Industrial vision and fundamental research, the same battle
Since 2017, APREX Solutions has been developing and marketing machine vision solutions for the most diverse needs (citer ceux qui font le plus sens : plasturgy, trefilage….???). The power and versatility of our algorithms allow us to meet the highest requirements, such as those encountered in fundamental research. Thus, in addition to our activities and expertise in industrial vision, we are happy to assist researchers from different countries to meet their needs in video analysis. Our APREX Track software is at the heart of two recent studies published in leading scientific journals, one in nuclear fusion and the other in fluid mechanics.
Since tomorrow’s applications for industry are being invented today in research centres,
APREX Solutions is involved in cutting-edge research. The founders of APREX are all doctors in science. Some of the algorithms we develop at APREX Solutions result from 10 years of R&D in nuclear fusion.
We are proud to put our technology at the service of many sectors of activity, and of the international recognition of the value of the technology transfer thus achieved (in 2019, APREX Solutions received the EUROfusion FUTTA award for technology transfer from fusion research to optical fibre quality control, read the article – in French).
Today, we continue to develop cutting-edge solutions for industry and research, particularly in nuclear fusion. Our tools have enabled to carry out an international study that has just been published in the leading scientific journal Nuclear Fusion [1]. The work presented in this study was carried out at the DIFFER laboratory (Eindhoven, The Netherlands) in collaboration with the Jean Lamour Institute (IJL – Nancy, France).
The objective of this work was to study the formation of gas bubbles and the ejection of molten metal droplets from a liquid metal sample placed in a nuclear fusion plasma simulator. The use of liquid metals as the first wall in nuclear fusion reactors is indeed a new and promising avenue that could lead to the construction of nuclear fusion power plants within the next twelve years (for more information, see for example the website of the Renaissance Fusion start-up).
In the study carried out in the DIFFER laboratory, we used our expertise to analyse recordings made with a high-speed camera filming at 4000 frames/s. A major difficulty was the simultaneous tracking of several hundred molten metal droplets with complex trajectories (occlusions, cusps, see the video below), against a spatially and temporally variable light background. The accurate reconstruction of the trajectories under different experimental conditions was indeed a fundamental point for this study. Technically, this challenge was met thanks to the use of self-adaptive multiscale thresholding and a hybrid tracking system made in APREX, combining a « classical » MSD approach and Bayesian theory. At APREX, we (almost) always have the solution, whatever your videos are!
Did you know that from a physical point of view, water is not an « ordinary » fluid? Water belongs to the family of so-called Newtonian fluids, which includes many gases, most solvents, and some low-viscosity oils. In these fluids, viscosity depends only on temperature and pressure, not on external forces or stresses.
Most fluids are non-Newtonian, i.e. their viscosity changes as they flow. The vast majority of fluids used in industry fall into this second category: creams, gels, suspensions, paints, where the rotation of the fluid results in either a trough or a rise of the fluid around the rotor axis.
Understanding the flow of these fluids is a matter of rheology research, and can have applications in fields as varied as the chemical, cosmetic, food and petroleum industries, and many others.
In fluid mechanics, two techniques exist to characterise the movement of a fluid: PIV (Particle Imaging Velocimetry), which provides velocity fields, and PTV (Particle Tracking Velocimetry), which provides individual trajectories. Both techniques generally require the addition of « passive » tracers (dye or particles) to the fluid.
While PIV is generally simpler to implement, PTV allows for much greater data exploitation.
The APREX Solutions team is proud to have contributed to the success of a study, led by Christel Métivier, from LEMTA laboratory (Nancy, France), and in collaboration with LRP (Laboratoire Rhéologie et Procédés – Grenoble, France) and IJL (Institut Jean Lamour – Nancy), recently published in the Journal of Non-Newtonian Fluid Mechanics [2].
This study required the ability to track tracers injected into the fluid and illuminated by a laser sheet over a long period of time, despite a high level of noise, spatially inhomogeneous, and locally comparable in intensity to the luminosity of the tracers.
This challenge was successfully met thanks to the combination of self-adaptive multiscale thresholding and predictive tracking based on the Bayesian approach [3] included in our APREX TRACK software.
Analysis of the recordings revealed the existence of an oscillatory convection regime in a viscoelastic gel subjected to a temperature gradient (heated from below): when the gel slides on the wall of the container, Rayleigh-Bénard convection can give rise to oscillatory movements in the gel in the form of convective structures that « wind » and « unwind » elastically, in the manner of a spring that stretches and relaxes.
This research is currently at an internationally recognised fundamental level, but could be used in the future to optimise the flow or mixing of fluids in many areas, both in R&D and in various industrial applications.
At APREX Solutions, we are already ready!
References