CFD I - Spring 2000

The syllabus Term project, in details next time CFD references Course Tools
University CourseNet: http://coursenet.ou.edu. College of Geoscience WebCT site: http://www.gcn.ou.edu:8900 Computing Facilities ECAS Lab Cray J90 and Departmental workstations Unix and Fortran Helps Consult Links at CFD Home page which is linked in WebCT.

Introduction Principle of Fluid Motion

1. 2. 3. 4. Mass Conservation Newtons Second of Law Energy Conservation Equation of State for Idealized Gas These laws are expressed in terms of mathematical equations, usually as partial differential equations. Most important equations the Navier-Stokes equations

Approaches for Understanding Fluid Motion

Traditional Approaches Theoretical Experimental Newer Approach Computational
CFD emerged as the primary tool for engineering design, weather prediction et al, because of the advent of digital computers

Theoretical FD
Science for finding solutions of governing equations in different categories and studying the associated approxiations / assumptions;

h = d/2,

c1 2gh / 0

Experimental FD
Understanding fluid behavior using laboratory models and experiments. Important for validating theoretical solutions. E.g., Water tanks, wind tunnels

Computational FD

A Science of Finding numerical solutions of governing equations, using high-speed digital computers

Why Computational FD
Analytical solutions exist only for a handful of typically boring problems Can control numerical experiments and perform sensitivity studies, some in very ideaized settings. Can study something that is not directly observable (black holes). Computer solutions provide a more complete sets of data in time and space We can perform realistic experiments on phenomena that are not possible to reproduce in reality, e.g., the weather Much cheaper than laboratory experiments Much more flexible each change of configurations, parameters We can now use computers to DISCOVER new things (drugs, sub-atomic particles, storm dynamics) much quicker

An Example Case for CFD Density Current Simulation

Thunderstorm Outflow in the Form of Density Currents

Positive Internal Shear

g=1

Negative Internal Shear

g=-1

T=12

Positive Internal Shear

g=1

Negative Internal Shear

g=-1

No Significant Circulation Induced by Cold Pool

Simulation of an Convective Squall Line in Atmosphere

Infrared Imagery Showing Squall Line at 12 UTC January 23, 1999.

ARPS 48 h Forecast at 6 km Resolution Shown are the Composite Reflectivity and Mean Sea-level Pressure.

Difficulties with CFD

PDE's are not well-suited for solution on computers - must approximate a continuous system by a discrete one. One must keep in mind that the governing equations themselves are approximations Most physically important problems are highly nonlinear - advantage of models is that you can examine the relative contribution of each term. But, how will you then know if the solution is correct? Validation?! It is impossible to represent all relevant scales in a given problem - there is strong coupling in atmospheric flows and most CFD problems. ENERGY TRANSFERS. (Transparency)

Most of the numerical techniques we use are inherently unstable - often we have to beat down the instability with a hammer, resulting in degraded solutions. We must deal with physical and computational instabilities.

Difficulties with CFD

We often have to impose nonphysical boundary conditions. We often have to parameterize processes which are not well understood (e.g., rain formation, chemical reactions, turbulence).

The real world is continuous, not discrete.

Often a numerical experiment raises more questions than it answers!!

POSITIVE OUTLOOK
New schemes / algorithms Bigger and faster computers Faster network access Better Desktop computers Better programming tools and environment Better understanding of dynamics / predictabilities etc.