MCE521 Advanced Computational Dynamics

Lecture 02

Dr. Ayokunle O. Balogun

1st College Building, FC003A
balogun.ayokunle@lmu.edu.ng

Class: Wednesday (8 – 10 am)

Venue: A04

MCE521: ADVANCED COMPUTATIONAL DYNAMICS

 Lecture content
Advanced Computational Dynamics
 Introduction – Practical Aspects of CFD
 Problem Setup – Pre-process
 Numerical Solution – CFD Solver
 Result Report and Visualization – Post-process

Recommended textbooks
1. Computational Fluid Dynamics: A Practical Approach by Tu J., Yeoh
G., and Liu C. 2nd Edition
2. Numerical Methods for Engineers by Chapra S., and Canale R.P. 5th
Edition
MCE521: ADVANCED COMPUTATIONAL DYNAMICS 2
INTRODUCTION – Practical Aspects of CFD
CFD Skill and Knowledge Acquisition
 Recent CFD codes have GUI that makes its study less intimidating
and more attractive to learning.
 In this practical sense CFD codes may be treated as blackboxes in
which a user learns to (i) Set-up fluid flow, heat transfer problems (ii)
Solve them numerically and (iii) Make a few graphical presentations.

 The theory of CFD is also very important because it forms the basis
for understanding the fundamental principles of transport (fluid
flow, heat transfer, exchange of chemical species) phenomena and
analyzing the computational solutions.

 In this course the CFD code of choice is COMSOL Multiphysics.

Generally, most CFD packages have user-friendly GUI applications
and environment for input of problem parameters and for examining
the computed results.

MCE521: ADVANCED COMPUTATIONAL DYNAMICS 3

INTRODUCTION – Practical Aspects of CFD
 The CFD code consist of three main elements (Fig 1)
1. Pre-processor
2. Solver
3. Post-processor
Shareware (Freeware) CFD
 Browse
www.cfd.online.com/Links/soft.
htm/ for a catalog of CFD codes
through the shareware, freeware
option link.

 A first-time user may visit

http://energy.concord.org/e
nergy2d/index.html
Fig 1: The interconnectivity functions of the three main elements within a CFD analysis framework

MCE521: ADVANCED COMPUTATIONAL DYNAMICS 4

INTRODUCTION – Practical aspects of CFD
Commercial CFD
 The software category of the web link
www.cfd.online.com/Links/soft.htm/ provides the list of commercial
codes that are currently available through the commercial option link.
 The table below presents the internet links for some popular CFD
codes
Table 1: Internet links to some popular commercial CFD packages

MCE521: ADVANCED COMPUTATIONAL DYNAMICS 5

PROBLEM SETUP – Pre-process
Creation of Geometry – Step 1
 Mostly the definition and creation of the geometry of the flow region,
i.e, the computational domain for the CFD calculations is the first
step (Fig 2).

 In the physical domain the

flow is 3-D but it is assumed
that flow is invariant in one
direction so it reduces to 2-D in
the computational domain.
 Fig 2 is analogous to fluid flow
in a circular pipe and it must
acquire a fully developed
regime as it exist the pipe.
Fig 2: Fluid flowing between two
stationary parallel plates

MCE521: ADVANCED COMPUTATIONAL DYNAMICS 6

PROBLEM SETUP – Pre-process
Mesh Generation – Step 2
 The computational domain is sub-divided into a number of
subdomains leading to the generation of a mesh (or grid) of cells
(elements or control volumes) over which the flow physics is solved
(Fig 3a).
 The essential fluid flows described in these cells are solved
numerically to obtain discrete values for the flow properties, such as
velocity, pressure, temperature, and other transport parameters of
interest, are determined.
 Generally, the accuracy of the CFD solution is influenced by the
number of cells in the mesh within the computational domain.

Fig 3a: Structured meshing for fluid flowing between two stationary parallel plates

MCE521: ADVANCED COMPUTATIONAL DYNAMICS 7

PROBLEM SETUP – Pre-process
Mesh Generation – Step 2
 Mesh generation entails either structured or unstructured cells. The
structured cells are usually used for relatively simpler geometries
and they generally consist a regular distribution of rectangular cells
(Fig 3b)
 For more complex geometries, and complicated shape boundaries
unstructured meshes (Fig 3b) allow for greater flexibility. It is also
possible to use a combination of both depending so as to attain a
more realistic simulation.

Fig 3b: Structured and unstructured meshing

MCE521: ADVANCED COMPUTATIONAL DYNAMICS 8

PROBLEM SETUP – Pre-process
Selection of Physics and Fluid Properties – Step 3
 It absolutely important that a CFD user carefully identifies the
underlying flow physics unique to the particular fluid-flow system.

 The various flow physics

that may be encountered in
CFD and heat transfer
processes are as shown in
Fig 4.

Fig 4: A flowchart encapsulating the

various flow physics in CFD

MCE521: ADVANCED COMPUTATIONAL DYNAMICS 9

ADVANCED COMPUTATIONAL DYNAMICS
Problem Setup – Pre-process
Specification of Boundary Conditions – Step 4
 A CFD user needs to define the appropriate boundary conditions that mimic
the real physical representation of the fluid flow in a solvable CFD problem.

 The different types of boundary conditions for a simple flow problems is

shown in Fig 6

Fig 6: Definition of symmetry Fig 5: Boundary conditions for

boundary condition an external flow problem

MCE521: ADVANCED COMPUTATIONAL DYNAMICS 10

NUMERICAL SOLUTION - CFD Solver
 The prerequisite processes in the solution procedure that have
implications for the computational solution are initialization,
solution control, monitoring solution, CFD calculation, and
checking for convergence.
 An understanding of the
prerequisite processes highlighted
above are necessary in order to
skillfully utilize the many solver
features and better navigate the
underlying “black box” operations
that resides in these codes.

Fig 7: An overview of the solution procedure

MCE521: ADVANCED COMPUTATIONAL DYNAMICS 11

NUMERICAL SOLUTION – CFD Solver
Initialization and Solution Control - Step 5
 Firstly, for an iterative solution, the discrete values of the flow
properties such as velocity, pressure, temperature and other transport
parameters of interest has to initialized before calculating a solution.
The appropriate selection of initial conditions (ICs) is important
because
 the closer the ICs are to the final steady-state solution, the quicker the
iterative procedure converges and yields results in a shorter
computational time.
 the farther away the ICs are from realty, the longer the computational
efforts and it may even lead to the iterative procedure misbehaving
or diverging.
 Secondly, in the solution control the appropriate parameters are
setup and this entails the specification of appropriate discretization
(interpolation) schemes and selection of suitable iterative solvers.

MCE521: ADVANCED COMPUTATIONAL DYNAMICS 12

NUMERICAL SOLUTION CFD Solver
Monitoring Convergence - Step 6
 This step involves the interlink of these three prerequisite operations:
monitoring solution, CFD calculation, and checking for convergence.
Convergence can be monitored by
1) imbalances, also known as residuals, which measure the overall
conservation of the flow properties. They can be viewed from the
commercial code GUIs.
 A converged solution is achieved when the residuals fall below some
convergence criteria or tolerance that is preset inside the solver-
controlling parameters of the iterative solvers (Fig 8).
2) monitoring variables such as lift, drag, or moment force to ascertain
the convergence of the numerical computations.
 Progress toward a converged solution can be greatly assisted by the
careful selection of various under-relaxation factors. These enhance
the stability of the numerical procedure and the convergence of the
iterative process.
MCE521: ADVANCED COMPUTATIONAL DYNAMICS 13
RESULT REPORT AND VISUALIZATION – Post-process
 Most of the commercial CFD codes have impressive visualization
tools that allow users to graphically view the results of CFD
calculation of a computational simulation.
X-Y plots
 These are mainly 2-D graphs that
represent the variation of one
dependent transport variable as
compared with another,
independent variable (Fig 9).
 These graphs represent a
popular way of directly
comparing the numerical data
with the experimentally
measured values.
Fig 9: X-Y plot of a parabolic laminar profile at the fully
developed region for flow through two-stationary plates.

MCE521: ADVANCED COMPUTATIONAL DYNAMICS 14

RESULT REPORT AND VISUALIZATION – Post-process
Vector plots
 A vector plot provides the means whereby a vector quantity (usually
velocity) is displayed at discrete points by an arrow, whose
orientation indicates direction and whose size indicates magnitude
(Fig 10).

Fig 10: Velocity vectors showing the flow development along the
parallel-plate channel for flow through two-stationary plates.

 The vector plot (Fig 10) also depicts the composite association of the
velocity vectors and distribution of dynamic pressure within the
flow domain that propels the fluid flow.

MCE521: ADVANCED COMPUTATIONAL DYNAMICS 15

RESULT REPORT AND VISUALIZATION – Post-process
Contour Plots
 A contour plot consist of contour lines also called isolines which are
described as lines indicative of some property that is constant in
space. The equivalent representation in three dimensions is an
isosurface.

Fig 11: Flooded contours on a rainbow – scale color map for the
normalized velocity of flow through two-stationary plates.

 Fig 11 is an example of flooded contour plots showing the

normalized velocity the flow through the two-stationary plates.
Animation
 CFD data fit very nicely into animation - moving pictures of the data
produced through the CFD simulation.
MCE521: ADVANCED COMPUTATIONAL DYNAMICS 16
ADVANCED COMPUTATIONAL DYNAMICS
REVIEW QUESTIONS
1. What are the main elements involved in a complete CFD analysis?
2. What is the main purpose of a CFD solver?
3. What is the consequence of using a very fine mesh (i.e., a very large
number of cells) compared with using a coarse mesh (i.e., a small
number of cells).
4. What is the main difference between structured and unstructured
mesh and when is each type applied to physical domains?
5. What types of boundary conditions can be imposed on the
computational domain?
6. What are the advantages of providing intelligent values for the
initial solution?

MCE521: ADVANCED COMPUTATIONAL DYNAMICS 17

ADVANCED COMPUTATIONAL DYNAMICS
DEMONSTRATION
6. Consider the base plate of a 800-W household iron with
a thickness of L = 0.6 cm, base area of 160 cm2, and
thermal conductivity of k = 20 W/m. C. The inner
surface of the base plate is subjected to uniform heat
flux generated by the resistance heaters inside. When
steady operating conditions are reached, the outer
surface temperature of the plate is measured to be 85 C.
Disregarding any heat loss through the upper part of
the iron, (a) express the differential equation and the
boundary conditions for steady one-dimensional heat
conduction through the plate, (b) obtain a relation for
the variation of temperature in the base of the plate by
solving the differential equation, and (c) evaluate the
surface inner temperature

MCE521: ADVANCED COMPUTATIONAL DYNAMICS 18

 Fig 8: A Typical ANSYS-CFX graphical user interface

MCE521: ADVANCED COMPUTATIONAL DYNAMICS 19

 Fig 8: Typical ANSYS-CFX GUIs for monitoring convergence
corresponding to the prescribed convergence criteria.

MCE521: ADVANCED COMPUTATIONAL DYNAMICS 20