Sites Inria

Version française



A computer full of sand

Tomography Tomography

What are the laws of physics that determine the movements of millions of grains in a silo?It’s a question that has baffled scientists for over a century.The work of Nicolas Brodu, a post-doctoral researcher in the Geostat project team at Inria Bordeaux - Sud-Ouest, has thrown a little more light on the matter.

We have all tried to take a handful of sand from a bucket and been surprised to find that there is a point at which it’s impossible to push your hand in any further, only to be further surprised when the grains of sand run out between our fingers like water. These strange effects are a result of the collective behaviour of the grains, which can change dramatically even though the grains themselves remain the same throughout the process. They are typical of complex systems, a field that Nicolas Brodu has studied for several years. His latest research has just been published in the prestigious journalNature Communications. By combining his skills as a computer scientist with those of the experimentalist Joshua Dijksman, they have put together a multidisciplinary project that also builds on the work of the American physicist Robert P. Behringer.

From experimentation …

In order to verify the behaviour of a granular material, Joshua Dijksman designed a system capable of deforming the material in a controlled manner. Nicolas Brodu developed the algorithms used to measure the forces acting on each individual grain. Together, they recreated deformations in the laboratory that simulate the passage of vehicles over a road, or the triggering of a landslide, and measured the three-dimensional forces acting between each grain deep within the pile. “The system works a bit in the same way as a body scanner”,explains Nicolas Brodu,“It uses tomography techniques to provide an accurate view of all the movements within the granular material, together with all the forces acting within it, with unprecedented accuracy”.In order to design the tomography scanner, Nicolas Brodu and Joshua Dijksman spent a year and a half at Duke University in Durham, North Carolina, in the group led by Robert P. Behringer, a leading international expert in the physics of granular materials. Nicolas Brodu then returned to Inria Bordeaux - Sud-Ouest, where he is continuing this work in active collaboration with his colleagues, while at the same time contributing to the research work of the Geostat team.

… to simulation

These results are of particular interest to physicists as this is the first system that is capable of directly and accurately measuring the three-dimensional forces acting on each individual grain while at the same controlling the deformation of the material. For the first time, it is thus possible to confirm numerical models of these deformations by means of actual measurements – and to determine their limitations. As a result, Nicolas Brodu has proposed a new model for these grain interactions, which better takes into account their multiple contacts within the granular material. He then simulated the experiment described inNature Communicationsas faithfully as possible. The very small differences between the predictions of the model and the measurements made at Durham are a promising indication of further advances in material physics. This work using the numerical simulation has also been published inPhysical Review E.

"Computer technology has now become a fundamental component of the scientific method. It is no longer just a tool, but an integral part of the process of measuring, modelling and predicting nature”,he concludes,“What we today refer to as complex systems are a reflection of this change, of our own inability to observe and understand collective phenomena without the help of computers.By making these effects accessible, scientific and multidisciplinary computing is already a standard for the future".

"We have progressed from 2D to 3D!"

“For the first time, researchers have succeeded in modelling the individual elements of granular materials in a more realistic manner than simply assuming them to be spherical. This is a considerable advance because, since the sixties, we have been restricted to two-dimensional systems. Now that we can work in 3D, we have a huge reserve of hypotheses to verify. Our priority is to concentrate on the transitions between the various states of a granular material, for example from a solid to a fluid. Many industrial systems fail, some with severe consequences, as a result of our continuing lack of understanding in this area. It is true that our work is only just beginning. In particular, we need to better take friction into account. But the tools developed as part of this collaboration are continuing to improve every day”.

Testimonial - Robert P. Behringer, Professor of Physics, Duke University, USA

Keywords: Computer science Physics Multidisciplinary collaboration Tomography