You are on page 1of 7

EME3156

COMPUTATIONAL METHODS FOR MECHANICAL ENGINEERING

CHAPTER 1

1. Introduction

This chapter begins with the review of differential equations, physical applications of
the equations in various fields of study, the boundary and initial conditions, the
numerical methods in solving these equations, and the software packages.

1.1 Differential equations

Differential equations are equations involving the derivatives of functions. They are
broadly classified into two: ordinary and partial. Ordinary differential equations (ODE)
contain derivatives of a dependent variable with respect to one independent variable.
For example, if y is the dependent variable and x is the independent variable, an ODE
takes the form of Eq. (1.1):

(1.1)

Usually, integration is performed to obtain y as a function of x. The difficulty of the


integration depends on the function f x, y . Specifically, if f is only a function of x in
the form of a polynomial, the basic integration technique is adequate. For example:

(1.2)

In Eq. (1.2), C is a constant that can be obtained after applying the additional condition
of the problem. This additional condition is named the boundary condition or the initial
condition depending on the physical representation of the problem, and will be
elaborated in Sec. 1.3.

Eq. (1.1) is named the ______________________, because the derivative is of the first
order. There are higher orders ODE, like second order, third order, etc. The higher order
ODE may contain lower order of derivatives in the equation, For example,
Trimester 2, 2015/2016

EME3156

COMPUTATIONAL METHODS FOR MECHANICAL ENGINEERING

CHAPTER 1

y y x 2 y is the second order ODE containing the first derivative, and also f is
now a function of x and y.
When f x, y 0 , the ODE is named ________________________ ODE. For example,
y y 0 is the second order homogeneous ODE. Conversely, an ODE with

f x, y 0 is named non-homogeneous ODE.

When the index of the derivatives is not equal to one, the ODE is a ________________
ODE. For example,

y2 x 2 1 .

A non-linear ODE is also obtained when the

derivatives multiply with each other, for example y y x 2 1 .

On the other hand, a partial differential equation (PDE) contains the partial derivatives
of one dependent variable with respect to several independent variables. For example, in
Eq. (1.3), u is the dependent variable while x and y are the independent variables.

(1.3)

As in ODE, PDE are also classified into first order, second order, etc. Among the
second order PDE, there is another classification depending on the equation. To explain
this classification, Eq. (1.4) shows the general equation of a second order PDE.

(1.4)
When B 2 4 AC 0 , it is an elliptic PDE.
When B 2 4 AC 0 , it is a parabolic PDE.
When B 2 4 AC 0 , it is a hyperbolic PDE.

1.2 Physical applications of differential equations

In Eq. (1.1), the independent variable is denoted by the symbol x. If x represents time, it
is usually changed to t, thus the derivative is now written as _______, meaning the
Trimester 2, 2015/2016

EME3156

COMPUTATIONAL METHODS FOR MECHANICAL ENGINEERING

CHAPTER 1

change of variable y with respect to time. Specifically, if y represents velocity, it is


changed to v, thus the derivative is now written as dv dt , meaning the change of
velocity with respect to time, or simply acceleration. This derivative may appear in the
equation representing the fall of an object with air resistance: dv dt g kv , where g is
the acceleration due to gravity, and k is a constant related to air resistance. This equation
is solved to get the velocity of the object as a function of time.

In another example, if y represents the mass of matter, it is changed to m, thus the


derivative is now written as dm dt , meaning the change of mass with respect to time.
This derivative usually appears in the nuclear decay equation: dm dt km , where k is
a constant related to the rate of decay of a radioactive matter. This equation is solved to
get the mass of the matter as a function of time.

An application of the second order ODE is in a mass-damper-spring oscillation system,


written in the form: ___________________, where m is the mass of the system, c is the
damping coefficient and k is the spring stiffness. Since y represents the displacement of
the mass, thus y and y represents acceleration and velocity of the mass respectively.
This equation is solved to get the displacement of the mass as a function of time.

The first order PDE takes the form u x v y 0 , which represents the continuity
equation in a fluid flow. There are two dependent variables in this equation: u and v,
representing the x-component and y-component of the velocity respectively. Knowing
that u and v cannot be solved by just having one equation, another equation is required
to complement the problem, named the momentum equation (a second order PDE):
uu x vuy dp dx u xx u yy X . The solution of u and v are usually used to

solve

another

related

equation

named

the

energy

equation:

C uTx vTy k Txx Tyy X , where T is the temperature of the domain. These three

equations (continuity, momentum and energy) frequently appear in fluid flow and heat
transfer problems.

Trimester 2, 2015/2016

EME3156

COMPUTATIONAL METHODS FOR MECHANICAL ENGINEERING

CHAPTER 1

The second order PDE shown in Eq. (1.4) is the most common among all types of PDE,
thus its physical application will be discussed as follows, and the corresponding
numerical methods used to analyse this type of PDE will be explained in later chapters.

A physical application of the elliptic PDE is the two-dimensional steady state heat
conduction, and the PDE usually takes the form u xx u yy f , where u represents the
temperature and f is the heat generation. The same equation can also be used to
represent the torsion of a bar with a non-circular cross-section, where u now represents
the stress and f is the angle of twist. In both examples, x and y are the coordinate system
of the domain.

The parabolic PDE takes the form u xx ut , and typically appears in one-dimensional
transient heat conduction. Thus, x and t represent the coordinate system and time
respectively.

A hyperbolic PDE takes the form utt u xx , usually appears in one-dimensional wave
equation. As in the parabolic PDE, x and t represent the coordinate system and time
respectively. However, u now represents the displacement of the wave.

1.3 Boundary and initial conditions

As shown in Eq. (1.2), the solution of a differential equation is not complete without
stating the additional conditions. Mathematically speaking, the value of C must be
obtained to complete the solution process.

Generally, the number of conditions required to complete the solution process depends
on the order of the differential equation. For an ODE, the number of conditions is equal
to the order of the equation, meaning that first order ODE requires one condition,
second order two conditions, etc.

Trimester 2, 2015/2016

EME3156

COMPUTATIONAL METHODS FOR MECHANICAL ENGINEERING

CHAPTER 1

For first order ODE, when the independent variable is time, this additional condition is
usually specified at the beginning of the time (t = 0), hence it is also named the
_____________________ (IC). Since almost all first order ODE is having time as the
independent variable, thus a first order ODE is also named the initial-value problem.
Using Eq. (1.2) as an example, the complete formulation can now be written as follows:
dy
t 2 1 ; IC: t 0, y 2 or in short ___________
dt

(1.5)

A second order ODE needs two conditions to completely solve the problem. Usually,
these conditions are specified at the boundary of the problem (the beginning and the end
of the independent variable). Hence this type of problem is also named the boundary
value problem, and the conditions are named the ______________________. An
example of complete statement of the boundary value problem is given in Eq. (1.6),
where the boundary is located at x 0 and x 1 .

d 2 y dy

x2 y
2
dx
dx

(1.6)

BC: x 0, y 2 and x 1, y 10 or in short ______________________

1.4 The meaning of domain and boundary

Before specifying the boundary conditions, it is important to understand the physical


location and meaning of the boundary of the problem, so that the BC is completely and
correctly stated to represent the physical problem. The boundary relates closely with the
physical shape of the domain. The meaning of boundary will be explained in terms of
the shape of the domain.

From the shape point of view, since this is a three-dimensional world, all objects
generally have a three-dimensional shape. For example, a cube has width, height and
depth. The boundary of the cube (three-dimensional shape) is the exterior surfaces of
Trimester 2, 2015/2016

EME3156

COMPUTATIONAL METHODS FOR MECHANICAL ENGINEERING

CHAPTER 1

the cube. When one of the dimensions are negligible compared to the other two (like a
piece of paper), the shape is two-dimensional. In this case, the boundary of the paper is
the edge of the paper. If another dimension is reduced, it becomes a one-dimensional
shape, like a long string. The boundary of the string is the end points of the string.
Finally, there is the zero-dimension where the shape is simply one solitary point.

1.5 Non-dimensional form of differential equations

Consider a one-dimensional heat transfer problem in Eq. (1.7), where u is the


temperature with the unit K or C; x is the distance with unit metres; Q is a constant
related to heat source with unit K/m2.

(1.7)

d 2u
Q 0 , u0 0 , u2 100
dx 2

The solution of this problem is only valid for the specifications mentioned above. If the
domain is changed to five metres long, theoretically the solution procedure needs to be
repeated. However, there is a generalised method to solve this governing equation that
can represent a general solution for the same type of problem.

Firstly, these variables can be non-dimenionalised by multiplying with constants. For


example, the non-dimensional form of x can be achieved by dividing x with the length L
of the domain, and given a new symbol X. Similarly, the temperature u can also be nondimensionalised by dividing with a constant Tref representing a reference temperature,
thus a given a new symbol . Eq. (1.8) is the non-dimensional form of (1.7):

(1.8)

Trimester 2, 2015/2016

EME3156

COMPUTATIONAL METHODS FOR MECHANICAL ENGINEERING

CHAPTER 1

After solving for , the actual temperature u can be obtained using the relation in Eq.
(1.8). The non-dimensionalisation of governing equations can also be done for higher
order ODE and PDE.

1.6 Engineering software

There are numerous commercial software that solve engineering problems. As in any
computer program, the user interface and data entry must be user friendly and easy to
use, and concentrate on the presentation of the physics of the problem more than the
mathematical operations in solving the problem. Therefore, these software do not
usually require the user to specify the type of equations to be solved; rather the physical
representation of the problem (for example fluid flow, heat transfer, structural,
vibration, etc). After specifying the type of problem, the software enters the input mode
where the geometry, material properties, meshing, boundary or initial conditions are
inserted. In each of the steps above, the software runs background operations like
transforming the parameters into mathematical equations. Finally, the results are
presented in graphical form including the use of colours to represent the values of the
dependent variables.

The role of an engineer does not stop at running the software and printing the results.
An engineer should also discuss the results obtained in terms of the accuracy and the
reliability of the results. Ultimately, the engineer should judge if the problem has been
accurately solved and the results can be used for further improvements or taken as the
final values for design and fabrication purposes. These additional steps require technical
knowledge on the mathematical (or numerical) analysis of the software, or the
knowledge on the physics of the problem.

Trimester 2, 2015/2016

You might also like