Fluid thinking

The analysis of real-world mixed physics, once the field where intractable problems were relegated, is fast becoming a must for engineers, explains Charles Clarke.

In our last exposé on CFD/FEA we mentioned that mother nature never does anything ‘discretely’. Natural phenomena derive from a complex interaction of physical attributes involving stress, vibration, fluid flow, heat transfer and more.

While it is true that computational fluid dynamics (CFD) software has come on in leaps and bounds in recent years, multi-physics analysis (MPA) — the analysis of real world mixed physics events — has become a realisable dream.

As we tend to ignore non-linearities in general purpose analysis, the effects of physical realms beyond the one we are investigating often escape attention. But anyone doing non-linear analysis will testify that there comes a point when you simply have to include complex phenomena. ‘Solving the problem that you can solve easily, rather than the one which needs solving, can only get you so far,’ said Laurence Marks, of UK multi-physics company Physica.

Many problems can be solved adequately in a single physical domain without considering the effects of cross coupling (as until now we could not do it cheaply or easily). There is, however, a whole range of problems in which the coupling between physical realms cannot be ignored. Advances in computing power mean that problems once thought intractable can now often be solved.

Vehicle ‘under hood’ thermal management is a classic example. The need for increased interior room and passenger comfort has resulted in more compact under-hood packaging. Also, emission requirements have led to catalytic converters being installed closer to the engine block, producing more heat radiation in the engine compartment.

The complexity of the geometries has made it expensive to couple CFD analysis of the engine compartment aerodynamics with convective and radiation heat transfer, in terms of problem preparation and parametric studies. Until now simulation has lagged behind design.

The ESI Group offers a solution based on CFD-VISCART, an automatic shrink-wrapping and 3D body fitted volume meshing tool, which reduces the time to conduct under-hood simulations from several weeks to a few days.

Using ESI’s CFD-ACE+, an advanced CFD and multi-physics solver for fluid, thermal, chemical, biological, electrical and mechanical phenomena, predictions made using this technique together with convective and radiative heat transfer came to within 3 per cent of physical vehicle tests.

Fluid structural interaction is well enough established to have its own acronym — FSI — and it is probably the first and most obvious example of MPA. The strong interactions between aerodynamics and the structure of an aerospace vehicle or device are an important design criterion. FSI is a non-linear problem and can lead to large structural oscillations and aeroelastic failure if it is not addressed early in the design cycle.

The aeroelastic performance represented by dynamic load prediction and vibration levels in aerodynamic surfaces has a great impact on vehicle design as well as in the cost and operational safety. Biomechanics provides a multitude of fluid structural interaction examples: animals are largely composed of bags of fluid, supported by more or less saturated structural elements. The simulation of bio-structures, such as drug delivery systems, often has to consider fluid structural interaction, especially if a dynamic response is of interest.

In blood flow, the flow regime is closely coupled to the structural response of the artery. Bone and other elements tend to be saturated and therefore any structural response is closely coupled to the fluid behaviour. In biomechanics, electro-chemical reaction adds another physical dimension, which must be considered for the simulation of critical physiological processes.

The performance of bio-analytic systems based on biological micro-electro-mechanical systems and biochips is being developed for monitoring, diagnostics and drug delivery. Multi-physics simulation can be effectively used for the design and optimisation of these devices.

The goal of all biochemical assay devices is to yield the maximum signal in the shortest time. The signal, however, depends on many variables such as geometry, flow rates and concentrations in an interdependent manner. If you can produce an accurate MPA simulation, then you can ring the changes geometrically to produce the optimum design.

The scripting capability and the simulation manager of CFD-ACE+ allows an automated design evaluation framework to be developed to facilitate rapid exploration of parametric optimisation studies cost-effectively.

Production processes are also a happy hunting ground for multi-physics; casting processes have been examined empirically for years but casting is a classic MPA issue. This can be extended into welding, where you not only have thermal, fluid, and structural coupling but also the effects of electro-magnetics. Consideration of each phenomenon in isolation will not allow you to optimise the process.

Continuous casting is a widely used production process. Its simulation involves the study of turbulent free surface flow with heat transfer, simulation of gas bubbles and solidification coupled to solid mechanics in the solid region. An involved simulation but the financial returns of being able to improve and optimise the process can make the investment in analysis trivial.

The strategy used to solve a multi-physics problem is arguably more dependent on the degree of coupling between the physics than the actual physics themselves. In CFD to finite element analysis (FEA), coupling the geometries is usually quite different. FEA tends to use elements from the surface of the tested object, which extend internally, whereas the CFD mesh is generally external to the object.

If coupling is in a single direction there are few problems. The issues become more complex if bi-directional coupling is necessary — differing time step requirements and mesh compatibility all start to come into play and, in some cases, tackling this becomes more challenging than the original analysis.

Complex MPA problems require fully coupled solvers, with all the physics in a single database. Many hi-tech companies have teams of analysts to tackle such problems. Production engineering environments tend to have many MPA processes but not so many skilled MPA practitioners. Hence the benefits of MPA can be lost if the results are not packaged for use by non-experts.

The general purpose solution of multi-physics problems is at an early stage development. As computer hardware gets more capable, the scope of ‘possible’ simulations broadens. We gain the ability to optimise processes further by increasing the number of parameters that we can consider.