm j}} \tanh |\vec{k_{\mathrm i}}+\vec{k_{\mathrm j}}|h } . \tag {7.42} \]" src="form_730.png"/>
</p>
<p>The solutions for both <img class="formulaInl" alt="$ \eta^{(2)} $" src="form_731.png"/> and <img class="formulaInl" alt="$ \Phi^{(2)} $" src="form_732.png"/> depend on products, sums and differences of Fourier components. Because the two-sided spectrum can be expressed as a one-sided spectrum, the second order contributions are naturally separated into sum-frequency and difference-frequency contributions.</p>
<p>Because the sum and difference-frequency components only depend on the sum and difference of frequencies (wavenumbers) and the Fourier components satisfy <img class="formulaInl" alt="$ a_{-\mathrm i} = \bar{a}_{\mathrm i} $" src="form_712.png"/>, the calculations of second order contributions to surface elevation, velocities and accelerations can be considerably simplified.</p>
<h4><a class="anchor" id="tm_load_wave_second_stokes_sum"></a>
The sum-frequency and difference-frequency transfer functions</h4>
<p>The kernels from <a class="el" href="a00014.html#tm_eq_7_41">Eq. (7.41)</a> and <a class="el" href="a00014.html#tm_eq_7_42">Eq. (7.42)</a> can be written in terms of sums and differences of frequencies and corresponding wavenumbers,</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ E_{\mathrm{ij}}^{\pm} = \frac{\omega_{\mathrm i} \pm \omega_{\mathrm j}}{g} P_{\mathrm{ij}}^{\pm} - \frac{g}{4} \frac{\vec{k_{\mathrm i}} \cdot \vec{k_{\mathrm j}}} {\omega_{\mathrm i} \omega_{\mathrm j}} - \frac{1}{4g} (\omega_{\mathrm i}^2 + \omega_{\mathrm j}^2 \pm \omega_{\mathrm i} \omega_{\mathrm j}) , \tag {7.43} \]" src="form_733.png"/>
</p>
<p>and</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ P_{\mathrm{ij}}^{\pm} = \frac { \frac{g^2}{2} \frac{\vec{k_{\mathrm i}} \cdot \vec{k_{\mathrm j}}} {\omega_{\mathrm i} \omega_{\mathrm j}} - \frac{1}{4} (\omega_{\mathrm i}^2 + \omega_{\mathrm j}^2 \pm \omega_{\mathrm i} \omega_{\mathrm j}) + \frac{g^2}{4} \frac{\omega_{\mathrm j} k_{\mathrm i}^2 \pm \omega_{\mathrm i} k_{\mathrm j}^2} {\omega_{\mathrm i} \omega_{\mathrm j} (\omega_{\mathrm i} \pm \omega_{\mathrm j})} } { \omega_{\mathrm i} \pm \omega_{\mathrm j} - g \frac{|\vec{k_{\mathrm i}} \pm \vec{k_{\mathrm j}}|} {\omega_{\mathrm i} \pm \omega_{\mathrm j}} \tanh |\vec{k_{\mathrm i}} \pm \vec{k_{\mathrm j}}|h } , \tag {7.44} \]" src="form_734.png"/>
</p>
<p>where <img class="formulaInl" alt="$ \mathrm{i,j} > 0 $" src="form_735.png"/>.</p>
<p>The symmetry properties of the second order kernels are</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ E_{\mathrm{ij}}^+ = E_{\mathrm{i,j}} = E_{\mathrm{j,i}} , \tag {7.45} \]" src="form_736.png"/>
</p>
 <p class="formulaDsp">
<img class="formulaDsp" alt="\[ E_{\mathrm{ij}}^- = E_{\mathrm{i,-j}} = - E_{\mathrm{j,-i}} , \tag {7.46} \]" src="form_737.png"/>
</p>
 <p class="formulaDsp">
<img class="formulaDsp" alt="\[ P_{\mathrm{ij}}^+ = P_{\mathrm{i,j}} = P_{\mathrm{j,i}} , \tag {7.47} \]" src="form_738.png"/>
</p>
 <p class="formulaDsp">
<img class="formulaDsp" alt="\[ P_{\mathrm{ij}}^- = P_{\mathrm{i,-j}} = - P_{\mathrm{j,-i}} , \tag {7.48} \]" src="form_739.png"/>
</p>
<p>where <img class="formulaInl" alt="$ \mathrm{i,j} > 0 $" src="form_735.png"/>.</p>
<p>For the second order wave elevation of <a class="el" href="a00014.html#tm_eq_7_39">Eq. (7.39)</a> we now have <img class="formulaInl" alt="$ \eta^{(2)}=\eta^{+} + \eta^{-} $" src="form_740.png"/> with</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \eta^{\pm} = \sum_{{\mathrm{i,j}} = 1}^{N} a_{\mathrm i} a_{\mathrm j} E_{\mathrm{i,j}}^{\pm} \cos (\psi_{\mathrm i} \pm \psi_{\mathrm j}) . \tag {7.49} \]" src="form_741.png"/>
</p>
<p>The second order velocity potential of <a class="el" href="a00014.html#tm_eq_7_40">Eq. (7.40)</a> can now be written as <img class="formulaInl" alt="$ \Phi^{(2)}=\Phi^{+} + \Phi^{-} $" src="form_742.png"/> with</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \Phi^{\pm}(\vec{x},t) = - \sum_{{\mathrm{i,j}} = 1}^{N} a_{\mathrm i} a_{\mathrm j} P_{\mathrm{i,j}}^{\pm} \frac{\cosh |\vec{k_{\mathrm i}} \pm \vec{k_{\mathrm j}}|(z+h)} {\cosh |\vec{k_{\mathrm i}} \pm \vec{k_{\mathrm j}}|h} \sin (\psi_{\mathrm i} \pm \psi_{\mathrm j}) , \tag {7.50} \]" src="form_743.png"/>
</p>
<p>where space-independent terms have been omitted.</p>
<h4><a class="anchor" id="tm_load_wave_second_stokes_diagonal"></a>
The diagonal of the difference-frequency kernels</h4>
<p>The difference-frequency kernels <img class="formulaInl" alt="$ E_{\mathrm{ij}}^- $" src="form_744.png"/> and <img class="formulaInl" alt="$ P_{\mathrm{ij}}^- $" src="form_745.png"/> are singular for <img class="formulaInl" alt="$ \mathrm{i} = \mathrm{j} $" src="form_746.png"/>. If one chooses to take the analytic limit, one gets a non-zero contribution to the zero-frequency contribution of the difference-frequency transfer functions. For <img class="formulaInl" alt="$ E_{\mathrm{ii}}^- $" src="form_747.png"/> this gives</p>
<p><a class="anchor" id="tm_eq_7_51"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ E_{\mathrm{ii}}^- \equiv \lim_{\omega_{\mathrm{j}} \to -\omega_{\mathrm{i}}} E_{\mathrm{i,j}} = \frac{1}{2} \frac{ \frac{gk_{\mathrm{i}}^2}{2\omega_{\mathrm{i}}^2} - \frac{\omega_{\mathrm{i}}^2}{2g} + \frac{gk_{\mathrm{i}}}{\omega c_{g\mathrm{i}}} }{1 - \frac{gh}{c_{g\mathrm{i}}^2}} - \frac{k_{\mathrm{i}}}{2 \sinh 2k_{\mathrm{i}}h} , \tag {7.51} \]" src="form_748.png"/>
</p>
<p>where</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ c_g = \frac{\mathrm{d}\omega}{\mathrm{d}k} , \tag {7.52} \]" src="form_749.png"/>
</p>
<p>is the group velocity.</p>
<p><a class="el" href="a00014.html#tm_eq_7_51">Eq. (7.51)</a> follows Marthinsen and Winterstein (1992).</p>
<p>With non-zero contributions from the difference-frequency kernels when <img class="formulaInl" alt="$ \mathrm{i} = \mathrm{j} $" src="form_746.png"/>, the resulting transfer functions give finite contributions to the zero-frequency component of the second order elevation and velocity potential. This corresponds to finite mean-values for <img class="formulaInl" alt="$ \eta $" src="form_686.png"/>, <img class="formulaInl" alt="$ u_x $" src="form_750.png"/> and <img class="formulaInl" alt="$ u_y $" src="form_751.png"/>. This can optionally be avoided by setting all zero-frequency components to zero, (equivalently, setting the diagonal entries of the difference-frequency kernels to zero) however this procedure lacks physical or theoretical justification.</p>
<p>In particular, the so-called mean sea state set-down leads to an inconsistency with the boundary condition applied at the sea floor, i.e. at <img class="formulaInl" alt="$ z=-h $" src="form_752.png"/>, as the sea floor depth <img class="formulaInl" alt="$ h $" src="form_679.png"/> should properly be measured from the mean sea level, <img class="formulaInl" alt="$ <\eta> $" src="form_753.png"/>, not from the first order contribution to the same, <img class="formulaInl" alt="$ <\eta^{(1)}> $" src="form_754.png"/>. The inconsistency might be small for most practical uses, but deserves mentioning. This is discussed occasionally in the literature, e.g., <a class="el" href="a00014.html#tm_load_references">Mei (1989)</a> and <a class="el" href="a00014.html#tm_load_references">Dalzell (1999)</a>.</p>
<p>The default behaviour of <code>WaveKin</code> in <code>RIFLEX</code> is to set the diagonal terms equal to zero.</p>
<h4><a class="anchor" id="tm_load_wave_second_stokes_high"></a>
High frequency contributions from the sum-frequency transfer functions</h4>
<p>Depending on how the target spectrum is filtered to obtain the first order wave elevation defined by <img class="formulaInl" alt="$ \{a_{\mathrm{i}}\} $" src="form_755.png"/>, the sum-frequency transfer functions may result in significant high frequency contributions to the second order wave elevation and kinematics. The acceleration terms are particularly sensitive to the high frequency part of the spectrum. Care must be taken when defining the first order wave elevation to avoid un-physical high frequency terms arising at second order. Otherwise, un-physically large accelerations could result.</p>
<p>The issue of large accelerations, i.e., rapidly varying velocities, is most acute near the mean sea surface and above ( <img class="formulaInl" alt="$ z \ge 0 $" src="form_756.png"/>).</p>
<h3><a class="anchor" id="tm_load_wave_second_spectrum"></a>
Spectrum</h3>
<p>The input to <code>WaveKin</code> is the first order wave elevation. From this, the second order contribution to wave elevation is calculated and added to the first order part. Also, kinematics are calculated complete to second order.</p>
<p>The input wave elevation may be provided as a directional spectrum, in which case the 2D array (frequency and direction) of Fourier components must be accompanied by an array of directions. The generation of the first order wave elevation is done outside of <code>WaveKin</code>.</p>
<p>The Fourier components may be taken deterministically from the power spectrum. For a time series of finite length there will be some variability of the power spectrum with respect to the target spectrum, and thus individual Fourier components should be <img class="formulaInl" alt="$ \chi^2 $" src="form_757.png"/> distributed. However, it should be noted that when <img class="formulaInl" alt="$ \eta $" src="form_686.png"/> is given as a sum of directional components this in itself results in a natural source of variability for the Fourier components at any given frequency.</p>
<h4><a class="anchor" id="tm_load_wave_second_spectrum_linear"></a>
The linear wave and the filtering procedure</h4>
<p>By target spectrum is meant the empirically established power spectrum <img class="formulaInl" alt="$ S(\omega) $" src="form_758.png"/> for long-crested waves or the directional spectrum <img class="formulaInl" alt="$ S(\omega,\theta) $" src="form_759.png"/> for directional sea states.</p>
<p>In order to generate an irregular sea state, phases are typically drawn from a uniform distribution, with the amplitude of Fourier components fixed by a filtered power spectrum. The resulting wave elevation is then referred to as a linear wave and identified as the first order contribution in the perturbative expansion of <img class="formulaInl" alt="$ \eta $" src="form_686.png"/>.</p>
<p>Various heuristics are presented in the literature for how the filtering should be done, see e.g. <a class="el" href="a00014.html#tm_load_references">Forristall (2000)</a> and <a class="el" href="a00014.html#tm_load_references">Stansberg, Gudmunstad and Haver (2008)</a>, but no solid theoretical foundation underlies these methods. For many purposes, results may be insensitive to the details of the filtering, however results from <code>WaveKin</code> indicate that filtering has a significant effect on wave kinematics near the sea surface at water depths of <img class="formulaInl" alt="$ 20-40 {\mathrm{~m}} $" src="form_760.png"/> and significant wave heights in the range of <img class="formulaInl" alt="$ 5-6 {\mathrm{~m}} $" src="form_761.png"/>.</p>
<p>Equating the linear wave with the first order wave elevation is a matter of convenience, but could under-estimate the non-linear nature of Stokes' waves. The assumption of un-correlated phases is justified by very weak wave-wave interactions, which in turn applies to low wave steepness. While the first order wave elevation will always have a more narrow banded spectrum than the full wave, it will nevertheless contain steep waves.</p>
<h4><a class="anchor" id="tm_load_wave_second_spectrum_separability"></a>
Separability of the directional spectrum</h4>
<p>A directional spreading function <img class="formulaInl" alt="$ D(\theta,\omega) $" src="form_762.png"/> is defined by</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ S(\omega, \theta) = S(\omega) \cdot D(\theta,\omega) , \tag {7.53} \]" src="form_763.png"/>
</p>
<p>together with the normalization condition</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \int^{2\pi}_0 D(\theta,\omega) {\mathrm d}\theta = 1 . \tag {7.54} \]" src="form_764.png"/>
</p>
<p>The power spectrum <img class="formulaInl" alt="$ S(\omega) $" src="form_758.png"/> for directional sea states is obtained by integrating over the directional coordinate</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ S(\omega) = \int^{2\pi}_0 S(\omega,\theta) {\mathrm d}\theta . \tag {7.55} \]" src="form_765.png"/>
</p>
<p>The simplest approach to describing a directional sea state is to assume a separable directional spectrum,</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ S(\omega, \theta) = S(\omega) \cdot D(\theta) , \tag {7.56} \]" src="form_766.png"/>
</p>
<p>where the power spectrum <img class="formulaInl" alt="$ S(\omega) $" src="form_758.png"/> (e.g., a JONSWAP spectrum) and the directional spreading function <img class="formulaInl" alt="$ D(\theta) $" src="form_767.png"/> are given separately and together define the sea state.</p>
<p>A common choice for <img class="formulaInl" alt="$ D $" src="form_625.png"/> is a frequency-independent cosine-2s distribution</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ D_s(\theta) = C_s \cos^{2s} (\theta - \theta_0) , \tag {7.57} \]" src="form_768.png"/>
</p>
<p>for <img class="formulaInl" alt="$ |\theta - \theta_0| \leq \frac{\pi}{2} $" src="form_769.png"/> and zero otherwise. <img class="formulaInl" alt="$ C_s $" src="form_770.png"/> is a normalization factor. The width of the distribution is parametrized by <img class="formulaInl" alt="$ s $" src="form_771.png"/> and the mean direction is given by <img class="formulaInl" alt="$ \theta_0 $" src="form_772.png"/>.</p>
<h3><a class="anchor" id="tm_load_wave_second_kinematics"></a>
Kinematics</h3>
<p>Velocities and accelerations are obtained from applying partial derivatives to the wave potential, which is known analytically up to and including second order.</p>
<p>The velocity and acceleration to second order are given by</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \vec{u}^{(1+2)} = \vec{u}^{(1)} + \vec{u}^{(2)} = \nabla (\Phi^{(1)} + \Phi^{(2)}) , \tag {7.58} \]" src="form_773.png"/>
</p>
 <p class="formulaDsp">
<img class="formulaDsp" alt="\[ \vec{a}^{(1+2)} = \frac{\mathrm d}{\mathrm dt} (\vec{u}^{(1)} + \vec{u}^{(2)}) . \tag {7.59} \]" src="form_774.png"/>
</p>
<p><code>WaveKin</code> by default performs differentiation by applying partial derivatives in Fourier space. This is efficient and accurate, and has been validated against differentiating along temporal and spatial coordinates. An option exists to perform differentiation numerically along the temporal coordinate, which is turned off by default.</p>
<h4><a class="anchor" id="tm_load_wave_second_kinematics_convection"></a>
Convection terms</h4>
<p>Frequently, convective terms are neglected in the calculation of wave kinematics. In the current implementation of <code>WaveKin</code> the inclusion of these terms is optional. If they are omitted, only the contribution from partial time-derivatives are included</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \vec{a}^{(1+2)} = \frac{\mathrm d}{\mathrm dt} (\vec{u}^{(1)} + \vec{u}^{(2)}) \approx \frac{\partial}{\partial t} (\vec{u}^{(1)} + \vec{u}^{(2)}) . \tag {7.60} \]" src="form_775.png"/>
</p>
<p>Convection terms can optionally be included in the accelerations returned by <code>WaveKin</code>. These are calculated by taking spatial derivatives of the velocity field (in the frequency domain) and multiplying with the velocity field (in the time domain).</p>
<p>Convection terms contribute to the accelerations of momentum-carrying fluid particles, and thus contribute to the inertia term of Morison's load model. They emerge when we take the total derivative of velocity with respect to time,</p>
<p><a class="anchor" id="tm_eq_7_61"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \vec{a} = \dot{\vec{u}} = \frac{\mathrm d}{\mathrm dt} \vec{u} = \frac{\partial}{\partial t} \vec{u} + \vec{v} \cdot \nabla \vec{u} , \tag {7.61} \]" src="form_776.png"/>
</p>
<p>where <img class="formulaInl" alt="$ \vec{v} $" src="form_777.png"/> is the velocity vector and <img class="formulaInl" alt="$ \vec{u} = \vec{u}(\vec{x},t) $" src="form_778.png"/> is the velocity field.</p>
<p>The inclusion of convection terms is turned off by default in <code>WaveKin</code>. This follows an assumption that they are small and can be neglected for practical purposes. This assumption should be questioned for severe sea states. The last term in <a class="el" href="a00014.html#tm_eq_7_61">Eq. (7.61)</a> constitutes the convection terms. It should be seen that its magnitude depends on the gradient of the velocity field, meaning that rapidly varying velocities, as found near the sea surface in severe sea states, may lead to significant contributions from convection terms.</p>
<h4><a class="anchor" id="tm_load_wave_second_kinematics_near"></a>
Kinematics near to and above the mean surface level</h4>
<p>The velocity potential <img class="formulaInl" alt="$ \Phi(x,y,z,t) $" src="form_779.png"/> is extrapolated above <img class="formulaInl" alt="$ z=0 $" src="form_691.png"/> according to the Taylor expansion of <a class="el" href="a00014.html#tm_eq_7_31">Eq. (7.31)</a>. Kinematics are obtained from the extrapolated potential by spatial and temporal differentiation in the same manner as for the velocity potential at and below <img class="formulaInl" alt="$ z=0 $" src="form_691.png"/>. However, it should be noted that the second order contribution at <img class="formulaInl" alt="$ z>0 $" src="form_780.png"/> only involves the first order velocity potential <img class="formulaInl" alt="$ \Phi^{(1)} $" src="form_781.png"/>. There are no quadratic transfer functions involved above mean surface level until third order contributions.</p>
<p>High frequencies resulting from providing an improperly filtered spectrum as the input first order wave may cause unphysical accelerations near to and above the mean surface level. To compensate for this, an option is provided to filter out large frequencies prior to extrapolating the kinematics above the mean surface level.</p>
<p>Together with, or instead of this option, there is an option to set a threshold surface level <img class="formulaInl" alt="$ z_{\mathrm{lim}}<0 $" src="form_782.png"/>, above which kinematics are extrapolated from the potential at <img class="formulaInl" alt="$ z=z_{\mathrm{lim}} $" src="form_783.png"/>. As the large frequencies are rapidly dampened even at very small values of <img class="formulaInl" alt="$ z<0 $" src="form_784.png"/>, setting <img class="formulaInl" alt="$ z_{\mathrm{lim}} $" src="form_785.png"/> to only a few centimeters below <img class="formulaInl" alt="$ z=0 $" src="form_691.png"/> may be sufficient to avoid the excessive spikes in acceleration.</p>
<p>The default behaviour of <code>WaveKin</code> in <code>RIFLEX</code> is to use the entire provided spectrum when extrapolating kinematics above the mean surface level, and to extrapolate from <img class="formulaInl" alt="$ z=0 $" src="form_691.png"/>.</p>
<h4><a class="anchor" id="tm_load_wave_second_kinematics_above"></a>
Kinematics above the sea surface</h4>
<p>In order to extrapolate kinematics between two nodes, one of which is dry, it can be impractical to set kinematics to zero above the sea surface <img class="formulaInl" alt="$ z > \eta $" src="form_786.png"/> (sometimes referred to as a "dry node"). Instead, the option exists to set kinematics at <img class="formulaInl" alt="$ z > \eta $" src="form_786.png"/> equal to the the kinematics at the surface: <img class="formulaInl" alt="$ z=\eta $" src="form_722.png"/>. This is the default behaviour of <code>WaveKin</code> in <code>RIFLEX</code>.</p>
<h4><a class="anchor" id="tm_load_wave_second_kinematics_below"></a>
Kinematics on the surface and below the mean surface level</h4>
<p>For kinematics on the surface and at or above the mean surface level, <img class="formulaInl" alt="$ z = \eta \ge 0 $" src="form_787.png"/>, the kinematics are obtained at second order by extrapolation of the first order kinematics at <img class="formulaInl" alt="$ z=0 $" src="form_691.png"/>. For kinematics on the surface and below the mean surface level, <img class="formulaInl" alt="$ z = \eta < 0 $" src="form_788.png"/>, there is a choice between using the second order potential, which exists at all <img class="formulaInl" alt="$ z \le 0 $" src="form_789.png"/> or using the first order potential at <img class="formulaInl" alt="$ z = \eta < 0 $" src="form_788.png"/> together with the first order potential extrapolated from <img class="formulaInl" alt="$ z = 0 $" src="form_790.png"/>. The latter method provides a consistent calculation of kinematics on the surface regardless of whether it is above or below <img class="formulaInl" alt="$ z=0 $" src="form_691.png"/>. However, it causes a discontinuity in the kinematics on and directly beneath the surface.</p>
<p>The magnitude of the discrepancy between the two methods should be of third order, if the series expansion behaves consistently. Thus, a comparison of the two methods is a test of the series expansion.</p>
<h1><a class="anchor" id="tm_load_current"></a>
Current Description</h1>
<p>The current velocity is normally assumed to be constant with time. A quasi-static analysis can, however, include the effect from a quasi-static current variation, and it is also possible to describe time-dependent current variation in dynamic analysis.</p>
<p>The current velocity at a given position is described by speed and direction. This can be done by input of discrete values and interpolation to actual node positions, or by definition of standard profiles.</p>
<p>The velocity components in global directions are denoted</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{U}^\textrm{c} = [U^{c}_x, U^{c}_y, U^{c}_z] \tag{7.62} \]" src="form_791.png"/>
</p>
<p>The vertical component <img class="formulaInl" alt="$ U^{c}_z $" src="form_792.png"/> is always assumed to be zero.</p>
<h1><a class="anchor" id="tm_load_hydrodynamic_models"></a>
Hydrodynamic Load Models for Submerged Elements</h1>
<p>This section will give an introduction to hydrodynamic load models for submerged or surface-piercing normal slender structures (i.e. pipes, cables, risers, etc.).</p>
<h2><a class="anchor" id="tm_genmori"></a>
Generalized Morison's Equation</h2>
<p>The velocity potential <img class="formulaInl" alt="$ \phi^I $" src="form_793.png"/> of the undisturbed wave motion will be written as a superposition of linear waves of different circular frequencies <img class="formulaInl" alt="$ \omega_\textrm{i} $" src="form_794.png"/> and wave propagation angles <img class="formulaInl" alt="$ \omega_\textrm{j} $" src="form_795.png"/> ,</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \phi^I = \sum_{\textrm{j} =1}^{N_\beta}\, \sum_{\textrm{k} =1}^{N_\infty}\, \frac{gA_\textrm{jk} }{\omega_\textrm{k} } \frac{\textrm{cosh} \left[k_\textrm{k} (Z+D) \right] }{\textrm{cosh} \left( k_\textrm{k} D \right) }\cos (k_\textrm{k} X \cos \beta_\textrm{j} + k_\textrm{k} Y \sin \beta_\textrm{j} - \omega_\textrm{k} t -\phi_\textrm{jk}) \tag{7.63} \]" src="form_796.png"/>
</p>
<p>Here <img class="formulaInl" alt="$ g $" src="form_273.png"/> is the acceleration due to gravity, <img class="formulaInl" alt="$ t $" src="form_10.png"/> is the time variable and <img class="formulaInl" alt="$ (X,Y,Z) $" src="form_797.png"/> is a Cartesian coordinate system with <img class="formulaInl" alt="$ Z =0 $" src="form_798.png"/> in the mean free surface level.</p>
<p>Positive Z-axis is upwards. The mean water depth is <img class="formulaInl" alt="$ D $" src="form_625.png"/> . The wave numbers <img class="formulaInl" alt="$ k_\textrm{k} $" src="form_799.png"/> are related to the wave frequency <img class="formulaInl" alt="$ \omega_\textrm{k} $" src="form_800.png"/> by the dispersion relationship <img class="formulaInl" alt="$ \omega_\textrm{k}^2/g = k_\textrm{k} \textrm{tanh}(k_\textrm{k}h) $" src="form_801.png"/>. <img class="formulaInl" alt="$ \phi_\textrm{jk} $" src="form_649.png"/> are random phase angles and <img class="formulaInl" alt="$ A_\textrm{jk} $" src="form_802.png"/> are determined by a wave spectrum for short crested sea.</p>
<p>When calculating the free surface elevation <img class="formulaInl" alt="$ \eta $" src="form_686.png"/> we will use that </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{\eta }= -\frac{1}{g} \frac{\partial \boldsymbol{\Phi}^I}{\partial t} \bigg|_{Z=0} \tag{7.64} \]" src="form_803.png"/>
</p>
<p> This value will be used to position the calculated velocity and acceleration profiles, which enables us in an easy way to calculate the loads on structural surface-piercing parts.</p>
<p><a class="anchor" id="Flexible_riser_system"></a></p>
<div class="image">
<img src="RIFLEX_TheoryManual_42_v0_796.svg" alt="RIFLEX_TheoryManual_42_v0_796.svg"/>
<div class="caption">
Flexible riser system</div></div>
<p><a class="anchor" id="Riser_cross_sections"></a></p>
<div class="image">
<img src="RIFLEX_TheoryManual_42_v0_797.svg" alt="RIFLEX_TheoryManual_42_v0_797.svg"/>
<div class="caption">
Riser cross sections</div></div>
<p>A schematic picture of a possible flexible riser system is shown in figure <a class="el" href="a00014.html#Flexible_riser_system">Flexible riser system</a>. Cross-sections of typical flexible pipe configurations are shown in figure <a class="el" href="a00014.html#Riser_cross_sections">Riser cross sections</a>.</p>
<p>The hydrodynamic forces are calculated according to a generalized Morison's equation formulation, with the optional addition of a linear drag force term. A local coordinate system <img class="formulaInl" alt="$ (x,y,z) $" src="form_804.png"/> is defined. See figure <a class="el" href="a00014.html#Flexible_riser_system">Flexible riser system</a> where <img class="formulaInl" alt="$ e(y,z) $" src="form_805.png"/> is in the cross-sectional plane. Note that small letters indicate the local coordinate system while large letters <img class="formulaInl" alt="$ (X,Y,Z) $" src="form_797.png"/> indicate the global coordinate system.</p>
<h3>Bi-symmetric cross section</h3>
<p>The motion and the water velocity are decomposed into 3 components in the local element coordinate system, with <img class="formulaInl" alt="$ x $" src="form_163.png"/> in the axial direction, <img class="formulaInl" alt="$ y $" src="form_683.png"/> and <img class="formulaInl" alt="$ z $" src="form_387.png"/> normal to the axis, cfr. <a class="el" href="a00014.html#Riser_cross_sections">Riser cross sections</a>.</p>
<p>It is assumed that both the <img class="formulaInl" alt="$ y $" src="form_683.png"/> - and <img class="formulaInl" alt="$ z $" src="form_387.png"/> -axis are symmetry-axes for the cross-sectional surface geometry. The longitudinal, <img class="formulaInl" alt="$ F_x $" src="form_806.png"/> , force components and transverse force components <img class="formulaInl" alt="$ F_y $" src="form_807.png"/> and <img class="formulaInl" alt="$ F_z $" src="form_808.png"/> , on an element of length <img class="formulaInl" alt="$ dx $" src="form_809.png"/> subjected to waves and current can then be written:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ F_x = dx [( \rho A + m_{a}^x) \dot{u}_{x}^I- m_{a}^x \ddot{r}_x + C_D^x |U_x^c + u_x^I - \dot{r}_x| (U_x^c +u_x^I-\dot r_x) + C_{DL}^x (U_x^c +u_x^I-\dot r_x)] \tag{7.65} \]" src="form_810.png"/>
</p>
 <p class="formulaDsp">
<img class="formulaDsp" alt="\[ F_y = dx [( \rho A + m_{a}^y) \dot{u}_{y}^I- m_{a}^y \ddot{r}_y + C_D^y |U_y^c + u_y^I - \dot{r}_y| (U_y^c +u_y^I-\dot r_y) + C_{DL}^y (U_y^c +u_y^I-\dot r_y)] \tag{7.66} \]" src="form_811.png"/>
</p>
 <p class="formulaDsp">
<img class="formulaDsp" alt="\[ F_z = dx [( \rho A + m_{a}^z) \dot{u}_{z}^I- m_{a}^z \ddot{r}_z + C_D^z |U_z^c + u_z^I - \dot{r}_z| (U_z^c +u_z^I-\dot r_z) + C_{DL}^z (U_z^c +u_z^I-\dot r_z)] \tag{7.67} \]" src="form_812.png"/>
</p>
<p>here dots denote time derivatives and</p>
<ul>
<li><img class="formulaInl" alt="$ \rho \qquad \qquad \qquad \, $" src="form_813.png"/> : mass density of water</li>
<li><img class="formulaInl" alt="$ A \qquad\qquad\qquad $" src="form_814.png"/> : external cross-sectional area</li>
<li><img class="formulaInl" alt="$ m_a^x, m_a^y, m_a^z \quad \;\;\,\, $" src="form_815.png"/> : 2D added mass in local x-, y-and z-directions</li>
<li><img class="formulaInl" alt="$ (r_x, r_y, r_z) \qquad \:\: $" src="form_816.png"/> : translational structural displacement components in the local coordinate system</li>
<li><img class="formulaInl" alt="$ C_D^x, C_D^y, C_D^z \quad \;\;\: $" src="form_817.png"/> : dimensional quadratic drag coefficients in local x-, y- and z-directions</li>
<li><img class="formulaInl" alt="$ C_{DL}^x, C_{DL}^y, C_{DL}^z $" src="form_818.png"/> : dimensional linear drag coefficients in local x-, y- and z-directions</li>
<li><img class="formulaInl" alt="$ (U_x^c, U_y^c, U_z^c) \quad \;\; $" src="form_819.png"/> : current velocity components in the local coordinate system</li>
<li><img class="formulaInl" alt="$ (u_x^I, u_y^I, u_z^I) \qquad \; $" src="form_820.png"/> : velocity component of undisturbed wave field in the local coordinate system</li>
</ul>
<p>See also the User Manual for the relation between these force coefficients and commonly used non-dimensional coefficients.</p>
<p>It should be noted that oscillatory forces due to vortex shedding are neglected. The drag coefficients and added mass coefficients will in general depend on structural form, Reynolds number, roughness ratio, Keulegan-Carpenter number, current velocity relative to wave induced velocity, etc. The nature of the orbital motion of the undisturbed local wave field will have an influence. For circular cylinders there is some information available in the literature.</p>
<h3>Rotation symmetric cross section</h3>
<p>The linear part of the drag force and the inertia forces, which are also linear, are calculated as for the bi-symmetric cross section. The longitudinal quadratic drag force is also calculated as for the bisymmetric cross section.</p>
<p>The transverse quadratic drag force is calculated differently, using the total relative velocity normal to the line axis:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ F_y^D = F\cos\phi \tag{7.68} \]" src="form_821.png"/>
</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ F_z^D = F\sin\phi \tag{7.69} \]" src="form_822.png"/>
</p>
<p>where</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ F = dx C_D^y [(U_y^c + u_y^I - \dot r_y)^2 + (U_z^c + u_z^I - \dot r_z )^2] \tag{7.70} \]" src="form_823.png"/>
</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \phi = \textrm{arctan} \left[\frac{U_z^c + u_z^I - \dot r_z}{U_y^c +u_y^I - \dot r_y} \right], \quad (0 \leq \phi \leq 2\pi ) \tag{7.71} \]" src="form_824.png"/>
</p>
<p>The hydrodynamic roll-moment (i.e. moment about the x-axis) on an element of length <img class="formulaInl" alt="$ dx $" src="form_809.png"/> is given by:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ M_x = dx \left( \rho \left[ \iint\limits_A \textrm{d}{y}\textrm{d}{z} (u^2 - z^2) + A_{44}^{(2D)} \right] \frac{\partial}{\partial y} \dot u_z^I -A_{44}^{(2D)} \ddot \eta_4 \right) \tag{7.72} \]" src="form_825.png"/>
</p>
<p>Here <img class="formulaInl" alt="$ \ddot \eta_4 $" src="form_826.png"/> is the roll angle. Further <img class="formulaInl" alt="$ A_{44}^{(2D)} $" src="form_827.png"/> is the two-dimensional added moment in roll with respect to the local x-axis. For steady current we can write the Munk moment as</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ M_x = \left( A_{33}^{(2D)} - A_{22}^{(2D)} \right) U_z^cU_y^c \tag{7.73} \]" src="form_828.png"/>
</p>
<p><img class="formulaInl" alt="$ M_x $" src="form_829.png"/> is not implemented in the present version.</p>
<p>At every time step, forces and moments are calculated along the whole instantaneously submerged length of the riser. It is earlier discussed how the free surface position <img class="formulaInl" alt="$ \eta $" src="form_686.png"/> is calculated, which is important information to calculate the submerged length of the riser.</p>
<p>In the case of additional buoys, the hydrodynamic forces is handled in a similar way as described for the pipe cross sections.</p>
<h2><a class="anchor" id="tm_mfloads"></a>
MacCamy-Fuchs with Quadratic Drag</h2>
<p>For structures with relatively large diameter compared to the wave length, near-field diffraction effects become more important. An analytical solution for the potential around a vertical, surface-piercing circular cylinder was developed by MacCamy and Fuchs.</p>
<p>For elements with MacCamy-Fuchs type loading, the formulation for the inertia and diffraction forces follows (MacCamy and Fuchs, 1954 and Dean and Dalrymple, 1991), with modifications for the irregular wave history. The force per unit length ( <img class="formulaInl" alt="$ dF $" src="form_830.png"/>) at <img class="formulaInl" alt="$ x=0 $" src="form_831.png"/> for a given elevation <img class="formulaInl" alt="$ z $" src="form_387.png"/> is given by:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ dF(z) = \sum_{n=1}^{N} \frac{4 \rho g a_n}{k_n}\frac{\cosh{k_n(z+h)}}{\cosh{k_n h}}G \cos{(\omega_n t - \epsilon_n - \alpha_n)} \tag{7.74} \]" src="form_832.png"/>
</p>
<p>where </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \tan{\alpha_n} = \frac{J_1^\prime(k_n a)}{Y_1^\prime(k_n a)} \tag{7.75} \]" src="form_833.png"/>
</p>
<p>and </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ G = \frac{1}{\sqrt{\left(J_1^\prime(k_n a)\right)^2 + \left(Y_1^\prime(k_n a)\right)^2}} \tag{7.76} \]" src="form_834.png"/>
</p>
<p>and the wave components are defined as for the wave potential.</p>
<p>The resulting <img class="formulaInl" alt="$ C_m $" src="form_835.png"/> and phase angle for a section with diameter 10 m are shown in <a class="el" href="a00014.html#maccamy_force_figure">Effective mass coefficient and force phase, D=10m</a>.</p>
<p><a class="anchor" id="maccamy_force_figure"></a></p>
<div class="image">
<img src="maccamy_fuchs.svg" alt="maccamy_fuchs.svg"/>
<div class="caption">
Effective mass coefficient and force phase, D=10m</div></div>
<p>The model is only applicable for circular cross-sections and approximately vertical sections. The forces are computed based on the z-coordinate in the center of the element and are applied in a lumped manner.</p>
<p>In addition to the inertia force, a quadratic drag force based on linear wave kinematics may also be included when using MacCamy-Fuchs. The drag force is according to Morison's Equation, with the same implementation as in section <a class="el" href="a00014.html#tm_genmori">Generalized Morison's Equation</a>.</p>
<h1><a class="anchor" id="tm_load_forced"></a>
Forced Vessel Motion</h1>
<p>In many cases, the slender structure will be connected to a vessel at one end. Motions of this vessel must therefore be known during a dynamic riser analysis. Generation of motion time series will be consistent with generated time series for wave-induced water particle velocities and accelerations. The rigid body motion responses consist of 6 degrees of freedom: surge, sway, heave, roll, pitch and yaw referred to the global <img class="formulaInl" alt="$ (X,Y,Z) $" src="form_797.png"/> coordinate system. The motions are in this chapter denoted by <img class="formulaInl" alt="$ \boldsymbol{x} $" src="form_836.png"/> .</p>
<p>The motion model consists of a set of high-frequency (wave frequency) motions in all 6 degrees of freedom and a set of low-frequency motions in the 3 horizontal degrees of freedom: surge, sway and yaw. These two sets of motions are referred to as HF-motions and LF-motions, respectively. For most dynamic line problems it is sufficient to include only the HF-motions. The effects of typical LF-motions, with periods of 60-180 s, can often be covered by suitable selection of static (mean) position.</p>
<p><a class="anchor" id="HF_Vessel_Motion_Model"></a></p>
<h2><a class="anchor" id="tm_load_forced_hf"></a>
HF Vessel Motion Model</h2>
<p>The wave frequency-, or HF motions are treated as linear responses to the waves. Thus, the HF motions are described by a set of complex transfer functions <img class="formulaInl" alt="$ j = 1, 2.....6$" src="form_837.png"/> where:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ H_{HFj}(\beta, \omega) = \frac{x_j(\beta, \omega)}{\zeta_a(\beta, \omega)} \tag{7.77} \]" src="form_838.png"/>
</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ S_{xj}(\beta, \omega) = |H_{HFj}(\beta, \omega )|^2 S_\zeta(\beta, \omega ) \tag{7.78} \]" src="form_839.png"/>
</p>
<h2><a class="anchor" id="tm_load_forced_lf"></a>
LF Vessel Motion Model</h2>
<p>The low-frequency-, or LF-motions, if included, are represented by directly input response spectra for surge, sway and yaw. The LF motions modelled this way are uncorrelated with the wave, and with each other.</p>
<p>Both the transfer functions of HF motions and spectra for LF motions refer to a specified vessel origin point. They are transformed to the structural attachment points on the vessel, (i.e. upper end terminal point for standard systems). This information together with the wave spectrum description, is sufficient to generate spectra and the sample time series of forced vessel motions.</p>
<h2><a class="anchor" id="tm_load_forced_frequency"></a>
Frequency-domain Vessel Motion Analysis</h2>
<h3>HF Responses</h3>
<p>In the linearized, frequency-domain analysis the wave frequency motions are completely described by their autospectra, <img class="formulaInl" alt="$ S_x $" src="form_840.png"/> .</p>
<p>Response spectra are generated for all 6 rigid-body motions.</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ S_{HFxj}(\omega) = \int_{\beta_1 - \frac{\pi}{2} }^{\beta_1 + \frac{\pi}{2}}\!{ | H_{xj}'(\beta, \omega)|^2 S_{\zeta, \textrm{TOT}} (\beta, \omega)}\textrm{d}{\beta} , \quad j = 1,2...6 \tag{7.79} \]" src="form_841.png"/>
</p>
<p>where <img class="formulaInl" alt="$H_{xj}'(\beta, \omega) $" src="form_842.png"/> is the transfer function referring to the top terminal point.</p>
<h3>LF-Responses</h3>
<p><img class="formulaInl" alt="$ S_{LFxj}(\omega)$" src="form_843.png"/> is given as direct input and subsequently used for time series generation.</p>
<h3>HF time series generation</h3>
<p>Input to the HF time series generation is the complex motion transfer function <img class="formulaInl" alt="$ H_j(\beta_k, \omega_l) $" src="form_844.png"/> , and the harmonic wave components, <img class="formulaInl" alt="$ Z_{kl} $" src="form_845.png"/> .</p>
<p>The harmonic components of the motion responses are simply:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ X_{j,l} = \sum_{\textrm{k} = 1}^{N_\textrm{k}}\, H_j(\beta_\textrm{k}, \omega_l)Z_{\textrm{k}l}, \quad j=1,2...6 \tag{7.80} \]" src="form_846.png"/>
</p>
<p>The transfer functions <img class="formulaInl" alt="$ H $" src="form_847.png"/> are stored on a tabular form for given arrays of directions and frequencies different from <img class="formulaInl" alt="$ \beta_\textrm{k} $" src="form_848.png"/> and <img class="formulaInl" alt="$ \omega_l $" src="form_849.png"/> , intermediate values are obtained by interpolation (3rd order spline interpolation). Symmetry properties of the platform are utilized.</p>
<p>Time series are obtained by adding the harmonic components by means of an FFT algorithm.</p>
<p>For regular wave analysis, the HF-time series may optionally be given as direct input of amplitudes and phase angles of the top terminal point of the structure.</p>
<h3>LF time series generation</h3>
<p>Input to the LF time series generation are the complex motion response spectra <img class="formulaInl" alt="$ S_{LF,xj} (\omega), j = 1,2,6 $" src="form_850.png"/></p>
<p>The harmonic components of the motion responses are:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{array}{l} |x_{j\textrm{k}}| = \sqrt{2S_{LF,xj}(\omega_\textrm{k}) \Delta \omega} \\ \\ \textrm{arg} (x_{j\textrm{k}}) = \phi_{xj\textrm{k}} \end{array} \tag{7.81} \]" src="form_851.png"/>
</p>
<p>The random phase angles, <img class="formulaInl" alt="$ \phi_{xjk} $" src="form_852.png"/> , are sampled from a uniform distribution over <img class="formulaInl" alt="$ [-\pi,\pi] $" src="form_853.png"/> .</p>
<p>Addition of the harmonic components to obtain time series is performed by a Fourier Transform algorithm.</p>
<p>The relations between time increment, <img class="formulaInl" alt="$ \Delta t_{LF} $" src="form_854.png"/> , frequency increment <img class="formulaInl" alt="$ \Delta \omega_{LF} $" src="form_855.png"/> and number of time steps for the LF-simulations is equivalent to those described in <a class="el" href="a00014.html#tm_eq_7_20">Eq. (7.20)</a> except that, <img class="formulaInl" alt="$ t_{\textrm{lim} } $" src="form_856.png"/> , the minimum interesting low frequency response period is greater, typically 20-30 sec.</p>
<h3>Total Responses</h3>
<p>Time series of total responses for the surge, sway and yaw motions are obtained by adding the HF and LF contributions directly.</p>
<p>If LF-responses have been generated with larger time increment than the HF-responses, a 3rd order spline interpolation is applied to obtain identical time increments.</p>
<h1><a class="anchor" id="tm_load_effects"></a>
Effects from Internal Fluid Flow</h1>
<p>The riser system model includes specification of internal fluid flow by:</p>
<table class="doxtable">
<tr>
<th>Parameter </th><th>Symbol </th><th>Dimension  </th></tr>
<tr>
<td> Cross section area of internal volume</td><td> <img class="formulaInl" alt="$ A_i $" src="form_857.png"/></td><td> <img class="formulaInl" alt="$ \mathrm{L^2} $" src="form_858.png"/> </td></tr>
<tr>
<td> Density </td><td> <img class="formulaInl" alt="$ \rho_i $" src="form_859.png"/> </td><td> <img class="formulaInl" alt="$ \mathrm{M/L^2} $" src="form_860.png"/> </td></tr>
<tr>
<td> Velocity</td><td> <img class="formulaInl" alt="$ v_i $" src="form_67.png"/> </td><td> <img class="formulaInl" alt="$ \mathrm{L/T} $" src="form_861.png"/> </td></tr>
<tr>
<td> Pressure gradient due to flow resistance </td><td> <img class="formulaInl" alt="$ \rho d_i $" src="form_862.png"/> </td><td> <img class="formulaInl" alt="$ \mathrm{F/L^3}$" src="form_863.png"/> </td></tr>
<tr>
<td> Pressure at end 1, including the effect of flow density</td><td><img class="formulaInl" alt="$ \rho_{1i} $" src="form_864.png"/> </td><td> <img class="formulaInl" alt="$ \mathrm{F/L^2} $" src="form_865.png"/> </td></tr>
</table>
<p>The following simplifications are made:</p>
<ol type="1">
<li>The fluid is incompressible.</li>
<li>The velocity and pressure distribution are assumed to be un-affected by the riser motion.</li>
<li>The pressure is included only in order to account for possible future pressure-dependent stiffness of nonbounded flexible pipes and for axial wall force evaluation.</li>
</ol>
<p>The internal flow as modelled will affect the riser configuration and behaviour in the following ways:</p>
<p>1) The effect of weight and mass</p>
<p>Vertical force per unit length due to internal fluid:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ q_i = A_i\rho_ig \tag{7.82} \]" src="form_866.png"/>
</p>
<p>This weight will affect the effective tension and thus the configuration of the riser. The weight will also affect the internal pressure distribution.</p>
<p>2) The effect of internal pressure</p>
<p>The bending stiffness of the nonbounded flexible pipes is assumed to be dependent upon the pressure differential,</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{array}{l} p_d = p_e-p_i \\ \\ p_e = \begin{cases} p_wg(-z), &amp; z<0 \\ 0, &amp; z\geq0 \end{cases} \\ \\ p_i = p_{1i} - \rho_igz - p_{di}s \end{array} \tag{7.83} \]" src="form_867.png"/>
</p>
<p>where</p>
<ul>
<li><img class="formulaInl" alt="$ \rho_w,\rho_i $" src="form_868.png"/> = density of water and internal fluid, respectively</li>
<li><img class="formulaInl" alt="$ p_{di} $" src="form_869.png"/> = pressure drop per unit length due to flow resistance</li>
<li><img class="formulaInl" alt="$ s $" src="form_771.png"/> = coordinate along the riser</li>
<li><img class="formulaInl" alt="$ p_{1i} $" src="form_870.png"/> = internal pressure at lower end (end 1)</li>
</ul>
<p>This information is not utilized in the present version.</p>
<p>3) The static effect of the impulse and centrifugal force</p>
<p>The centrifugal force effect may be included in the riser loads. This force depends upon local curvature, velocity and mass of fluid flow.</p>
<p><a class="anchor" id="Centrifugal_force_effect"></a></p>
<div class="image">
<img src="RIFLEX_TheoryManual_42_v0_869.svg" alt="RIFLEX_TheoryManual_42_v0_869.svg"/>
<div class="caption">
Centrifugal force effect</div></div>
<p>The centrifugal force per unit length can be expressed by:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{array}{l} q_c = m_iv_i^2/R \\ m_i = \rho_iA_i \end{array} \tag{7.84} \]" src="form_871.png"/>
</p>
<p>This force will act radially on any curved part of the riser. However, the force will also contribute to increase the effective riser tension:</p>
<p><a class="anchor" id="tm_eq_7_85"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ f_e = f_{eo} + mv_i^2 \tag{7.85} \]" src="form_872.png"/>
</p>
<p>Comparing the two cases a) and b) in figure <a class="el" href="a00014.html#Centrifugal_force_effect">Centrifugal force effect</a> for a case with zero bending stiffness, the radii of curvature can be expressed by:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{array}{l} \textrm{a)} \quad R_o = (f_{eo} /q_o) \\ \\ \textrm{b)} \quad R = (f_{eo} + mv_i^2 ) / (q_o + mv_i^2/R) = f_{eo} /q_o = R_o \end{array} \tag{7.86} \]" src="form_873.png"/>
</p>
<p>Thus, the centrifugal force will not contribute to change the static configuration. The only effect is the increase of wall tension due to the impulse force, according to <a class="el" href="a00014.html#tm_eq_7_85">Eq. (7.85)</a> written:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{q_I} = -m_i \boldsymbol{\dot v} \tag{7.87} \]" src="form_874.png"/>
</p>
<p>where <img class="formulaInl" alt="$ \boldsymbol{\dot v} $" src="form_875.png"/> is the acceleration vector of the fluid. Assuming stationary, incompressible flow, the acceleration can be written:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{\dot v} = v_i^2 \frac{\partial \boldsymbol{e} }{\partial s} + \boldsymbol{\ddot r} - 2v_i (\boldsymbol{\omega} \times \boldsymbol{e}) \tag{7.88} \]" src="form_876.png"/>
</p>
<p>where</p>
<p><img class="formulaInl" alt="$ \frac{\partial \boldsymbol{e} }{\partial s} $" src="form_877.png"/> = curvature vector</p>
<p><img class="formulaInl" alt="$ \boldsymbol{r} $" src="form_452.png"/> = acceleration of pipe</p>
<p><img class="formulaInl" alt="$ \boldsymbol{\omega} $" src="form_878.png"/> = rotation of pipe</p>
<p><img class="formulaInl" alt="$ \boldsymbol{e} $" src="form_879.png"/> = pipe tangent unit vector</p>
<p><img class="formulaInl" alt="$ s $" src="form_771.png"/> = curve coordinate along the pipe</p>
<p>Thus, the following force terms are derived:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{q_I} = -m_i \boldsymbol{\ddot r} -m_iv_i^2 \frac{\partial \boldsymbol{e} }{\partial s} -2m_i v_i (\boldsymbol{\omega} \times \boldsymbol{e} ) \tag{7.89} \]" src="form_880.png"/>
</p>
<p>The force terms accounts for:</p>
<ol type="1">
<li>Force due to pipe acceleration</li>
<li>Force due to centripetal acceleration</li>
<li>Force due to pipe rotation (Coriolis force)</li>
</ol>
<p>The present program version includes only the first term. It is included directly in the mass matrix of the elements.</p>
<p>A special slug modelling option is included to simulate the effect of a relatively short fluid slug moving through a riser. This is done by updating the mass matrix according to a specified slug speed in the pipe.</p>
<p>The second term is identical with the static centrifugal force discussed previously, but the curvature will be time-dependent in the dynamic case.</p>
<p>The third term may have more complicated effects.</p>
<p>Both the second and the third terms may be included directly as forces in the time domain analysis in a possible future extension of the program.</p>
<h1><a class="anchor" id="tm_load_hydrodynamic_on"></a>
Hydrodynamic Load on Partly Submerged, Floating Elements</h1>
<p>This section will give an introduction to the hydrodynamic load models for floating partly submerged elements, which are relevant for analysis of floating flexible fish farm systems, floating offshore loading hoses, etc. The presentation is to a large extent based on Ormberg (1991).</p>
<p>A floating body of small volume may, during a wave cycle, accomplish large motions from its mean position. Thus, the hydrodynamic loads should be computed at the instantaneous position rather than at the mean position, which means that the assumptions within linear potential theory are violated. The applied hydrodynamic load model, described by Aarsnes et al. (1988), is an engineering approach to a solution of the problem. It involves hydrodynamic coefficients deduced by linear theory. In addition, an adequate description of the wave kinematics in the wave zone is required. Application of the model including comparison with model test results is described by Ormberg (1991).</p>
<p>The hydrodynamic forces are calculated based on two-dimensional strip theory, which requires the structural members to be long and slender. Hydrodynamic interaction between structural parts is neglected. The wave-induced excitation forces (Froude-Krylov and diffraction forces) are computed by a long wavelength approximation which involves added mass and potential damping of the actual cross section together with the wave kinematics. The viscous loads are computed using the drag term in Morison's equation.</p>
<p>In the following, a rigid structural member (section) with constant cross sectional dimensions, is considered. The section with local (co-rotating) coordinate system <img class="formulaInl" alt="$ (x, y, z) $" src="form_881.png"/> , is illustrated in figure <a class="el" href="a00014.html#Structural_member_divided_into_sub-elements">Structural member divided into sub-elements</a>. The figure also defines the space-fixed (global) coordinate system with X- and Y-axes in the still water plane and with the Z-axis pointing upwards.</p>
<p><a class="anchor" id="Structural_member_divided_into_sub-elements"></a></p>
<div class="image">
<img src="RIFLEX_TheoryManual_42_v0_883.svg" alt="RIFLEX_TheoryManual_42_v0_883.svg"/>
<div class="caption">
Structural member divided into sub-elements, for hydrodynamic load calculation</div></div>
<p>As indicated in <a class="el" href="a00014.html#Structural_member_divided_into_sub-elements">Structural member divided into sub-elements, for hydrodynamic load calculation</a> , the section is divided into sub-elements for hydrodynamic load calculation. The forces acting on a sub-element are calculated based on the instantaneous orientation and position relative to the free surface. The water particle kinematics, which are assumed to be constant within the sub-element, are calculated at the buoyancy center of the sub-element. Correspondingly, the load per unit length is constant over the length of the sub-element and attacks through the buoyancy center. The total force normal to the section is found by simply adding the force contribution on each sub-element.</p>
<p>For the computer program implementation, a table of the two-dimensional added mass and damping coefficients for different levels of immersion at the actual wave frequency is pre-computed by use of the Frank close fit method, Frank (1967).</p>
<p>In the following, the hydrodynamic forces are understood as the total forces acting on the member from the surrounding fluid. This implies that buoyancy and current forces are also included in the hydrodynamic forces. Thus, the total hydrodynamic forces are given as</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{F^H} =\boldsymbol{F^{\mathrm{Pot} }} + \boldsymbol{F^D} = \boldsymbol{F^{FK}} + \boldsymbol{F^S} + \boldsymbol{F^R} + \boldsymbol{F^D} \tag{7.90} \]" src="form_882.png"/>
</p>
<p>where:</p>
<p><img class="formulaInl" alt="$ \boldsymbol{F^{Pot}} $" src="form_883.png"/> : the potential flow contribution to hydrodynamic forces</p>
<p><img class="formulaInl" alt="$ \boldsymbol{F^{FK} } $" src="form_884.png"/> : Froude-Krylov forces including buoyancy forces</p>
<p><img class="formulaInl" alt="$ \boldsymbol{F^S} $" src="form_885.png"/> : diffraction forces</p>
<p><img class="formulaInl" alt="$ \boldsymbol{F^R} $" src="form_886.png"/> : added mass and damping forces</p>
<p><img class="formulaInl" alt="$ \boldsymbol{F^D} $" src="form_887.png"/> : the viscous forces or drag forces</p>
<h2><a class="anchor" id="tm_load_hydrodynamic_on_potential"></a>
Potential Flow Contribution</h2>
<h3>Loads in longitudinal direction</h3>
<p>The long and slender shape assumption implies that the forces on the end surfaces are dominated by the Froude-Krylov contribution. Hence, other contributions are neglected. The importance of including the pressure force on end plates is thoroughly discussed by Hooft (1972).</p>
<p>The end pressure forces are calculated based on dynamic Froude-Krylov pressure and the hydrostatic pressure</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ p_{\textrm{tot}} + p_{\textrm{dyn}} + p_{\textrm{st}} = p_{\textrm{dyn}} -\rho gZ \tag{7.91} \]" src="form_888.png"/>
</p>
<p>where <img class="formulaInl" alt="$ g $" src="form_273.png"/> is the acceleration due to gravity and <img class="formulaInl" alt="$ \rho $" src="form_889.png"/> is the density of water.</p>
<p>Thus, the total potential contribution to load in local x-direction of the structural member is found by the vector sum of the end pressure forces, which formally is written</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{F_x^{FK}} = \left( \int_{A_w^a}\!{p_{\mathrm{tot}}^{(a)}}\textrm{d}{A} - \int_{A_w^b}\!{p_{\mathrm{tot}}^{(b)}}\textrm{d}{A} \right) \boldsymbol{i_I} \tag{7.92} \]" src="form_890.png"/>
</p>
<p>where <img class="formulaInl" alt="$ A_w $" src="form_891.png"/> is the instantaneous immersed area of the end surface. The indexes (a) and (b) indicate the end surface number, see <a class="el" href="a00014.html#End_pressure_forces">End pressure forces</a> For practical purposes it is assumed that the following simplified formula is sufficient</p>
<p><a class="anchor" id="tm_eq_7_93"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{F_x^{FK} } = \left( p_{\textrm{tot}}^{(a)} A_w^{(a)} - p_{\textrm{tot}}^{(b)} A_w^{(b)} \right) \boldsymbol{i_I} \tag{7.93} \]" src="form_892.png"/>
</p>
<p>where <img class="formulaInl" alt="$ p_\textrm{tot} $" src="form_893.png"/> is calculated at the wetted area center.</p>
<p><a class="anchor" id="End_pressure_forces"></a></p>
<div class="image">
<img src="RIFLEX_TheoryManual_42_v0_897.svg" alt="RIFLEX_TheoryManual_42_v0_897.svg"/>
<div class="caption">
End pressure forces</div></div>
<h3>Froude-Krylov (FK) forces:</h3>
<p>The Froude-Krylov forces including the buoyancy forces acting normal to a sub-element may according to Kaplan and Hue (1959) and Dixon et al. (1979), be computed by</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \Delta \boldsymbol{F_n^{FK} } = \boldsymbol{f_n^{FK} } \Delta x = \left( \boldsymbol{f_n^{\mathrm{Buoy} } } + \rho A_S \boldsymbol{\dot u_n} \right) \Delta x \tag{7.94} \]" src="form_894.png"/>
</p>
<p>where <img class="formulaInl" alt="$ \boldsymbol{f_n^{\textrm{Buoy} } } $" src="form_895.png"/> is the buoyancy force calculated based on the instantaneous immersed cross section area. <img class="formulaInl" alt="$\Delta x $" src="form_896.png"/> is the length of the sub-element. <img class="formulaInl" alt="$ \boldsymbol{\dot u_n} $" src="form_897.png"/> is the water particle acceleration vector component normal to the sub-element, i.e.</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{\dot u_n} = \dot u_y \boldsymbol{i_2} + \dot u_z \boldsymbol{i_3} \tag{7.95} \]" src="form_898.png"/>
</p>
<p>The buoyancy force referred to the global system is simply calculated by</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{\tilde{f}^{\mathrm{\,Buoy} }} = \rho g A_S \boldsymbol{I_3} \tag{7.96} \]" src="form_899.png"/>
</p>
<p>where <img class="formulaInl" alt="$ A_S $" src="form_900.png"/> is the immersed cross section area of the sub-element.</p>
<p>Transforming the buoyancy load to local system, neglecting the axial contribution to avoid double representation of this term, the FK-force referred to local system is written</p>
<p><a class="anchor" id="tm_eq_7_97"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \Delta \boldsymbol{F_n^{FK}} = \boldsymbol{f_n^{FK}} \Delta x = \left( \rho g A_S T_{23} + \rho A_S \dot u_y \right) \Delta x\boldsymbol{i_2} + \left( \rho g A_S T_{33} + \rho A_S \dot u_z \right) \Delta x \boldsymbol{i_3} \tag{7.97} \]" src="form_901.png"/>
</p>
<p>where <img class="formulaInl" alt="$ \boldsymbol{T_{ij} } $" src="form_902.png"/> is the transformation matrix between the global and local coordinate systems, i.e.</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{i_i} = \boldsymbol{T_{ij} I_j} \tag{7.98} \]" src="form_903.png"/>
</p>
<h3>Diffraction forces:</h3>
<p>According to the long wave assumption, the diffraction forces may be expressed in terms of two-dimensional added mass and damping</p>
<p><a class="anchor" id="tm_eq_7_99"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \Delta \boldsymbol{F_n^S} = \boldsymbol{f_n^S} \Delta x= \left( A_{22}^{(2D)} \boldsymbol{\dot u_y} + B_{22}^{(2D)} \boldsymbol{u_y} \right) \Delta x\boldsymbol{i_2} + \left( A_{33}^{(2D)} \boldsymbol{\dot u_z} + B_{33}^{(2D)} \boldsymbol{u_z} \right) \Delta x\boldsymbol{i_3} \tag{7.99} \]" src="form_904.png"/>
</p>
<p>where <img class="formulaInl" alt="$ A^{(2D)} = A^{(2D)} (Z_\mathrm{rel} )$" src="form_905.png"/> and <img class="formulaInl" alt="$ B^{(2D)} = B^{(2D)}(Z_{\mathrm{rel} }) $" src="form_906.png"/> are position-dependent added mass and damping per unit length, respectively. <img class="formulaInl" alt="$ Z_\mathrm{rel} $" src="form_907.png"/> is the vertical position relative the free surface.</p>
<h3>Added mass and damping forces:</h3>
<p>The added mass and damping forces including moment around local x-axis are calculated according to</p>
<p><a class="anchor" id="tm_eq_7_100"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{bmatrix} \Delta F_y^R\\ \Delta F_z^R\\ \Delta F_{\theta x}^R \end{bmatrix} = \left( \begin{bmatrix} A_{22}^{(2D)} &amp; 0 &amp; A_{24}^{(2D)} \\ 0 &amp; A_{33}^{(2D)} &amp; 0 \\ A_{42}^{(2D)} &amp; 0 &amp; A_{44}^{(2D)} \end{bmatrix} \begin{bmatrix} \ddot v_y \\ \ddot v_z \\ \ddot v_{\theta x} \end{bmatrix} + \begin{bmatrix} B_{22}^{(2D)} &amp; 0 &amp; B_{24}^{(2D)}\\ 0 &amp; B_{33}^{(2D)} &amp; 0 \\ B_{42}^{(2D)} &amp; 0 &amp; B_{44}^{(2D)} \end{bmatrix} \begin{bmatrix} \dot v_y \\ \dot v_z \\ \dot v_{\theta x} \end{bmatrix} \right) \tag{7.100} \]" src="form_908.png"/>
</p>
<p>where <img class="formulaInl" alt="$ \boldsymbol{\dot v} $" src="form_875.png"/> and <img class="formulaInl" alt="$ \boldsymbol{\ddot v} $" src="form_909.png"/> are the structural velocity and acceleration, respectively.</p>
<h2><a class="anchor" id="tm_load_hydrodynamic_on_simplified"></a>
Simplified Approach</h2>
<p>A further simplification of the hydrodynamic problem is included by introducing constant added mass coefficients, i.e. neglecting the depth dependency. This means that the added mass is taken as proportional to the submerged cross sectional area. Accordingly <a class="el" href="a00014.html#tm_eq_7_99">(Eq. 7.99)</a> and <a class="el" href="a00014.html#tm_eq_7_100">(Eq. (7.100)</a> simplify to</p>
<p><a class="anchor" id="tm_eq_7_101"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \Delta \boldsymbol{F_n^S} = \boldsymbol{f_n^S} \Delta x = \rho A_S C_{my} \dot u_y \Delta x \boldsymbol{i_2} + \rho A_S C_{mz} \dot u_z \Delta x \boldsymbol{i_3} \tag{7.101} \]" src="form_910.png"/>
</p>
<p>and</p>
<p><a class="anchor" id="tm_eq_7_102"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \Delta \boldsymbol{F_n^R} = \boldsymbol{f_n^R} \Delta x = -\rho A_S C_{my} \ddot v_y \Delta x \boldsymbol{i_2} - \rho A_S C_{mz} \ddot v_z \Delta x \boldsymbol{i_3} \tag{7.102} \]" src="form_911.png"/>
</p>
<p>where <img class="formulaInl" alt="$ A_S $" src="form_900.png"/> is submerged cross section area. <img class="formulaInl" alt="$ C_{my} $" src="form_912.png"/> and <img class="formulaInl" alt="$ C_{mz} $" src="form_913.png"/> are constant added mass coefficients referred to completely submerged cross section.</p>
<h2><a class="anchor" id="tm_load_hydrodynamic_on_viscous"></a>
Viscous Forces</h2>
<p>The viscous forces are computed using the drag force term of Morison's equation. The cross flow principle first introduced by Hoerner (1965), is used to determine the forces in transverse direction of the sub-elements. The drag coefficients are assumed to be constant and given for a submerged cross section.</p>
<p>The viscous forces or drag forces are calculated based on relative velocity ( <img class="formulaInl" alt="$ \boldsymbol{u_r} $" src="form_914.png"/>) in the local system expressed by</p>
<p><a class="anchor" id="tm_eq_7_103"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{u_r} = \boldsymbol{u_c} + \boldsymbol{u_w} - \boldsymbol{\dot v} = u_{rx} \boldsymbol{i_1} + u_{ry}\boldsymbol{i_2} + u_{rz}\boldsymbol{i_3} \tag{7.103} \]" src="form_915.png"/>
</p>
<p>where <img class="formulaInl" alt="$ u_c $" src="form_916.png"/> is the current velocity, <img class="formulaInl" alt="$ u_w $" src="form_917.png"/> is the wave particle velocity and <img class="formulaInl" alt="$ \dot v $" src="form_918.png"/> is the structural velocity.</p>
<h3>Loads in longitudinal direction</h3>
<p>A friction force contribution is included in the axial direction</p>
<p><a class="anchor" id="tm_eq_7_104"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \Delta \boldsymbol{F^D_x} = \boldsymbol{f_x^D} \Delta x = \frac{1}{2} \rho C_{Dx}L_W |u_{rx}|u_{rx}\Delta x \boldsymbol{i_1} \tag{7.104} \]" src="form_919.png"/>
</p>
<p>where <img class="formulaInl" alt="$ C_{Dx} $" src="form_920.png"/> is the skin friction coefficient and <img class="formulaInl" alt="$ L_W $" src="form_921.png"/> is the instantaneous wetted part of the cross section circumference.</p>
<h3>Transverse loads</h3>
<p>From <a class="el" href="a00014.html#tm_eq_7_103">Eq. (7.103)</a>, the transverse relative velocity vector is written</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{u_{rn}} = u_{ry} \boldsymbol{i_2} + u_{rz}\boldsymbol{i_3} \tag{7.105} \]" src="form_922.png"/>
</p>
<p>The transverse drag force is calculated according to</p>
<p><a class="anchor" id="tm_eq_7_106"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \Delta \boldsymbol{F_n^D} = \boldsymbol{f_n^D} \Delta x = \frac{1}{2} \rho C_{Dy} h_{\mathrm{rel} } |\boldsymbol{u_{rn}}| u_{ry} \Delta x \boldsymbol{i_2} + \frac{1}{2} \rho C_{Dz} b_\mathrm{rel} |\boldsymbol{u}_{rn}| u_{rz} \Delta x \boldsymbol{i_3} \tag{7.106} \]" src="form_923.png"/>
</p>
<p>where</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{array}{l} \displaystyle h_\mathrm{rel} = \frac{A_S}{A} h \\ \\ \displaystyle b_\mathrm{rel}\, = \frac{A_S}{A} b \end{array} \]" src="form_924.png"/>
</p>
<p><img class="formulaInl" alt="$ C_{Dy} $" src="form_925.png"/> and <img class="formulaInl" alt="$ C_{Dz} $" src="form_926.png"/> are drag coefficients in local y and z directions respectively, <img class="formulaInl" alt="$ A_S $" src="form_900.png"/> is the instantaneous submerged cross section area and <img class="formulaInl" alt="$ A $" src="form_243.png"/> is the cross section area. <img class="formulaInl" alt="$ b $" src="form_927.png"/> and <img class="formulaInl" alt="$ h $" src="form_679.png"/> are characteristic width (y-direction) and height (z-direction) of the section respectively.</p>
<h2><a class="anchor" id="tm_load_hydrodynamic_on_moment"></a>
Moment around Local x-axis</h2>
<p>Assuming the transverse forces attack at the center of buoyancy, see <a class="el" href="a00014.html#Excitation_forces_acting_at_the_buoyancy_center">Excitation forces acting at the buoyancy center</a>, the moment around the local x-axis due to wave excitation forces and drag forces is approximated by</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \Delta \boldsymbol{F_{\theta x}^h} = \boldsymbol{f_{\theta x}^h} \Delta x = \left( \boldsymbol{r_B} \times \boldsymbol{f_n^h} \right) \Delta x \tag{7.107} \]" src="form_928.png"/>
</p>
<p>where <img class="formulaInl" alt="$ \boldsymbol{r_B} = y_B \boldsymbol{i_2} + z_B \boldsymbol{i_3} $" src="form_929.png"/> is the position vector of the buoyancy center and the transverse hydrodynamic excitation forces are given by</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{f_n^h} = \boldsymbol{f_n^{FK}} + \boldsymbol{f_n^S} + \boldsymbol{f_n^D} \tag{7.108} \]" src="form_930.png"/>
</p>
<p><a class="anchor" id="Excitation_forces_acting_at_the_buoyancy_center"></a></p>
<div class="image">
<img src="RIFLEX_TheoryManual_42_v0_938.svg" alt="RIFLEX_TheoryManual_42_v0_938.svg"/>
<div class="caption">
Excitation forces acting at the buoyancy center</div></div>
<h2><a class="anchor" id="tm_load_hydrodynamic_on_discussion"></a>
Discussion</h2>
<p>Computation of added mass and potential damping according to the Frank close fit method requires that the structural member is horizontal, i.e. the local xy-plane is parallel with the still water plane, and that the cross section is symmetric about its local x-z plane. During time domain simulation, the added mass and damping are taken as direction-dependent coefficients connected to local (structural) axis directions, while strictly they are valid for global directions. Hence, introduced as a practical limit, the instantaneous pitch and roll angles have to be within <img class="formulaInl" alt="$ \pm 30 ^{\circ} $" src="form_931.png"/> inclination to the surface.</p>
<p>The loads due to slamming or water impact are not included. These may be included as the time rate of change of momentum associated with the immersed portion of the structure. Calculating the vertical impact forces on horizontal members for the water entry case is well documented by e.g. Greenhow (1987), Kaplan and Silbert (1976), Faltinsen (1977). However, in this more general case regarding structure orientation, water exit as well as entry, further investigation is needed. Hence, this term is neglected.</p>
<p>The long wave length assumption is used in the calculation of potential flow acting on the sub-element, which normally requires the wave length / diameter ratio to be greater than 5. It is further assumed that the local angular displacements are not too large relative to the free surface. This assumption is due to that the computation of hydrodynamic coefficients are precalculated for a horizontal structural member.</p>
<p>Experiments indicate that the drag coefficients for a cross section in the free surface zone are strongly dependent on the submergence, see e.g. Øritsland (1986). These effects are not included. The dependency on Reynolds number, Keulegan-Carpenter number and roughness should be taken into account when specifying the constant drag coefficients as input.</p>
<h2><a class="anchor" id="tm_load_hydrodynamic_on_implementation"></a>
Implementation in Finite Element Formulation</h2>
<p>Different formulations might be used for the numerical implementation of hydrodynamic loads. These consider the order of the polynomial used for the displacement functions and the representation of the distribution of the hydrodynamic element loads. In the current version of the program, a consistent formulation is adopted for the numerical implementation of hydrodynamic loads, i.e. using the same displacement functions as for the element deformation description. The contributions from added mass and damping forces are included in the inertia and damping load vectors of the system, while the other hydrodynamical loads contribute to the external load vector.</p>
<p>Analogous to hydrodynamic strip theory, the load distribution is taken care of by dividing the element into a specified number of sub-elements for load calculation, see figure <a class="el" href="a00014.html#Element_divided_into_sub-elements_for_hydrodynamic_load_calculation">Element divided into sub-elements for hydrodynamic load calculation</a>. Within a sub-element, the hydrodynamic loads are assumed to be constant. The corresponding hydrodynamic mass- and damping matrices and nodal load vector are computed by numerical integration simply by adding up the contribution from each sub-element. This is certainly not the most accurate method for numerical integration, but is found to be adequate considering the uncertainties in the hydrodynamic load computation.</p>
<p>Based on this formulation, the element length may be chosen with regard to structural deformations of the actual problem, while the lengths of the sub-elements are chosen with regard to hydrodynamic load distribution independently of the element length.</p>
<p>The hydrodynamic loads may in general be described as non-conservative loads i.e. dependent on position, direction and/or loaded area. By linearizing the incremental dynamic equation, tangential mass, damping and stiffness matrices are introduced, see chapters <a class="el" href="a00004.html#finite_element">Finite Element Formulation</a> and <a class="el" href="a00010.html#dynamic_time">Dynamic Time Domain Analysis</a>. However, the linearization also introduces an additional stiffness due to the displacement dependent loads. This stiffness is called load correction stiffness, are described in section <a class="el" href="a00008.html#tm_static_finite_incremental">Incremental Equilibrium Iterations</a>. The inclusion of the load correction stiffness only has a consequence for the rate of convergence in a numerical solution procedure. Provided uniqueness of the solution, it is the equilibrium equation alone that governs the final solution. A thorough discussion regarding type of loads and the consequence of implementing the corresponding stiffness matrix is presented by Mathisen (1990).</p>
<p>In the current version of the program, a load correction stiffness matrix associated with the water plane stiffness (buoyancy springs) is included.</p>
<p>In the following, the formulation of hydrodynamic loads are described as developed for beam elements. This formulation applies for the bar elements by use of corresponding displacement functions and by omitting rotational degrees of freedom.</p>
<p><a class="anchor" id="Element_divided_into_sub-elements_for_hydrodynamic_load_calculation"></a></p>
<div class="image">
<img src="RIFLEX_TheoryManual_42_v0_940.svg" alt="RIFLEX_TheoryManual_42_v0_940.svg"/>
<div class="caption">
Element divided into sub-elements for hydrodynamic load calculation</div></div>
<h3>Hydrodynamic mass and damping matrices</h3>
<p>The hydrodynamic (added) mass and damping matrices are established based on the hydrodynamic mass and damping calculated along the secant length of the deformed element.</p>
<p>The components of these matrices are computed according to</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{array}{l} \begin{aligned} m_{vv}^h &amp;= \sum_{\textrm{i = 1} }^{\textrm{nsub}}\, \boldsymbol{N_v^T} (x_\textrm{i} ) \Delta m_{vv}^h (x_i)\boldsymbol{N_v} (x_i)\\ m_{ww}^h &amp;= \sum_{\textrm{i = 1} }^{\textrm{nsub}}\, \boldsymbol{N_w^T} (x_\textrm{i} ) \Delta m_{ww}^h (x_i)\boldsymbol{N_w} (x_i)\\ m_{{\theta_x}{\theta_x}}^h &amp;= \sum_{\textrm{i = 1} }^{\textrm{nsub}}\, \boldsymbol{N_{\theta_x}^T} (x_\textrm{i} ) \Delta m_{{\theta_x}{\theta_x}}^h (x_i)\boldsymbol{N_{\theta_x}} (x_i)\\ m_{v{\theta_x}}^h &amp;= \sum_{\textrm{i = 1} }^{\textrm{nsub}}\, \boldsymbol{N_v^T} (x_\textrm{i} ) \Delta m_{v{\theta_x}}^h (x_i)\boldsymbol{N_{\theta_x}} (x_i)\\ \\ \end{aligned} \end{array} \tag{7.109} \]" src="form_932.png"/>
</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{array}{l} \begin{aligned} c_{vv}^h &amp;= \sum_{\textrm{i = 1} }^{\textrm{nsub}}\, \boldsymbol{N_v^T} (x_\textrm{i} ) \Delta b_{vv}^h (x_i)\boldsymbol{N_v} (x_i)\\ c_{ww}^h &amp;= \sum_{\textrm{i = 1} }^{\textrm{nsub}}\, \boldsymbol{N_w^T} (x_\textrm{i} ) \Delta b_{ww}^h (x_i)\boldsymbol{N_w} (x_i)\\ c_{{\theta_x}{\theta_x}}^h &amp;= \sum_{\textrm{i = 1} }^{\textrm{nsub}}\, \boldsymbol{N_{\theta_x}^T} (x_\textrm{i} ) \Delta b_{{\theta_x}{\theta_x}}^h (x_i)\boldsymbol{N_{\theta_x}} (x_i)\\ c_{v{\theta_x}}^h &amp;= \sum_{\textrm{i = 1} }^{\textrm{nsub}}\, \boldsymbol{N_v^T} (x_\textrm{i} ) \Delta b_{v{\theta_x}}^h (x_i)\boldsymbol{N_{\theta_x}} (x_i)\\ \end{aligned} \end{array} \tag{7.110} \]" src="form_933.png"/>
</p>
<p>where <img class="formulaInl" alt="$ x_i $" src="form_934.png"/> is taken at the buoyancy center of the sub-element. <img class="formulaInl" alt="$ \Delta m_{\alpha \gamma}^h $" src="form_935.png"/> and <img class="formulaInl" alt="$ \Delta b_{\alpha \gamma} $" src="form_936.png"/> are added mass and damping for the sub-element. These terms are calculated according to the applied hydrodynamic load model.</p>
<p>If the added mass and damping forces are established according to <a class="el" href="a00014.html#tm_eq_7_97">Eq. (7.97)</a> the added mass for the sub-element is computed by</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{array}{l} \begin{aligned} &amp;\Delta m_{vv}^h &amp;= A_{22}^{(2D)}\Delta x_0 \\ &amp;\Delta m_{ww}^h &amp;= A_{33}^{(2D)}\Delta x_0 \\ &amp;\Delta m_{\theta_x\theta_x}^h &amp;= A_{44}^{(2D)}\Delta x_0 \\ &amp;\Delta m_{v\theta_x}^h &amp;= A_{24}^{(2D)}\Delta x_0 \\ \end{aligned} \end{array} \tag{7.111} \]" src="form_937.png"/>
</p>
<p>and the damping computed by</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{array}{l} \begin{aligned} &amp;\Delta b_{vv}^h &amp;= B_{22}^{(2D)}\Delta x_0 \\ &amp;\Delta b_{ww}^h &amp;= B_{33}^{(2D)}\Delta x_0 \\ &amp;\Delta b_{\theta_x\theta_x}^h &amp;= B_{44}^{(2D)}\Delta x_0 \\ &amp;\Delta b_{v\theta_x}^h &amp;= B_{24}^{(2D)}\Delta x_0 \\ \end{aligned} \end{array} \tag{7.112} \]" src="form_938.png"/>
</p>
<p>where <img class="formulaInl" alt="$ A_{ij}^{(2D)} $" src="form_939.png"/> and <img class="formulaInl" alt="$ B_{ij}^{(2D)} $" src="form_940.png"/> are 2-dimensional added mass and damping per unit length dependent on the instantaneous immersion. <img class="formulaInl" alt="$ \Delta x_0 $" src="form_941.png"/> is the length of the sub-element in the initial condition.</p>
<p>If the simplified approach is applied for the hydrodynamic load calculations, <a class="el" href="a00014.html#tm_eq_7_102">Eq. (7.102)</a>, no hydrodynamic damping is considered. The added mass for the sub-element is established according to</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{array}{l} \begin{aligned} &amp;\Delta m_{vv}^h &amp;= \rho A_S C_{my}\Delta x_0 \\ &amp;\Delta m_{ww}^h &amp;= \rho A_S C_{mz}\Delta x_0 \\ \end{aligned} \end{array} \tag{7.113} \]" src="form_942.png"/>
</p>
<p>where <img class="formulaInl" alt="$ C_{my} $" src="form_912.png"/> and <img class="formulaInl" alt="$ C_{mz} $" src="form_913.png"/> are directional dependent added mass coefficients, and <img class="formulaInl" alt="$ A_S $" src="form_900.png"/> is submerged cross section area.</p>
<h3>External hydrodynamic load vector</h3>
<p>The hydrodynamic loads are computed along the secant length of the deformed element.</p>
<p>Considering the transverse load contributions the hydrodynamic load on each sub-element is written</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{f_n^h} = \boldsymbol{f_n^{FK}} + \boldsymbol{f_n^S} + \boldsymbol{f_n^D} \tag{7.114} \]" src="form_943.png"/>
</p>
<p>where</p>
<p><img class="formulaInl" alt="$ \boldsymbol{f_n^{FK}} $" src="form_944.png"/> : Froude-Krylov force including buoyancy, computed according to <a class="el" href="a00014.html#tm_eq_7_97">Eq. (7.97)</a>.</p>
<p><img class="formulaInl" alt="$ \boldsymbol{f_n^S} $" src="form_945.png"/> : diffraction force computed according to <a class="el" href="a00014.html#tm_eq_7_99">Eq. (7.99)</a> or <a class="el" href="a00014.html#tm_eq_7_101">Eq. (7.101)</a>.</p>
<p><img class="formulaInl" alt="$ \boldsymbol{f_n^D} $" src="form_946.png"/> : drag force computed according to <a class="el" href="a00014.html#tm_eq_7_106">Eq. (7.106)</a>.</p>
<p>The external moment around local x-axis is approximated according to</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{f_{\theta_x}^h} = \boldsymbol{r^B} \times \boldsymbol{f_n^h} \tag{7.115} \]" src="form_947.png"/>
</p>
<p>where <img class="formulaInl" alt="$ \boldsymbol{r^B} $" src="form_948.png"/> is the vector from the principal axis to the buoyancy center, see <a class="el" href="a00014.html#Element_divided_into_sub-elements_for_hydrodynamic_load_calculation">Element divided into sub-elements for hydrodynamic load calculation</a>. In the axial direction, <img class="formulaInl" alt="$ F_x^{FK} $" src="form_949.png"/> and <img class="formulaInl" alt="$ f_x^D $" src="form_950.png"/> are computed according to <a class="el" href="a00014.html#tm_eq_7_93">Eq. (7.93)</a> and <a class="el" href="a00014.html#tm_eq_7_104">Eq. (7.104)</a>, respectively.</p>
<p>The element end forces, <img class="formulaInl" alt="$ F_x^{FK} $" src="form_949.png"/> , Froude-Krylov forces including hydrostatic pressure, are assumed to be constant over the element length. Hence, the cross sectional axial force (stress resultant) is expressed in terms of effective axial force, see section <a class="el" href="a00014.html#tm_load_overview">Overview Of Load Effects</a>. The effect axial force is used directly in the expression for the geometric stiffness.</p>
<p>Based on these considerations, the components of the external hydrodynamic load vector is computed according to</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \begin{array}{l} \displaystyle \boldsymbol{S_u^h} = \frac{1}{2} F_ x^{FK} \begin{bmatrix} 1 \\ 1 \end{bmatrix} \sum_{\mathrm{i}=1}^{\textrm{nsub}}\, \boldsymbol{N_u^T} (x_\mathrm{i})f_x^D (x_\mathrm{i}) \Delta x_0\\ \\ \displaystyle \boldsymbol{S_v^h} = \sum_{\mathrm{i} = 1}^{\textrm{nsub}}\, \boldsymbol{N_v^T} (x_\mathrm{i}) f_y^h(x_\mathrm{i}) \Delta x_0\\ \\ \displaystyle \boldsymbol{S_w^h} = \sum_{\mathrm{i} = 1}^{\textrm{nsub}}\, \boldsymbol{N_w^T} (x_\mathrm{i}) f_z^h(x_\mathrm{i}) \Delta x_0\\ \\ \displaystyle \boldsymbol{S_{\theta_x}^h} = \sum_{\mathrm{i} = 1}^{\textrm{nsub}}\, \boldsymbol{N_{\theta_x}^T} (x_\mathrm{i}) f_{\theta_x}^ h(x_\mathrm{i}) \Delta x_0\\ \\ \displaystyle \boldsymbol{S_{\theta_y}^h} = 0 \\ \\ \displaystyle \boldsymbol{S_{\theta_z}^h} = 0 \end{array} \tag{7.116} \]" src="form_951.png"/>
</p>
<p>where <img class="formulaInl" alt="$ N $" src="form_434.png"/> denotes the actual displacement function. <img class="formulaInl" alt="$ N $" src="form_434.png"/> and <img class="formulaInl" alt="$ f $" src="form_331.png"/> are taken at the buoyancy center <img class="formulaInl" alt="$ x_i $" src="form_934.png"/> of each sub-element. <img class="formulaInl" alt="$ \Delta x_0 $" src="form_941.png"/> is the initial length of the sub-element.</p>
<h3>Hydrodynamic stiffness matrix</h3>
<p>The stiffness associated with water plane stiffness (buoyancy springs) is included in the stiffness matrix. The stiffness terms are limited to contributions due to translational displacement in global Z-direction and torsional rotation around local x-axis. For the computation of the hydrodynamic stiffness matrix, linear displacement functions are applied.</p>
<p>Thus, the element hydrodynamic stiffness matrix with regard to Z-displacement, <img class="formulaInl" alt="$ \boldsymbol{\tilde{k}_{ww}^{ E}} $" src="form_952.png"/> , is calculated directly in global system according to</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{\tilde{k}_{ww}^{ E}} = \sum_{\mathrm{i} = 1}^{\textrm{nsub}}\, \boldsymbol{N^T} (x_\mathrm{i} ) \frac{\partial\, \tilde{f}_Z^{ \mathrm{\,Buoy} }(x_\mathrm{i} )}{\partial Z} \boldsymbol{N} (x_\mathrm{i} ) \Delta x_0 \tag{7.117} \]" src="form_953.png"/>
</p>
<p>where <img class="formulaInl" alt="$ \frac{\partial\, \tilde{f}_Z^{~\textrm{Buoy} }}{\partial Z} $" src="form_954.png"/> is the water plane stiffness of the sub-element based on the water plane area in the global XY-plane</p>
<p>The local torsional hydrodynamic stiffness matrix is calculated according to</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{k_{\theta_x \theta_x}^E} = \sum_{\mathrm{i} = 1}^{\textrm{nsub}}\, \boldsymbol{N^T} (x_\mathrm{i} ) \left( \frac{\partial f ^\textrm{Buoy}_{\theta_x}(x_\mathrm{i} )}{\partial \theta_x} \right) \boldsymbol{N}(x_\mathrm{i} ) \Delta x_0 \tag{7.118} \]" src="form_955.png"/>
</p>
<p>Note that this formulation results in a symmetric hydrodynamic stiffness matrix.</p>
<h1><a class="anchor" id="aerodynamicLoads"></a>
Aerodynamic loads</h1>
<h2><a class="anchor" id="AeroElements"></a>
Lift forces, drag forces, and torsional moments on elements</h2>
<p>Airfoil sections which are not a part of a wind turbine may be subjected to aerodynamic loads based on interpolation from the static lift, drag, and moment curves, or Morison-type quadratic drag loads.</p>
<h3><a class="anchor" id="MorisonAeroDrag"></a>
Morison-type aerodynamic drag loads (CHTYPE=MORI)</h3>
<p>The drag load calculation for Morison-type aerodynamic drag is based on the relative wind velocity in the local system:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{u_r} = \boldsymbol{u_{wind}} - \boldsymbol{\dot{v}} = u_{rx}\boldsymbol{i_1} + u_{ry}\boldsymbol{i_2} + u_{rz}\boldsymbol{i_3} \tag{7.119} \]" src="form_956.png"/>
</p>
<p>where <img class="formulaInl" alt="$\boldsymbol{u_r}$" src="form_957.png"/> is the relative velocity at the element center, <img class="formulaInl" alt="$\boldsymbol{u_{wind}}$" src="form_958.png"/> is the undisturbed incoming wind velocity at the element center, and <img class="formulaInl" alt="$\boldsymbol{\dot{v}}$" src="form_959.png"/> is the structural velocity at the element center. In the local system, the x-axis is along the element.</p>
<p>The magnitude of the wind velocity normal to the elements is used in the calculation of the drag load:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ u_{rn} = \sqrt{u_{ry}^2 + u_{rz}^2} \tag{7.120} \]" src="form_960.png"/>
</p>
<p>Using the dimensional drag coefficients ( <img class="formulaInl" alt="$C_{Dx}$" src="form_961.png"/>, <img class="formulaInl" alt="$C_{Dy}$" src="form_962.png"/>, <img class="formulaInl" alt="$C_{Dz}$" src="form_963.png"/>), the aerodynamic load per unit length is calculated as:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ f = C_{Dx}u_{rx}|u_{rx}|\boldsymbol{i_1} + C_{Dy}u_{ry}u_{rn}\boldsymbol{i_2} + C_{Dz}u_{rz}u_{rn}\boldsymbol{i_3}. \tag{7.121} \]" src="form_964.png"/>
</p>
<p>The force per unit length is multiplied by the dry length (following the same water level integration as the wave loads) and applied in a lumped manner.</p>
<p>For an axisymmetric cross section, it is possible to give non-dimensional aerodynamic coefficients ( <img class="formulaInl" alt="$C_{dx}$" src="form_965.png"/>, <img class="formulaInl" alt="$C_{dy}$" src="form_966.png"/>, <img class="formulaInl" alt="$C_{dz}$" src="form_967.png"/>) as input rather than dimensional coefficients:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ C_{Dx} = \frac{1}{2}\rho_{air}S_{2D}C_{dx} \tag{7.122} \]" src="form_968.png"/>
</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ C_{Dy} = \frac{1}{2}\rho_{air}B_y C_{dx} \tag{7.123} \]" src="form_969.png"/>
</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ C_{Dz} = \frac{1}{2}\rho_{air}B_z C_{dx} \tag{7.124} \]" src="form_970.png"/>
</p>
<p>where <img class="formulaInl" alt="$S_{2D}$" src="form_971.png"/> is the cross-sectional dry surface, <img class="formulaInl" alt="$B_y$" src="form_972.png"/> is the projected area per unit length in the local <img class="formulaInl" alt="$y$" src="form_204.png"/> direction, and <img class="formulaInl" alt="$B_z$" src="form_973.png"/> is the projected area per unit length in the local <img class="formulaInl" alt="$z$" src="form_205.png"/> direction. For a circular cross section with diameter <img class="formulaInl" alt="$D$" src="form_974.png"/>,</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ S_{2D} = \pi D \tag{7.125} \]" src="form_975.png"/>
</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ B_y = B_z = D. \tag{7.126} \]" src="form_976.png"/>
</p>
<h3><a class="anchor" id="liftDragAirfoils"></a>
Lift forces, drag forces, and torsional moments on elements (CHTYPE=AIRF)</h3>
<p>For airfoil sections which are not a part of a wind turbine, aerodynamic loads are based on interpolation from the static lift, drag, and moment curves.The wind velocity and structural velocity are computed at the center of each airfoil element based on the undisturbed wind speed and structural velocity at the nodes. All computations are carried out in the local airfoil coordinate system. The lift force per unit length is denoted <img class="formulaInl" alt="$ F_L $" src="form_977.png"/>, the drag force per unit length is <img class="formulaInl" alt="$ F_D $" src="form_978.png"/>, and the aerodynamic moment per unit length is <img class="formulaInl" alt="$ M $" src="form_702.png"/>.</p>
<p><a class="anchor" id="Local_airfoil_coordinate_system"></a></p>
<div class="image">
<img src="airfoil_coordinate_system.svg" alt="airfoil_coordinate_system.svg"/>
<div class="caption">
Local airfoil coordinate system</div></div>
<p>The relative velocity in the airfoil coordinate system is computed as:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{V_{r}} = \boldsymbol{V_{\mathrm{wind} }} - \boldsymbol{V_{\mathrm{struc} }} \tag{7.127} \]" src="form_979.png"/>
</p>
<p>The spanwise ( <img class="formulaInl" alt="$ V_{r_{z}} $" src="form_980.png"/>) component of the velocity is assumed to be small, and an error is returned if the spanwise component is large compared to the velocity magnitude. The angle of attack ( <img class="formulaInl" alt="$ \alpha $" src="form_981.png"/>) is computed based on the local <img class="formulaInl" alt="$ V_{r_{x}} $" src="form_982.png"/> and <img class="formulaInl" alt="$ V_{r_{y}} $" src="form_983.png"/> components of the velocity. The Reynolds number (Re) is computed as</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \text{Re} = \frac{\rho_{\mathrm{air} } c \sqrt{\left(V_{r_{x}}^2 + V_{r_{y}}^2 \right)} }{\nu_{\mathrm{air} }} \tag{7.128} \]" src="form_984.png"/>
</p>
<p>where <img class="formulaInl" alt="$ \rho_{air} $" src="form_985.png"/> is the air density, <img class="formulaInl" alt="$ c $" src="form_986.png"/> is the chord length, and <img class="formulaInl" alt="$ \nu_{air} $" src="form_987.png"/> is the viscosity of air.</p>
<p>Lift ( <img class="formulaInl" alt="$ C_L $" src="form_988.png"/>), drag ( <img class="formulaInl" alt="$ C_D $" src="form_989.png"/>), and moment ( <img class="formulaInl" alt="$ C_M $" src="form_990.png"/>) coefficients are then obtained by interpolation in the airfoil tables based on Re and <img class="formulaInl" alt="$ \alpha $" src="form_981.png"/>. The forces and moment (about the airfoil coordinate system) per unit length are:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ F_L = \frac{1}{2} C_L \rho_{\mathrm{air} } c \left(V_{r_{x}}^2 + V_{r_{y}}^2 \right) \tag{7.129} \]" src="form_991.png"/>
</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ F_D = \frac{1}{2} C_D \rho_{\mathrm{air} } c \left(V_{r_{x}}^2 + V_{r_{y}}^2 \right) \tag{7.130} \]" src="form_992.png"/>
</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ M = \frac{1}{2} C_M \rho_{\mathrm{air} } c^2 \left(V_{r_{x}}^2 + V_{r_{y}}^2 \right) \tag{7.130} \]" src="form_993.png"/>
</p>
<p>Then, <img class="formulaInl" alt="$ F_L $" src="form_977.png"/>, <img class="formulaInl" alt="$ F_D $" src="form_978.png"/>, and <img class="formulaInl" alt="$ M $" src="form_702.png"/> are multiplied by the dry length of the element, transformed into the appropriate coordinate system, and are applied to the nodes in a lumped manner. The dry length of the element is determined consistent with the wave load integration definition.</p>
<h2><a class="anchor" id="BEM"></a>
Blade element/momentum (BEM) theory for wind turbines</h2>
<p>The aerodynamic load model used for wind turbines in RIFLEX and SIMO is based on the blade element momentum (BEM) theory. This theory combines momentum theory and blade element theory. The general principle of the theory is that forces developed locally at the airfoil, based upon empirical lift and drag coefficients, are balanced with the change in momentum of the air flowing through the rotor disk.</p>
<h3><a class="anchor" id="outlineOfBEM"></a>
Outline of BEM calculation</h3>
<p>The balance of airfoil forces with the change in momentum of the air is the basic step in the calculation. In the implemented calculation, the dynamic wake method is used. Only one solution sequence is performed for each time step, and time is used a relaxation parameter. In a quasi-static analysis, an iterative calculation would be performed for each time step.</p>
<p>An outline of the basic force balance is given below. Each wind turbine blade is divided into blade elements. For each blade element, aerodynamic airfoil coefficients (lift, drag and moment) are defined.</p>
<ol type="1">
<li>The calculation begins with a suitable guess for the induced velocity at each blade element. Induced velocity is defined as the difference between the velocity of air at the rotor plane and the velocity of air far upstream of the rotor plane.</li>
<li>The relative air velocity with respect to each blade element is calculated.</li>
<li>The angle-of-attack with respect to the airfoil is calculated.</li>
<li>Using airfoil coefficient data, the aerodynamic lift force, drag force, and moment are calculated using the Øye dynamic stall model.</li>
<li>Airfoil forces are rotated into the rotor plane coordinate system.</li>
<li>Annulus-average forces and velocities are computed, for use in the momentum balance. Rotor plane-average remote and induced velocities are computed, for use in computing the Prandtl factor.</li>
<li>The Prandtl factor is calculated. This factor modifies the basic momentum balance to account for a finite number of blades.</li>
<li>Induced velocity (the average over each annulus) is updated based upon momentum balance, using annulus-average forces.</li>
<li>In cases where the rotor is yawed with respect to the remote wind, the wake skew angle is computed, and the variation of induced velocity with blade azimuth angle is computed. Based upon the newly calculated induced velocity, a guess can be made to feed the next iteration.</li>
<li>The calculation is so far quasi-static. The final step is the Øye dynamic wake model, which acts as a filter for the induced velocity. The inputs are the quasi static induced velocity and the induced velocity from the last time step.</li>
</ol>
<h3><a class="anchor" id="BEMCorrectionFactors"></a>
Correction factors</h3>
<p>A number of correction factors are applied in the BEM method.</p>
<h4>Glauert correction</h4>
<p>BEM theory is not valid for induction factors greater than 0.4. The Glauert correction is used for large induction factors: for <img class="formulaInl" alt="$ a>0.4$" src="form_994.png"/>, the empirical thrust curve recommended by Burton et al (2011, p67) is used:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ a = \frac{(C_T/F - C_{T1})}{C_{T2}-C_{T1}}\left(a_2-a_1 \right) + a_1 \tag{7.132} \]" src="form_995.png"/>
</p>
<p>where <img class="formulaInl" alt="$ a_2=1.0 $" src="form_996.png"/>, <img class="formulaInl" alt="$ C_{T2}=1.82 $" src="form_997.png"/>, <img class="formulaInl" alt="$ a_1=1.0-0.5\sqrt{C_{T2}} $" src="form_998.png"/>, <img class="formulaInl" alt="$ C_{T1}=4a_1(1-a_1) $" src="form_999.png"/>, and <img class="formulaInl" alt="$ F$" src="form_1000.png"/> is the Prandtl factor <a class="el" href="a00014.html#tm_eq_7_133">Eq. (7.133)</a>. For <img class="formulaInl" alt="$ F=1 $" src="form_1001.png"/>, the thrust curve used in the calculations is compared to the momentum theory result shown in the figure <a class="el" href="a00014.html#Glauert_correction">Glauert correction</a>.</p>
<p><a class="anchor" id="Glauert_correction"></a></p>
<div class="image">
<img src="glauert_correction.svg" alt="glauert_correction.svg"/>
<div class="caption">
Glauert correction</div></div>
<h4>Prandtl factor</h4>
<p>To account for the tip and hub loss on a blade due to a finite number of blades, the Prandtl factor is implemented. The Prandtl factor <img class="formulaInl" alt="$ F $" src="form_1002.png"/> as a function of the radial location <img class="formulaInl" alt="$ r $" src="form_685.png"/> is computed as:</p>
<p><a class="anchor" id="tm_eq_7_133"></a> </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ F = \frac{2}{\pi}\cos^{-1}{\left(e^{-f} \right)} \tag{7.133} \]" src="form_1003.png"/>
</p>
<p>where</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ f = \frac{B}{2}\frac{R-r}{2r\sin{\phi}} \tag{7.134} \]" src="form_1004.png"/>
</p>
<p><img class="formulaInl" alt="$ B $" src="form_1005.png"/> is the number of blades, <img class="formulaInl" alt="$ R $" src="form_1006.png"/> is the outer radius of the rotor, and <img class="formulaInl" alt="$\phi$" src="form_1007.png"/> is the angle that the trailing vorticies make with the rotorplane. <img class="formulaInl" alt="$ \phi $" src="form_634.png"/> is assumed to be the same as the angle of incoming flow, neglecting tangential induced velocity. The default program behavior is to correct <img class="formulaInl" alt="$ \phi $" src="form_634.png"/> for yawed inflow.</p>
<h4><a class="anchor" id="DynamicWake"></a>
Dynamic wake</h4>
<p>The dynamic wake effect is the time lag in induced velocities due to the shedding and downstream convection of vorticity. Dynamic wake effects are most pronounced for heavily loaded rotors, corresponding to high induction factors (low wind speeds). This effect can be modeled by the Stig Øye dynamic inflow model, which acts as a filter for induced velocities, as shown below (Snel and Schepers, 1995). <img class="formulaInl" alt="$ \boldsymbol{W_{qs}} $" src="form_1008.png"/> is the quasi static induced velocity vector, and <img class="formulaInl" alt="$ \boldsymbol{W} $" src="form_282.png"/> is the new induced velocity vector. <img class="formulaInl" alt="$ \tau_1 $" src="form_1009.png"/> and <img class="formulaInl" alt="$ \tau_2 $" src="form_1010.png"/> are time constants.</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{W} + \tau_2 \frac{d\boldsymbol{W}}{dt} = \boldsymbol{W}_{\mathrm{int} } \tag{7.135} \]" src="form_1011.png"/>
</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \boldsymbol{W}_{\mathrm{int} } + \tau_1 \frac{d\boldsymbol{W}_{\mathrm{int} }}{dt} = \boldsymbol{W}_{qs} + 0.6 \tau_1 \frac{d\boldsymbol{W}_{qs}}{dt} \tag{7.136} \]" src="form_1012.png"/>
</p>
<h4><a class="anchor" id="DynamicStall"></a>
Dynamic stall</h4>
<p>Dynamic stall refers to time-dependent variations in the lift and drag coefficients of an airfoil due to changes in the angle of attack. The Stig Øye model is implemented and gives unsteady lift by filtering the trailing edge separation point with an empirical time constant (see also Hansen, 2008).</p>
<p>The degree of stall ( <img class="formulaInl" alt="$ f_s $" src="form_1013.png"/>) affects the lift as</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ C_L = f_s C_{L,\mathrm{inv} }(\alpha) + (1-f_s)C_{L,fs}(\alpha) \tag{7.137} \]" src="form_1014.png"/>
</p>
<p>where <img class="formulaInl" alt="$ C_{L,\mathrm{inv} } $" src="form_1015.png"/> is the lift coefficient without separation, and <img class="formulaInl" alt="$ C_{L,fs} $" src="form_1016.png"/> is the fully separated lift coefficient. A static value <img class="formulaInl" alt="$ f_{s}^{st} $" src="form_1017.png"/> can be found to reproduce the static airfoil data. The degree of stall is assumed to chase the static value:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ \frac{df_s}{dt} = \frac{f_s^{st}-f_s}{\tau} \tag{7.138} \]" src="form_1018.png"/>
</p>
<p>where <img class="formulaInl" alt="$ \tau $" src="form_1019.png"/> is a time constant. By integration:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ f_s(t+\Delta t) = f_s^{st} + (f_s(t)-f_s^{st})e^{-\Delta t/\tau}. \tag{7.139} \]" src="form_1020.png"/>
</p>
<p>For flow that is not fully stalled, the lift coefficient is computed as: </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ C_L = \frac{1}{4}\frac{dC_L}{d\alpha} (\alpha-\alpha_0)\left(1+\sqrt{1-|f_s|} \right)^2 \tag{7.140} \]" src="form_1021.png"/>
</p>
<p>where <img class="formulaInl" alt="$ \frac{dC_L}{d\alpha} $" src="form_1022.png"/> is computed at the full-stall limit (depending on the sign of <img class="formulaInl" alt="$ \alpha $" src="form_981.png"/>), and <img class="formulaInl" alt="$ \alpha_0 $" src="form_1023.png"/> is the angle of attack where the lift is zero.</p>
<p>For fully stalled flow: </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ C_L = C_{L,qs}\left(1+\sqrt{1-|f_s|} \right)^2 \tag{7.141} \]" src="form_1024.png"/>
</p>
<p>where <img class="formulaInl" alt="$ C_{L,qs} $" src="form_1025.png"/> is the quasi-static lift.</p>
<p>In order to initialize the dynamic stall calculation, the program identifies several important points in the lift curve: the angle of attack where there is zero lift ( <img class="formulaInl" alt="$ \alpha_0 $" src="form_1023.png"/>), the maximum slope of the lift curve in the linear region for both positive and negative lift values ( <img class="formulaInl" alt="$ \left[ dC_L/ d\alpha \right] _{\mathrm{max} }(1) $" src="form_1026.png"/> and <img class="formulaInl" alt="$ \left[ dC_L/ d\alpha \right] _{\mathrm{max} }(2) $" src="form_1027.png"/>), and the angle of attack corresponding to full separation for both positive and negative lift values ( <img class="formulaInl" alt="$ \alpha_{fs}(1) $" src="form_1028.png"/> and ( <img class="formulaInl" alt="$ \alpha_{fs}(2) $" src="form_1029.png"/>)) For typical airfoils, these points are relatively easy to identify. For more complex airfoils, the user may input these values directly.</p>
<div class="image">
<img src="dynstall_param.svg" alt="dynstall_param.svg"/>
<div class="caption">
Illustration of dynamic stall initialization parameters.</div></div>
<h4><a class="anchor" id="SkewedInflow"></a>
Skewed inflow</h4>
<p>Skewed inflow may occur due to rotor tilt or a yaw angle between the rotor and the oncoming wind. Glauert developed the basic formulation for correcting the induction factor due to skewed inflow: <img class="formulaInl" alt="$ a $" src="form_716.png"/> is the induction factor, <img class="formulaInl" alt="$ \chi $" src="form_1030.png"/> is the wake skew angle, and <img class="formulaInl" alt="$ \Psi $" src="form_1031.png"/> is the azimuth angle that is zero at the most downwind position of the rotor.</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ a_{\mathrm{skew} } = a\left[1+ \tan{\left(\frac{\chi}{2} \right)}\left(\frac{r}{R}\right)\cos{\left(\Psi \right)}\right] \tag{7.142} \]" src="form_1032.png"/>
</p>
<h4><a class="anchor" id="TowerInfluence"></a>
Upwind tower influence</h4>
<p>The tower influence (shadow) effect is implemented using potential flow theory. Since the incoming wind has to travel around the tower, the tower has an effect on the local inflow, even for an upwind turbine. The velocity deficit due to the tower is based on the 2D potential solution for constant flow around a circle, with the possibility to include corrections for the tower drag and the Bak coefficient. The coordinate system for the tower influence calculation is shown in figure <a class="el" href="a00014.html#Tower_influence_coordinate_system">Tower influence coordinate system</a>.</p>
<p><a class="anchor" id="Tower_influence_coordinate_system"></a></p>
<div class="image">
<img src="towShad1.svg" alt="towShad1.svg"/>
<div class="caption">
Tower influence coordinate system</div></div>
<p>For an upwind wind turbine, the non-dimensional influence at a location <img class="formulaInl" alt="$(x,y,z)$" src="form_1033.png"/> is evaluated as:</p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ x_w = \frac{2x}{D_{\mathrm{tow}}} \tag{7.143}\]" src="form_1034.png"/>
</p>
 <p class="formulaDsp">
<img class="formulaDsp" alt="\[y_w = \frac{2y}{D_{\mathrm{tow}}} \tag{7.144} \]" src="form_1035.png"/>
</p>
 <p class="formulaDsp">
<img class="formulaDsp" alt="\[x_{\mathrm{infl}} = \left[ 1 - \frac{(x_w+b)^2-y_w^2}{\left((x_w+b)^2 + y_w^2\right)^2} + \frac{C_D}{2\pi} \frac{x_w+b}{(x_w+b)^2 + {y_w}^2}\right] \tag{7.145}\]" src="form_1036.png"/>
</p>
 <p class="formulaDsp">
<img class="formulaDsp" alt="\[y_{\mathrm{infl}} = 2\left[ \frac{(x_w+b) y_w}{\left((x_w+b)^2 + y_w^2\right)^2} + \frac{C_D}{2\pi} \frac{y_w}{(x_w+b)^2 + {y_w}^2} \right] \tag{7.146} \]" src="form_1037.png"/>
</p>
<p>where <img class="formulaInl" alt="$D_{\mathrm{tow}}$" src="form_1038.png"/> is the local tower diameter for the given <img class="formulaInl" alt="$z$" src="form_205.png"/> level, <img class="formulaInl" alt="$b$" src="form_1039.png"/> is the Bak coefficient, and <img class="formulaInl" alt="$C_D$" src="form_1040.png"/> is the local tower drag coefficient for the given <img class="formulaInl" alt="$z$" src="form_205.png"/> level.</p>
<p>In practice, the horizontal wind direction and speed are computed at the location <img class="formulaInl" alt="$(x,y,z)$" src="form_1033.png"/>. This speed is multiplied by the factors <img class="formulaInl" alt="$x_{\mathrm{infl} }$" src="form_1041.png"/> and <img class="formulaInl" alt="$y_{\mathrm{infl} }$" src="form_1042.png"/>, and the modified speed is then transposed back to the initial wind direction (Moriarty and Hansen, 2005).</p>
<p>Note that the <img class="formulaInl" alt="$x$" src="form_1043.png"/>-coordinate here is upwind and the velocity is set to zero for any point inside the tower. The tower is assumed to be approximately vertical and no effect on the vertical wind speed is included.</p>
<p>The resulting velocity deficit is shown in figure <a class="el" href="a00014.html#Velocity_deficit_due_to_the_tower">Velocity deficit due to the tower</a>.</p>
<p><a class="anchor" id="Velocity_deficit_due_to_the_tower"></a></p>
<div class="image">
<img src="towerInfluence.svg" alt="towerInfluence.svg"/>
<div class="caption">
Velocity deficit due to the tower</div></div>
<h4><a class="anchor" id="DWTowerInfluence"></a>
Downwind tower influence</h4>
<p>In order to model downwind wind turbines, a different method for computing the velocity deficit due to the presence of the tower is needed. The most common models, which are implemented in AeroDyn and Bladed, for example, are based on work by Powles. Powles' model, as implemented in RIFLEX, assumes a cosine-squared shape for the tower shadow.</p>
<p>The cosine-squared model for the wake factor <img class="formulaInl" alt="$u_{\mathrm{wake}}$" src="form_1044.png"/> is: </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ u_{\mathrm{wake}} = \begin{cases} \displaystyle\frac{C_D}{\sqrt{d}}\cos{\left( \frac{\pi y}{2 d} \right)}, &amp; |y|<\sqrt{d} \\ 0, &amp; |y|\geq\sqrt{d} \end{cases} \tag{7.147} \]" src="form_1045.png"/>
</p>
<p> where <img class="formulaInl" alt="$ d = \sqrt{x^2+y^2} $" src="form_1046.png"/>. Together with the potential flow model, the local wind velocity is found as: </p>
<p class="formulaDsp">
<img class="formulaDsp" alt="\[ U_{\mathrm{local}} = (u-u_{\mathrm{wake}})U_\infty \tag{7.148} \]" src="form_1047.png"/>
</p>
 <p class="formulaDsp">
<img class="formulaDsp" alt="\[ V_{\mathrm{local}} = (v-u_{\mathrm{wake}})U_\infty \tag{7.149} \]" src="form_1048.png"/>
</p>
<p>where <img class="formulaInl" alt="$U_\infty$" src="form_1049.png"/> is the total instantaneous horizontal velocity, and <img class="formulaInl" alt="$u=1$" src="form_1050.png"/> when the cosine-squared model is in effect. In Powles' model, the flow is assumed to not reverse: for <img class="formulaInl" alt="$u-u_{\mathrm{wake}}<0$" src="form_1051.png"/>, we assume <img class="formulaInl" alt="$U_{\mathrm{local}}=0$" src="form_1052.png"/>. The computed <img class="formulaInl" alt="$U_{\mathrm{local}}$" src="form_1053.png"/> and <img class="formulaInl" alt="$V_{\mathrm{local}}$" src="form_1054.png"/> are then transformed back to the local global reference frame. Thus, the wake is assumed to align with the instantaneous horizontal wind vector.</p>
<p>The resulting velocity deficit for a downwind turbine is shown in figure <a class="el" href="a00014.html#Velocity_deficit_due_to_the_tower_downwind">Velocity deficit due to the tower, downwind wind turbine</a>.</p>
<p><a class="anchor" id="Velocity_deficit_due_to_the_tower_downwind"></a></p>
<div class="image">
<img src="towerInfluenceDownwind.svg" alt="towerInfluenceDownwind.svg"/>
<div class="caption">
Velocity deficit due to the tower, downwind wind turbine</div></div>
<h3><a class="anchor" id="BEMLimitations"></a>
BEM Limitations</h3>
<p>The BEM aerodynamic load module has the following limitations:</p>
<ol type="1">
<li>The code does not truly account for blade cone or large deflections</li>
<li>The vortical wake structure must be preserved. For this to be the case, the ratio of blade tip speed to wind speed must not be too low, and the majority of the blade must not be stalled.</li>
<li>The cone angle should be small. If the rotor is coned: for purposes of load calculation, define the aerodynamic analysis as if the cone angle were zero; for a lower bound on power output, define the aerodynamic analysis with the actual cone angle.</li>
</ol>
<h1><a class="anchor" id="tm_load_references"></a>
References</h1>
<p>Burton, T., Jenkins, N., Sharpe, D. and Bossanyi, E. (2011): Wind Energy Handbook (Wiley)</p>
<p>Dalzell, J.F. (1999): A note on finite depth second order wave-wave interactions. Applied Ocean Research <b>21</b> 105.</p>
<p>Dean, R. G. &amp; Dalrymple, R. A. (1991): Water Wave Mechanics for Engineers and Scientists (World Scientific)</p>
<p>Dixon, A.G., Greasted, C.A. and Salter, S.H. (1979): Wave Forces on Partially Submerged Cylinders ASCE, Journal of the Waterway, Port, Coastal and Ocean Division, Nov.</p>
<p>Engseth, A., Bech, A. and Larsen, C.M. (1988): Efficient Method for Analysis of Flexible Risers. Proc. of the Int. Conf. on Behaviour of Offshore Structures (BOSS), Vol. 3, pp. 1357-1371, Tapir, Trondheim.</p>
<p>Faltinsen, O.M. (1977): Water impact loads and dynamic response of horizontal cylinders in offshore structures. OTC-paper No. 2741, Texas.</p>
<p>Forristall, G.Z. (2000): Wave Crest Distributions: Observations and Second-Order Theory. Journal of Physical Oceanography <b>30</b> 1931.</p>
<p>Frank, W. (1967): Oscillation of Cylinders in or below the Free Surface of Deep Fluids Report 2375. Washington DC: Naval Ship Research and Development Center.</p>
<p>Fylling, I.J., Sødahl, N. and Bech, A. (1986): On the Effects of Slug Flow on the Dynamic Response of Flexible Risers and Submerged Hoses. IBC Technical Services, London.</p>
<p>Greenhow, M. (1987): Water -entry and -exit of horizontal circular cylinder. Technical Report, Marintek, Trondheim.</p>
<p>Hansen, M.O.L. (2008): Aerodynamics of Wind Turbines. 2nd Edition.</p>
<p>Hoerner, S.F. (1965): Fluid-dynamic Drag. Published by the author, Washington D.C.</p>
<p>Hooft, J.P. (1972): Hydrodynamic Aspects of Semi-submersible platform in waves. H. Veeman on Zonen N.V., Wagningen.</p>
<p>Kaplan, P. and Hue, P.N. (1959): Virtual mass and slender body theory for bodies in waves. Proc. Sixth An. Conference on Fluid Mechanics, University of Texas, Texas.</p>
<p>Kaplan, P. and Silbert, M.N. (1976): Impact forces on platform horizontal members in the splash zone. OTC-paper No. 2498, Texas.</p>
<p>MacCamy, R. C. &amp; Fuchs, R. A. (1954): Wave Diffraction on Piles: A Diffraction Theory. Beach Erosion Board; Corps of Engineers.</p>
<p>Marthinsen, T. and Winterstein, S.R. (1992): On the skewness of random surface waves. The Second International Offshore and Polar Engineering Conference, International Society of Offshore and Polar Engineers.</p>
<p>Mathisen, K.M. (1990): Large Displacement Analysis of Flexible and Rigid Systems considering Displacement-Dependent Loads and Nonlinear Constraints. Doctoral Thesis, Div. of Structural Mechanics, The Norwegian Institute of Technology, Trondheim.</p>
<p>Mathisen, K.M. and Bergan, P.G. (1986): Nonlinear static analysis of flexible risers. The International Symposium &amp; Exhibition, OMAE 1986, Tokyo.</p>
<p>Mei, C.C. (1989): The Applied Dynamics of Ocean Surface Waves. World Scientific.</p>
<p>Morgan, G.W. and Peret, J.W. (1975): Applied Mechanics of Marine Riser Systems. Petroleum Engineer, Oct. 1974 - Oct. 1975.</p>
<p>Moriarty, P.J. and Hansen, A.C. (2005): AeroDyn Theory Manual. NREL/TP-500-36881</p>
<p>Ormberg, H. (1991): Non-Linear Response Analysis of Floating Fish Farm Systems. Doctoral Thesis, Div. of Marine Structures, The Norwegian Institute of Technology, Trondheim.</p>
<p>Powles, S.R.J. (1983): The effects of tower shadow on the dynamics of a HAWT. Wind Engineering, 7:26–42, 1983.</p>
<p>Sharma, J. and Dean, R.G. (1981): Second-Order Directional Seas and Associated Wave Forces. Publication of: Offshore Technology Conference.</p>
<p>Skjelbreia, L. and Hendrickson, J. (1960): Fifth order gravity wave theory. Proceedings of 7th Conference on Coastal Engineering, Vol. 1, The Netherlands.</p>
<p>Snel H., and Schepers J.G. (1995): Joint Investigation of Dynamic Inflow Effects and Implementation of an Engineering Method. Report ECN-C–94-107, Energy Research Centre of the Netherlands.</p>
<p>Sparks, C.P. (1984): The influence of tension, pressure and weight on pipe and riser deformations and stresses. Journal of Energy Resources Technology, Vol. 106, pp. 46-54.</p>
<p>Stansberg, C.T., Gudmunstad, O.T. and Haver, S.K. (2008): Kinematics under extreme waves. Journal of Offshore Mechanics and Arctic Engineering <b>130</b> 021010.</p>
<p>Øritsland O. (1986): Wave forces on a sphere in the splash zone. Technical report, MARINTEK, Trondheim.</p>
<p>Aarsnes, J.V., Lien E. and Oltedal, G. (1988): SIFICA - Theory Manual MARINTEK report: 513004.05.02, Trondheim. </p>
</div></div><!-- contents -->
	</body>
</html>