| Computational Fluid Dynamics can be used for many different tasks: From fundamental
low-speed Direct Numerical Simulation of the driven flow in a very simple rectangular
cavity to a Reynolds-averaged steady-state Navier-Stokes calculation in or over a complex
"real world" geometry. A complex geometry in the industry can be anything from
an external shape like a body-wing fairing or an internal shape of a machine or even a
combination of both (the "under-the-hood" engine cooling flows of cars and
trucks). Examples of complex internal geometries can also be found in an engine block. The
internals of an engine is a challenging object for the fluid dynamics analyst. At the
air-side one encounters two-phase flows (evaporating direct injected fuel droplets),
combustion, moving meshes (opening and closing of valves combined with a squishing piston)
and the simple "complex geometry" of an intake port. At the other side of the
cast iron one finds water as the coolant medium. The water is forced through the block and
the head and cools the cylinders and exhaust ports. As the heat distribution influences
the distortion of the block and head it is an important design topic. Experimental
assessment of the flow in the engine is nearly impossible and computational fluid dynamics
is an interesting alternative.
The numerical analysis of the water coolant
flow starts with the generation of the grid. This task is the most difficult and
considerable time-consuming part of such a project. When no CAD-data are available the
geometry has to be defined first (generally speaking, the pre-processors of CFD-packages
are not very suited for the definition of complex geometries) and then transferred to the
grid-generation code. If CAD-data can be used it has to be imported into the
grid-generator using some standard like IGES or VDA/FS. A smooth image of just a
small part of this data is shown here.
From this point on one can start with the actual mesh generation. The three before
mentioned sub-tasks (reading in CAD data, modifying and/or creating surface data and
grid-generation) were performed using ICEM/CFD DDN and ICEM/CFD MULCAD. The use of this
grid-generation software suite, with embedded CAD functionality, has been shown and
discussed. ICEM/CFD MULCAD is a sophisticated block-modeller package and is known to
produce very good meshes. Although in the near future the use of special automatic
meshers, such as ICEM/CFD TETRA, will account for the bulk of the meshing of complex
geometries, traditional blockmeshers will remain in use when high quality grids are
needed. Also a hybrid use of hexahedral and tetrahedral meshes is foreseeable in the
future.
The created MULCAD-model of the presented engine water coolant
jacket contains about 300,000 hexahedral cells (and only 1004 non-hexahedral cells) made
out of about 750 MULCAD-meshblocks (which is considerable less than the number of
"ordinary" meshblocks which should otherwise have been allocated). Just one view
of the mesh:
 The mesh
of the complete model.
The flow was solved using a finite volume code and the standard k-e
turbulence model. Finally some examples were shown of the numerical results of the flow
through the engine (see the figures of the velocity vectors in the block and the
spagetti's in the head) and a statement about the accuracy of the simulation using
measured data has been made. Please note that the only thing which was additionally
specified besides the mass flow was the flow split between the head and a small secondary
outlet in the cyclinder jacket.
 Velocity
vectors in the jacket.
 Streamlines in the head.
 Total
pressure field.
The calculated pressure drop, dominated by the small passage from the block towards the
head (see the last figure), corresponds very well with the measured pressure drop. A fine
result of which we are proud of. |
