Sea Technology

NOV 2013

The industry's recognized authority for design, engineering and application of equipment and services in the global ocean community

Issue link:

Contents of this Issue


Page 14 of 75

behaviors. In the past decades, the commercial software tools based on a single method have revolutionized the entire engineering design stream and have proven to be very successful. They have been widely applied in nearly every electromagnetic-related initial engineering design. Popular simulation techniques, such as simulation programs with integrated circuit emphasis (SPICE), fnite difference timedomain (FDTD) method, fnite element method (FEM) and discontinuous Galerkin (DG) method, have received a lot of attention in both academic research and industry development. Beyond State of the Art Each method has supported the business of several successful engineering simulation software companies. However, as engineering simulation techniques develop, each will have its own limits, advantages and disadvantages. No one could handle all cases in all situations. For example, FDTD is simple to develop and is very effcient. However, its stair-cased grid and central difference approximation lead to a large dispersion error for an electrically large problem. FEM is robust in modeling the geometries and performing hp-refnements, hence the accuracy of FEM is highly controllable. However, FEM requires huge memories to store its system matrices and is very slow to obtain the solution when compared with FDTD. In the time-marching techniques, an explicit method is simple and effcient, but it is only conditionally stable, which means small and fne structures can lead to a very tiny time-step increment and can thus make the calculation time unrealistically long. An implicit method is unconditionally stable and is not limited by a tiny time-step increment on fne structures, but its operation requires inversion of denser matrices. The solution time can be prohibitive when running electrically large problems. Modern engineering simulation problems, such as SQIF overall performance in a platform, have many more challenging requirements on the simulators due to a multiscale nature. None of the current available simulators can solely resolve true multiscale problems accurately with reasonable computation cost. Realizing these limitations, researchers and engineers began to consider hybrid techniques to simulate multiscale problems. This need has been realized for several years, and numerous studies have been performed to arrive at a good strategy in hybridization of different simulation techniques. Nevertheless, the technology gap is still far from being completely flled in software implementation. Accurate communication between different transient simulators is a critical factor that can impact the accuracy and effciency of hybrid simulation techniques. To determine the best strategy of communication, a huge amount of fundamental research is needed. Even though the strategy of hybrid simulation techniques is becoming more mature today at the research level, the implementation tasks are tremendously challenging. Similar to applying high-performance computing techniques to matured codes, the real challenge of applying hybrid multiscale simulation techniques resides in revising the code, which is developed based on outdated simulation logic. The software framework determined years ago with the consideration of only a single simulation method is extremely hard to revise and adjust to the new design principle of hybrid simulation. Both revising the old software framework and a complete November 2013 / st 15

Articles in this issue

Links on this page

Archives of this issue

view archives of Sea Technology - NOV 2013