Direct Numerical Simulation of Reverse Transition in Rotating Pipe Flow

Christoph Brehm, University of Kentucky

Usage Details

Turbulent swirling and rotating flows involve intricate flow physics that are not well-understood, including turbulence suppression and relaminarization, one of the biggest mysteries in turbulence research. The axially rotating pipe is an exemplary prototypical model problem that exhibits these complex turbulent flow physics.  For this flow, the rotation of the pipe causes a region of turbulence suppression and relamiinarization which is particularly sensitive to the rotation rate and Reynolds number. 

While the laminar- turbulent transition process has been studied in great detail for decades, there has never been a DNS study that covers the entire reverse (turbulent to laminar) transition process. In conducting this study, a sufficiently high Reynolds number must be used to ensure an adequate separation of scales.  Significant temporal and spatial resolution is required to thoroughly study the intricate nature of turbulence, turbulence suppression and relaminarization.  Particularly, to obtain data that is suitable for the description of the entire statistical distribution of the dissipative scales of turbulence these DNS simulations will require sub-Kolmogorov scale grid resolution. 

Experiments by Murakami et al. showed that fully-developed flow cannot be achieved until L/D = 100, thus requiring a domain length of 120-140 to ensure the full reverse transition process is captured. The high resolution and large domain necessitate a grid size with 22 x 10⁹ points to sufficiently capture the turbulent scales at a Reynolds number of Re = 37,000. In addition to this size requirement, a temporal domain of 6 x 10⁹ time steps is required to adequately capture the physics of reverse transition, leading to a requirement of 368M core-hours. Leadership class computing allocation resources are required to conduct the first ever reverse transition simulations of rotating turbulent pipe flows. 

A key objective of this research is to obtain high-quality simulation data (in conjunction with ongoing experiments) that can be used to study the nature of the highly complicated flow physics of turbulent rotating flows. While numerous research studies have targeted the laminar to turbulence transition process, very few (almost no) studies were concerned with investigating the reverse process, particularly within the framework of stability analysis. 

As such, the intricate nature and relevant physical mechanisms of this process are not well understood and the proposed research offers an opportunity to understand the physics of this process. A profound understanding of the relevant mechanisms can be employed to devise passive or active flow control strategies for turbulent flows.  Another significant contribution of the proposed research is to improve current computational prediction capabilities of swirling and rotating flows.