. (A.32)
Hence, for α1 = 0, λi is proportional to ωi and for α2 = 0, λi is inverted pro-portional to ωi. If we know the overall damping level for two frequencies we can determine α1 and α2 by
α1 = 2ω1ω2 ω22− ω12
(λ1ω2− λ2ω1) , (A.33a) α2 = 2 (λ2ω2− λ1ω1)
ω22− ω12
. (A.33b)
The inertia term leads to damping from rigid body motions in the same way as these motions give inertia forces. This damping contribution is not wanted since structure damping is mainly caused by strains in the material of the structure.
Hence, the inertia term in Rayleigh damping should not be included. The stiffness term α2K is directly related to strains in the structure and can model interior structural damping adequately. The damping ratio will, however, increase with increasing frequency. The damping level, λi, must therefore be correct at the actual response frequency. The influence from damping at higher frequencies is without significance for the solution, and may in fact contribute to a more smooth time history since false oscillations caused by the numerical method might be suppressed.
A.3 Load on the Risers
The risers will experience load in the global horizontal direction due to currents.
The current loads will induce forces on the riser, which we will find by using Morison’s equation. The algorithm for loads, forces and displacements for the static initialization procedure could be summarized as follows:
1. Decompose the current from the global frame to the local frame.
2. Calculate the load at each node for each element locally.
3. Calculate the forces at each node for each element locally.
4. Transform the forces back to the global system.
5. Add the forces for each node to find the total force in the global frame.
6. Introduce load terms from for the specified motions from the TLP.
7. Find the displacement vector by r = K−1R.
Below each point is explained in detail.
Decompose the current from the global frame to the local frame. First the current, given in the global frame, has to be decomposed into the local frame of the node at which the current attacks. It is assumed that the global current velocity vector in each node i, vcur,if , only has a horizontal component, such that the vzf-component is zero. The flow is linear, i.e. there are no vortices in the flow.
The decomposed current in i-frame is then
vfcur,i =
Calculate the load at each node for each element locally. The loads in each element’s two nodes are found by Morison’s equation for the axial and transverse direction
where fext,xi is the external force in xi-direction for element i. When a force, stiffness matrix etc. is described in the i-frame, it is implicitly for element i only.
ρw is the density of the displaced fluid, Di the diameter of element i, and CDn and CDt the drag force coefficients in normal and tangential direction. vxi and vzi are the current velocities in the local i-frame. The load in each direction is given by
This gives the load in each of the two nodes of an element to be qxi,1 = 1
Calculate the forces at each node for each element locally. Assuming linearly varying transverse and axial load, the load on a bar is found in Fylling et al. (2005) and is given as
where fni is the normal and fti the tangential forces. The sum of forces acting on each local element in its frame is then given by
fi = fti+ fni. (A.48)
This could be simplified to include forces in the xi-direction only since these are far larger than the forces in the zi-direction. The forces in zi-directions will not have major impact on the dynamics, hence they could be neglected. The resulting
external force in each element is then given in the same direction as the load as
Transform the forces back to the global system. Before adding the forces acting at each node, they have to be represented in the same frame. The forces are transformed from its local frame to the global frame the same way as the displacements
fif=Tfifi, (A.50)
where fif is the external force in the two nodes in element i given in the global f -frame.
Add the forces for each node to find the total force in the global frame.
After the forces at all nodes are transformed back to the global coordinate system, the forces and moments are summarized in each node
fx,if = fxf1,i+ fxf2,i−1, (A.51) fz,if = fz1,if + fz2,i−1f . (A.52) After summarizing the forces at each node, the total force vector in f -frame due to the current forces is called fdrag.
Introduce load terms for the specified motions from the TLP. The specified motions for the quasi-static case are accounted for by correction terms in the load vector. These are found from
fcorr= fdrag− kspexT LP, (A.53) where fcorr is the corrected force vector, kspe is the specified stiffness vector and xT LP is the scalar TLP motion in surge. fcorr is a vector due to the two-dimensional model, and kspe is a vector due to the specified motion at the top node in surge only. All vectors are described in the global f -frame.
Find the displacement vector by r = K−1R. The displacements of the riser for the quasi-static case is then found to be
r = K−1R, (A.54)
where R is the force vector. Upper case is used here since this is most com-mon within the field of structural mechanics for load and force vectors. See the following iteration algorithm for equilibrium between internal and external forces.