739
Views
0
CrossRef citations to date
0
Altmetric
Editorial

Interdisciplinary CFD

Pages 333-335 | Published online: 30 Oct 2012

While CFD developers continue to improve solvers, increase capabilities and high performance efficiency, it is important to note that the CFD user community has significantly widened over the past few decades. CFD has become so ubiquitous that my kids discuss the merits of the portrayal of fluid behaviour in their video games with me on a routine basis. The purpose of this special issue on “Interdisciplinary CFD” is to (1). showcase a number of fairly different fields where CFD plays a major role; (2). illustrate different approaches for integration of CFD into complex problem investigations; and (3). encourage cross-fertilisation of ideas between the CFD community and other disciplines.

My own interest in interdisciplinary CFD came at a time of reduced funding in the aeronautics field in the US in the 1990s, and hence the need to look elsewhere for university research funding. There was no shortage of people asking for help with fluid modelling, be it in combustion, medical applications or virtual reality. I eventually settled on combustion, MEMS modelling, and, later, meteorology as “applications” of my advanced adaptive spectral element solvers. In each community, there was a steep learning curve for me, as well as for my collaborators from other disciplines. For students, they had to master mathematics, computer programming and fluid dynamics, and other disciplinary fields. The demands on students and developers continue to mount today with the flood of available tools and applications, and these are somewhat served by commercial, and more recently by open source, codes providing a comprehensive basis on which to build a wide range of CFD applications.

However, some applications cannot take as their starting point a CFD code based on the traditional Navier-Stokes equations. For example, in atmospheric science, I quickly learnt that the set of equations you start out with depends on the investigations to be pursued. Atmospheric models vary according to their intended use, with global circulation models being used for long-term climate studies, global weather forecasting for large-scale predictions and limited area models for local or national weather prediction. These different situations lead to different assumptions in the equation models from the outset. For example, while a hydrostatic assumption is the norm for global circulation or global weather models where the vertical direction is much smaller than the horizontal, the non-hydrostatic equations must be used for local and severe weather prediction. However, with an increased interest in higher resolution across the board, the need to treat the vertical direction more completely requires a non-hydrostatic assumption.

Furthermore, an increase in resolution is not the sole answer, nor is it attainable. We simply cannot resolve every feature of the atmosphere, whether it is for the entire globe or only for a continent. This brings up issues of operating at the limit of resolution, a topic common to several fields, be they weather modelling, ocean dynamics or biomedical modelling. Indeed, the last ten years of CFD development have focused on multiscale modelling, high performance computing and adjoint-based optimisation to tackle this issue.

Increase in resolution brings up other challenges to established modelling codes. Contemporary atmospheric models, for example, are made up of two basic components: the dynamics core which solves the fluid mechanics equation models of the desired atmospheric flows, and a “physics” package to treat the complex associated phenomena of radiative transfer, photochemistry, phase change, precipitation, etc. While most atmospheric modellers consider the fluids the “easy” part, the “physics” take up the bulk of the calculation and development time. These “physics” models have been built and tuned over many years on regular grid models with relatively low resolution and are now found to respond poorly to increased resolution. Clearly, this subject needs to be rethought, particularly in the context of adaptive grids.

Cross-disciplinary teams have succeeded in resolving some of these issues and in pushing the boundaries of established computational methodologies. For example, this fall, the Isaac Newton Institute for Mathematical Sciences at the University of Cambridge hosts a six-month program for Multiscale Numerics for the Atmosphere and Ocean, organised by academics, national laboratory and national weather service scientists, from mathematics, computer science and ocean and atmospheric science disciplines. One particular focus of this program was addressed in August by a workshop on adaptivity. Similar programs exist in other fields involving fluid dynamic modelling.

Over the last fifteen years, validation and verification in CFD has become a field in itself. Comparing our simulations with established test cases, experiments or real data has become more and more of a necessity, especially when using a commercial code for which one either does not have the source or for which the user is unaware of the underlying methodology. Moreover, in some fields, incorporation of observational data into models is a well-established tool to relate modelling to real data, such as the use of data assimilation techniques in atmospheric modelling.

The papers selected for this issue cover a wide range of fields: ocean dynamics, fire simulation, biomedical flows, helicopter design and analysis and computer animation. While many more application fields exist, this selection covers, in my opinion, many of the challenges facing CFD, while showcasing investigations of considerable interest to society that benefit from the advances of CFD over the last thirty years.

Özgkmen and Fischer's paper, “CFD application to oceanic mixed layer sampling with Lagrangian platforms”, highlights challenges in the ocean modelling community that share many aspects with atmospheric modelling as described above. The paper includes a discussion of the hydrostatic assumption and its limitations with regard to oceanic mixing layers at the sub-mesoscale range, degradation of hydrostatic models with increased resolution, established parameterisations and data assimilation techniques in conflict with increased resolution and accuracy developments, addressed here by using spectral element direct numerical simulation and large eddy simulation. The investigation shows a concerted effort to constrain the sampling of the modelled sub-mesoscale flows to the parameters and conditions of the experiments, that can be ship-borne, satellite-based, radar-based or clusters of drifters.

McGrattan et al's paper on “Computational fluid dynamics modelling of fire” describes the US National Institute for Standards and Technology's tried and tested method for practical simulations of fire in difficult environments, such as complex geometries of public buildings. As the code is intended as a tool for fire researchers and fire protection engineers for the design of fire protection systems and fire event reconstruction, the need for simplified modelling, in this case using the low Mach number formulation in order to capture large scale fire dynamics, with large eddy simulation modelling for the small scales, is justified. Combustion modelling, of course, requires simplification of the multiple species tracking, and modelling of heat release rate, radiation processes and heat transfer from surrounding surfaces. These simplified “physical” processes are needed in the same way as those described above for atmospheric modelling. Validation cases and comparison to real data are included.

Pollard et al's paper on “Recent advances and key challenges in investigations of the flow inside human oro-pharangeal-laryngeal airway” describes a tight connection between CFD modelling and experimental efforts to describe the complex mixed laminar/ turbulent flow inside the human upper airway. The paper makes a significant pitch for validation of modelling using idealised (rather than true anatomical) geometry experiments, and discussing the experimental data gathering methodologies as well. The relative merits of direct numerical simulation, large eddy simulation and Reynolds-averaged Navier-Stokes approaches are also discussed in this transitional context.

In “Rotorcraft simulations: a challenge for CFD”, Costes et al describe ONERA's modelling capabilities for rotorcraft in France and highlight the challenges of coupling aeroelastic and aerodynamic modelling capabilities with an eye towards acoustics, in the complex unsteady geometry of a complete helicopter. The challenges include the necessity for accuracy of vortex capturing and wake predictions, and the unknown blade motion determined by rotor trim. Adaptivity and higher accuracy for vortex and wake modelling are explored in relation to wind tunnel measurements. The complexity of the coupling for trim calculations necessitates simplified modelling, as in many of the other applications described in this special issue. Again, wind tunnel test results are used to validate the coupling with a caveat that the wind tunnel model rotor blades are generally not dynamically scaled.

Vines et al's paper on “Computer animation challenges for computational fluid dynamics” provides a review of the tremendous progress towards realistic CFD approaches to efficient, and in some cases real-time, simulation of fluid motion for computer animation. The many clever shortcuts that have been proposed by the computer graphics community are included here as a reminder of what can be simplified and at what cost, a lesson that can be beneficial to other fields where the efficiency to accuracy ratio may be slanted in order to produce practical modelling tools. The challenges that remain are described to incite more CFD developers to jump in to this potentially lucrative field. In the case of computer animation, few validation and verification procedures seem to be established, since realism to the observer and efficiency are the main criteria of performance.

Together, these five papers show the breadth of progress in CFD through its contribution to disparate fields with some unique but many common challenges. All but the computer animation papers speak of validation through planned comparison with experimental data and the challenges therein. All of the applications are so complex due to their interdisciplinarity that they necessitate simplifications and parameterisations that need to be rethought as modelling approaches improve in accuracy and resolution. Deciding on how to limit complexity in the context of finite resources is a common theme for all modellers. Lastly, complex geometry and grid treatment, again in the context of improved resolution capabilities, are being tackled in a number of ways with adaptivity in view, but are not yet providing significant performance improvements in a number of interdisciplinary fields.

I would like to express my sincere appreciation to the contributors to this special issue on “Interdisciplinary CFD”. While each of the authors has significant experience in CFD, they bring to us, in addition, the unique perspective from their respective application fields and have touched on significant challenges for all of us to tackle. I encourage you, the reader, to contact them or to get involved in interdisciplinary projects and to consider how you can contribute to enhancing the power of CFD in diverse fields from the Earth's core to the upper reaches of the atmosphere.

Reprints and Corporate Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

To request a reprint or corporate permissions for this article, please click on the relevant link below:

Academic Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

Obtain permissions instantly via Rightslink by clicking on the button below:

If you are unable to obtain permissions via Rightslink, please complete and submit this Permissions form. For more information, please visit our Permissions help page.