Although a lot of high quality tools for (mono-disciplinary) simulations are available today, it is not possible to do multi-disciplinary computations as described above. During the EU funded project CISPAR, researchers, software-engineers and end users cooperate to specify, implement and use an open interface (COCOLIB = Coupling Communication Library) for the coupling of arbitrary simulation codes.
Technical and Scientific Approach
Each code has its own internal data representation for the physical problems to be computed. The space where the fluid moves or where structure is deformed will be represented by different mesh types. In the example of the artificial heart valve the fluid space has the form of a regular grid mesh, whereas the flaps of the heart valve itself are defined by irregular triangles (finite elements). Each point in the grid or cell has several attributes: e.g. pressure or velocity for the fluid or force and deformation coefficient for the structure. Those values will be updated during the simulation run. To investigate the interaction between both parts the structure deformation has to be derived directly from the pressure value of the fluid; i.e. values (e.g. pressure) from one part have to be transformed into those of the other side (e.g. force). In order to transport values across the coupling border, first the neighborhood between grid points and triangle elements has to be computed. If they match exactly, the values can be transported without any change. Otherwise they have to be interpolated according to the geometric distance between point and triangle. All tasks - neighborhood computation, interpolation and communication - will be supported by COCOLIB. Additionally neighbourhood information has to be kept up to date all the time: in case of the opening heart valve former separated fluid grids (closed flap) have to be combined into one, if the flaps are totally open.
Besides the more numerical problems - coupling of simulation meshes - the integration of different software programs into one application has to be realized. Up to now a single simulation code has exclusive access to the parallel machine it is running on. But in the coupled case two or more codes have to share those resources. COCOLIB has control on this and the ongoing coupling communication.
Actual models of artificial heart valves are made of thermoplastic elastomere and resemble the natural heart valves in form. Functionality and operation mode are defined by the dynamics of the blood and the structural behavior of the flaps. The aim of an industrial design of such a valve therefore is a coupled simulation of fluid as well as structure. Structural analysis has to compute large deformations, time dependencies, a non-linear behavior of the material and contact problems at the tips of the flaps. On the other side the fluid is influenced by inlet and outlet conditions (of the heart regarded as a pump), turbulences, viscosity and the deformation of the closing and opening flaps.
Aerospatial - Cooling a hypersonic combustion chamber: The design of local cooling techniques (e.g. injection of hydrogen) is a necessity due to the very high level of convective fluxes in a scramjet combustion chamber and to the peaks of fluxes. Therefore, the main purpose of our participation within CISPAR is to determine the equilibrium heat fluxes and the strains and stresses occurring inside the wall between the hydrogen tank and the scramjet chamber.
Sulzer Innotec - Artificial heart valve: The design and development of artificial heart valves require a detailed examination of structural behaviour and fluid dynamics. Within CISPAR, coupled computation will provide information about structural loads, bending of stent cusps, opening/closing of the leaflet and flow conditions during operation. Simulation of fluid-structure interaction supports product design and process optimisation.
Mercedes-Benz - Rubber flaps and torque converter: In the shroud for the electrical fan in front of the condenser there are six openings to control cooling air. Rubber flaps moved by the pressure open/close these openings which requires the simultaneous solution of a fluid and mechanical problem. A second test case is the torque converter whose shape is being optimised in order to minimise the mass. The goal within CISPAR is to optimise the shroud design and torque converter before first prototyping in order to save costs.
Germanischer Lloyd - Slamming of a ship bow: When a ship encounters high waves its bow area might be elevated out of water. The hull falls down some seconds later and slams onto the water surface which causes high impact pressure at the slamming area. The analysis of the slamming induced impact pressure yields a fluid-structure problem. The goal within the CISPAR project is to investigate loads induced by slamming in order to take them into account in ship design.