Who can benefit from Simcenter Amesim - 3D CFD co-simulation?

A large portion of the industrial design cycle is progressively being replaced with, and complemented by, Computer-Aided Engineering tools – which include CFD, as well as 1D system engineering packages.

A particularity of the design cycle is that it incorporates different levels of detail: earlier stages look at the product in a more macroscopic manner, whereas later on in such a project, finer and finer levels of detail need to be taken into account.

In the context of 1D/3D CFD co-simulation CFD co-simulation , this implies that both the 1D systems engineer and the 3D fluids engineer working on the same project will benefit from carrying out co-simulation between their respective domains.

Indeed, the systems engineer may have to take multiple highly localized design approaches into account: not only CFD, but also Finite Element Analysis (FEA) for structural analysis, as well as specialized solutions for electrics, vehicle dynamics, and so on.

Directly receiving results from a CFD simulation of a given component will therefore more accurately inform the systems designer about the effect of this component on the global performance of the system, contributing to a better understanding of it as a whole.

Figure 6: The 1D user’s perspective

Analogously, the CFD engineer working on (for example) optimizing the design of an injector nozzle, will get a lot of useful information from knowing what the global injection system design is. This translates into boundary conditions which are more representative of the global system.

The dynamics between different components in a system can often be linked – in other terms, these components can be physically coupled. Ignoring this coupling may result in non-representative simulation results, which is one of the strongest arguments for using co-simulation as a means of exploring and fully accounting for such couplings.

Figure 7: The CFD user’s perspective

These different user communities clearly have common interests and use-cases where co-simulation may be of great benefit. Some of these real-world areas of interest are, for example:

• Lubrication

• Fuel injection

• Hydraulic component design

• Engine thermal management

• Electrics and electronics cooling

• Fluid-structure interactions for immersed structures

• etc.

In industry today, co-simulation is used at different stages within a design cycle, depending on the domain in question – below is a flowchart highlighting a realistic case where co-simulation is used part of the design process.

Figure 8: A typical industrial design process involving the use of 1D/3D CFD co-simulation

Both user communities mentioned above can extract a significant benefit from employing 1D/3D CFD co-simulation; however, the typical setups for simulating Simcenter Amesim models and 3D CFD models are exceedingly different:

• Simcenter Amesim models can generally be run on a single computer – dual- or quad-core processors and industry-average (or above-average) RAM are sufficient. This is becoming less true as users design larger and more complex models, but simple, basic models can still be run on a single machine, with runs usually lasting in the order of minutes (hours in the most extreme of cases).

• CFD programs solve highly non-linear algebraic equations in 3-dimensional space. Depending on the length scales and gradients involved, spatial discretization  Discretization is very often extremely fine – resulting in a calculation over millions of cells  Cell for advanced models. As a result, it is not at all unusual for industry-level simulations to run on HPC clusters, or highly parallelized computing setups dedicated to the CFD simulations. Even in these configurations, CFD simulations are expected to last in the order of hours, and can last up to days (depending on the model).

As a result, it can be said that given the computational intensity of CFD simulation, the CFD user community will suffer a negligible hit in terms or performance or cost, all the while incorporating the added value of having 1D simulation data available during CFD simulations (see the Motivations section for more information about this.)

On the other hand, 1D systems design teams do not currently deploy the same kind of computational setups, and indeed, CFD data may come with a significant cost associated with making such resources available. However, if a part of the system is particularly difficult to optimize or highly geometry-dependent, some 1D user communities may find that this investment is worthwhile.