Research Projects Supported by HKU's High Performance Computing Facilities
ˇ@
Researchers:
Dr C O Ng and Miss Emily W M Ho, Department of Mechanical Engineering
Project Title:
A Numerical Study On Turbulent Oscillatory Plane Couette flow
Project Description:

The objectives of research project are to study the turbulent structures and passive scalar transport (in the form of heat flux) in oscillatory plane Couette flow at Reynolds number equal to 5,000 by solving the unsteady incompressible Navier-Stokes equations with direct numerical simulation.

Plane Couette flow is the basic underlying flow of many mechanical systems such as hydrodynamic lubrication. It involves the flow of a fluid sandwiched between two plane rigid walls, where the flow is forced by one of the walls moving in its own plane.  When the flow is steady (i.e., the wall moves at a constant speed), the velocity profile is always monotonous and the shear stress distribution is constant across the channel, irrespective of the flow being laminar or turbulent.  In the steady case, turbulent plane Couette flow can be conveniently solved by assuming the classical log-laws to hold locally near the walls.  In sharp contrast, no simple analyses will come by when the flow becomes unsteady, as in the case when one of the walls oscillates periodically with time in its own plane.  Difficulties arise owing to the time-dependent accelerating terms in the Navier-Stokes equations. The oscillatory boundary may lead to periodic flow reversal and therefore detachment of the boundary layer from the wall. The classical theory of wall turbulence no longer applies here, and the problem can be solved only through some well-devised numerical computation. To this end, a contemporary numerical technique, known as the direct numerical simulation (DNS) will be adopted in this work.  DNS is computationally intensive and resources demanding, but it is free from empirical closure hypotheses. Results generated by a good quality DNS are usually highly regarded by the research community.

- Back to top- / Contents
Project Duration:
September 2002 to August 2004
Project Significance:
The proposed work will be of significance in two aspects: (i) a non-classical physical problem of oscillatory turbulent Couette flow is studied for its complex system of turbulence transport; (ii) DNS is used to produce the results, which will be of archive value serving as a benchmark for other similar studies. Findings from this work will promote a better understanding on a fundamental fluid mechanics problem that has not been satisfactorily solved thus far.
Results Achieved:
PUBLICATION LIST

1.  ˇ§Direct numerical simulation of turbulent plane Couette flow and scalar transfer at Reynolds number equal to 5000ˇ¨, Proceedings of 7th HKSTAM, 2003.
http://www.hkstam.org.hk/paper03/22.htm

2. ˇ§Numerical simulation of turbulent oscillatory plane Couette flowsˇ¨, Compilation of Abstracts, Second M.I.T. Conference on Computational Fluid and Solid Mechanics, Boston, USA, June 17-20, p. 129, 2003.

3. ˇ§A numerical study on turbulent oscillatory plane Couette flowˇ¨, Conference paper submitted to the 17th National Conference on Hydrodynamics and the 6th National Congress on Hydrodynamics, 2003.

PARTICIPATION IN SEMINAR AND CONFERENCE

1. ˇ§Direct numerical simulation of turbulent plane Couette flow and scalar transfer at Reynolds number equal to 5000ˇ¨ was presented in Mechanical Engineering departmental seminar (HKU) on March 7, 2003.

2.  ˇ§Numerical simulation of turbulent oscillatory plane Couette flowsˇ¨ was presented at Second M.I.T. Conference on Computational Fluid and Solid Mechanics, Boston, USA on June 17, 2003.

3.  ˇ§A numerical study on turbulent oscillatory plane Couette flowˇ¨ to be presented at the 17th National Conference on Hydrodynamics and the 6th National Congress on Hydrodynamics, 2003.
ˇ@

- Back to top- / Contents

Remarks on the Use of High Performance Computing Cluster:
The present computation has been performed by means of a super parallel computer cluster (hkuhpcc) housed in the Computer Centre of the University of Hong Kong. The system consists of 32 nodes of dual Pentium III 1-GHz CPUs equipped with 1 GB of memory. The peak performance of the system is about 76.8GFLOPS (Rpeak value).

It took more that 100 hrs (accumulative) of computational time using 32 processors to complete this calculation and required about 50M byte of memory for each sub-job.
ˇ@
Email Addresses:
cong@hku.hk (Dr Ng) and howaiman@graduate.hku.hk (Miss Ho)
- Back to top- / Contents