Page 301 - Practical Design Ships and Floating Structures
P. 301

216

             substantial restrictions  on the geometries handled,  lack of computing power and unreliable analysis
             tools. A short overview of previous work on automated optimization of offshore structures is given by
             Birk (1998).
             Today,  automated  optimization  is mature  enough to be  a valuable  tool  in designing  better floating
             systems (Birk and Harries, 2000). This approach shifts the main focus from repeated interactive work
             with  CAD  and  CFD  tools  to  the  more  important  definition  of  objective  functions,  parametric
             description of hull  properties and assessment of final results. The following section concentrates on
             rational  design  criteria  suitable  to  compare  motion  behavior  of  different  designs.  For  a  detailed
             description of the optimization  framework used in this research work the interested reader is referred
             to Birk (1998). Only a short review is presented below.

             2  RATIONAL DESIGN CRITERIA

             Automated assessment  of design variants relies on rational criteria,  i.e.  an objective  function which
              provides the scale necessary to compare different designs. Appropriate models of system behavior and
              environmental conditions have to be set up.
             2.1 ModelIing System Behaviour

              The  performance  assessment  of  each  design  variant  starts  with  computation  of  selected  response
              amplitude operators of forces and motions. The application presented below utilizes results obtained
              by the well established 3D-diraction-radiation software package WAMIT' (Newman and Sclavounos,
              1988). The program is based on linear potential theory and neglects viscous effects. Although higher
              order quantities and viscous effects may be important in special cases linear analysis based on potential
              theory will be generally sufficient when comparing different designs. The complex ratio of output and
              input  signals  s(w)  and  <(w)  for  each  wave  frequency  w  constitutes  the  transfer  function  or
              response  amplitude  operator  (RAO)  HSc(m)=- 40)  which  gives  a  complete  description  of  the
                                                 &w)
              corresponding hydrodynamic characteristics. Values of  HsC (0) are computed and stored for a larger
              number of wave frequencies and an appropriate number of wave headings.
              In many cases motions of selected points  5, are of special interest. Within linear theory they are easily
              deduced from the translational and rotational motion vectors of the centre of gravity  sT and  gI), i.e.
                        .
              s,=sI.+es,) The vector r describes the location of the reference point with respect to the centre of
              gravity.  Since  all  quantities  are  harmonic  with  wave  frequency  o amplitudes  of  velocities  and
              accelerations are computed by simple multiplication of  E, by  w  and  - o2 respectively.

              2.2 Modelling environmental conditions

              The hydrodynamic analysis yields transfer functions, which characterize the behavior in regular waves.
              In  reality,  the  elevation  of  the  ocean  surface  is  irregular  and  of  random  nature.  Hence,  rational
              seakeeping  criteria have  to  be  based  on  a probabilistic  description  of  random  seas.  For  intervals
              ranging from one to three hours the statistical parameters of irregular seas do not vary much, i.e. the
              process  is  stationary  (Barltrop  and  Adams,  1991).  Gaussian  distribution  of  wave  elevations  and
              Rayleigh distribution of wave heights are assumed. The description of these short- term sea states is
              commonly based on design spectra, representing the frequency dependent energy distribution of waves.
              The relevant parameters  are significant wave height H, and mean zero-up- crossing period  To. The
              significant wave height is linked to the variance  u'  of the random process by  H: = 160'  (Newland,
   296   297   298   299   300   301   302   303   304   305   306