Please return to your seats and fasten your seat belts…
Dr. Ugo Piomelli walks up to his window and points to a plume of vapour rising out of a duct into the cold February sky. “That,” he explains, “is what I study. Turbulence. Notice the way the vapour doesn’t rise in a straight line. It is not linear. There are instabilities that create a quasi-chaotic state.” Indeed, turbulence is the sort of motion that dominates the real world around us, and yet it is perhaps the longest running unsolved problem in classical physics.
As well as being a professor in the Department of Mechanical and Materials Engineering at Queen’s University, Piomelli holds the Canada Research Chair in Turbulence Simulation and Modelling and HPCVL-Sun Microsystems Chair in Computational Science and Engineering. Yet his interest in fluid and turbulent motions goes far back, perhaps born of a childhood fascination with airplanes. And not only the engineering and design of the flying machines, but their interactions with the invisible and uplifting fluid we call air. Not only were airplanes a beautiful human achievement, watching them would inevitably bring questions to mind.
Beyond clouds and airfoils, turbulence is found everywhere – from the way cream moves through your coffee, to the flow of water over a whale’s flipper, to the turbulent eddies caused by stents in vascular arteries. Each is a specific problem to which Piomelli has applied his understanding and his research tools.
Oddly perhaps, science has had equations to describe turbulence since the mid-1800s. However, using those equations to solve anything beyond the simplest real-world problems has been beyond human capacity up until recently. Real turbulence is just so complex and dynamic that the calculations require not only computers, but the combined number crunching power of arrays of computers, known as supercomputers.
To take a specific application (and one that Piomelli is working on at the moment), consider the wear and tear that the blades of a hydroelectric turbine undergo during years of operation. The pitting and rust formation on the blades amplify the turbulence and friction in the flow of water over the blades which negatively impacts their efficiency. Yet, resurfacing the blades is a huge undertaking that requires halting that particular turbine and all the lost income that implies. So Piomelli studies the exact relationship between the topology of corroded blades and water turbulence in order to develop useful guidelines for Hydro-determine when it makes sense to resurface a particular set of turbine blades.
The resolution that Piomelli works at is astounding – 600 million points are used to describe the flow over just a small section of a corroded turbine blade. And then the interaction of the roughness with the water flow is simulated over time at intervals of fractions of a second. Piomelli works in a world far far beyond pencil and paper or blackboard calculations. He and his team at the Turbulence Simulation Lab (TSL) at Queen’s make use of a 240-processor TSM Linux cluster, several clusters at the High Performance Computing Virtual Laboratory and a supercomputer at Hydro-Québec. Even with these supercomputers, running a simulation takes about a month to crunch all the data contained in barely four seconds of real time. Yet it is only at this resolution that the problem can truly be solved.
Of course, no single real world problem is ever as simple as just sending data to be modelled by the computers. One needs to know how to apply the equations to the specific problems in the right way. And that is what Piomelli and his graduate students work on, and undoubtedly the challenge that drives them from one problem to the next.
And while the turbulent flows and interactions under study seem to reduce to reams of data, Piomelli’s language often includes words like beauty and whirls and references to Leonardo Da Vinci’s notebooks. There is true beauty to the whirl of clouds, and the turbulent eddies of a rushing brook. And equally, there is a deep beauty in the understanding of these things.
(e)Affect Issue 7 Spring 2015
Learn more about:
Dr. Piomelli's research