NERSC Initiative for Scientific Exploration (NISE) 2011 Awards

High Fidelity Study of Multiscale Multiphase Lubricant Flow in High-Speed Gear Systems

Xiaoyi Li, United Technologies Corp Research Center

Associated NERSC Project: Multiscale/multiphase research for first-principles fuel atomization calculation (m945), Principal Investigator: Marco Arienti

High-speed interlocked gear systems have a wide spectrum of applications across many business units in United Technologies Corporation, including Sikorsky, Pratt Whitney, Hamilton Sundstrand and Clipper. The geared transmission systems are also heavily used by automotive industry. In all the gear systems, sufficient oil supply must be delivered to the gears and bearings to provide adequate lubrication and cooling throughout the operating range. On the other hand, excessive supply can lead to higher power loss by friction among oil, air and gears. Therefore, studying the oil lubrication of gears is crucial for improving its durability as well as performance. The proposed study can help fundamentally understanding the complex physical processes in gear systems and potentially impact the design of such systems in many industrial products.

The lubrication processes in complex gear applications physically pose a challenging problem of multiphase oil-air flow interacting with complicated solid gear moving at high speed. No simple analytical/numerical models can give reasonable prediction of the involved processes except some high-cost direct simulation approaches based on first principles. In this work, we will directly solve the fundamental multiphase Navier-Stokes flow, using a coupled level-set and volume-of-fluid method to track oil flow and an embedded boundary approach to account for the motion of solid gears. The typical simulation domain is large containing large gears. However, resolving the details of oil introduction pose a stringent demand for high grid resolution at the oil-air interface. In addition, the high rotating speed of gears induces complex turbulent oil-gas flows that also have to be well resolved using high grid resolution. We have first applied adaptive mesh refinement (AMR) to manage the grid count and simulation speed within feasible range. A low resolution simulation at typical high gear speed has been demonstrated on a small-scale (64 cores) cluster at United Technologies Research Center. Here we propose to run a set of full-scale simulations at NERSC supercomputing facilities to do parametric exploration. The results would benefit the design of lubrication devices for many gear applications in industry.