Modelling

CHAPTER 4: MODELLING OF SPRAY DEPOSIT
4
MODELLING OF SPRAY
DEPOSIT
4.1 FINITE ELEMENT ANALYSIS
The finite element method is a numerical procedure, that can be used to obtain
solutions to a large class of engineering systems, including stress analysis, heat
transfer, fluid flow and electromagnetism. In thermal spraying, finite element analysis
is used either to validate or predict experimental results, through numerical
formulation.
Residual stress is a major problem in thermal spraying where high
thickness coatings are required, due to the rapid solidification and cooling of sprayed
droplets during cooling. Experimental analysis may help the researcher understand
where these stresses arise in a component, hence lead to the prevention of such
outcomes. Finite element analysis may support experimental findings and predict
results for more complex situations. In general most of the analyses of residual stress,
consist of analytical models of the behaviour of molten droplets striking the substrate
surface, similar to that detailed by Fukanuma et al. [101], nevertheless finite element
analysis similar to that shown here has also been documented [102,103]. In this
research finite element analysis of heat transfer and residual stress in the coated sample
is carried out and compared to experimental data generated during the current research.
There are many finite element method software programs available to yield various
engineering solutions, however the ANSYS finite element program was used in the
current research, as it is widely available within the university. Hence the followings
sections are specific to the ANSYS program.
The origin of the modern finite element method may be traced back to the early 1900s,
when some investigators approximated and modelled elastic continua using discrete
equivalent elastic bars. The ANSYS finite element method program was released in
________________________________________________________________________________________________
Production of Coated and Free-Standing Engineering Components Using the HVOF (High Velocity Oxy-Fuel) Process
J. Stokes
91

1971 for the first time. ANSYS is a comprehensive general purpose finite element
computer program, which contains over 100,000 lines of code [104]. ANSYS is
capable of performing static, dynamic, heat transfer, fluid flow, and electromagnetism
analyses.
The current version of ANSYS has multiple windows incorporating a
Graphical User Interface (GUI), pull-down menus, dialog boxes, and a tool bar. The
objective of this section is to introduce the basic concepts in finite element formulation,
including numerical formulation. The following topics are addressed:
Engineering Problems
Numerical Methods
Steps in the Finite Element Method
Numerical Formulation of systems specific to this research
Verification of Results
________________________________________________________________________________________________
J. Stokes
92
4.2 ENGINEERING PROBLEMS

Engineering problems can be described in general as being mathematical models of
physical situations [105]. The mathematical models generally comprise of numerous
differential equations with sets of corresponding initial and boundary conditions. The
differential equations are derived by applying fundamental laws and principles of
nature to an engineering system. These equations represent the balance of mass, force,
or energy and when possible, the solution of these equations renders a detailed
behaviour of a system under a given set of conditions.
Analytical solutions show the exact behaviour of a system at any point within the
system. An analytical solution may be composed of two parts: firstly a homogenous
part and secondly a particular part. In any engineering system, there are two sets of
parameters that influence the way a system behaves. Firstly, there are those parameters
that provide information regarding the natural behaviour of a given system and always
appear in the homogeneous part of the solution. Examples of these parameters include
thermal conductivity, modulus of elasticity, and viscosity properties of a material, as
shown in table 6. On the other hand, there are parameters that produce disturbances in
a system and they appear in the particular part of the solution. Examples of disturbing
parameters include temperature difference across a medium, external forces, moments,
and pressure difference in a fluid flow, as shown in table 6.
________________________________________________________________________________________________
J. Stokes
93

Table 6,
Physical properties characterising and disturbing various engineering

systems.
Engineering Problem
Heat Transfer
Parameters that
Parameters that Disturb
Characterise a Problem
a Problem
Thermal Conductivity
Heat Input
Temperature
Difference
Solid Mechanics
Modulus of Elasticity
External Forces
Modulus of Rigidity
Moments
Second Moment of
Support Excitation
Area
Polar Moment of
Inertia
Fluid Flow
Electrical Networks
Viscosity
Pressure Difference
Relative Roughness
Rate of Flow
Solid Permeability
Resistance
Voltage Difference
Permeability
________________________________________________________________________________________________
J. Stokes
94
4.3 NUMERICAL METHODS

Many practical engineering problems can only be solved approximately. This inability
to obtain an exact solution may be attributed to either the complex nature of the
governing differential equations or the difficulties that arise from dealing with initial
and boundary conditions [105]. To deal with such problems, numerical approximations
are used. In contrast to analytical solutions, which show the exact behaviour of a
system at any point within the system, numerical solutions approximate exact solutions
only at discrete points. The first step in the numerical procedure is to discretize
(divide) a system into small subsystems known as elements, where their shape is
described by discrete points known as nodes.
There are two types of numerical methods, finite difference methods and finite element
methods. With finite difference methods, the differential equation is written at each
discrete point (node), and the derivatives are replaced by difference equations, this
approach results in a set of simultaneous linear equations [106]. Finite difference
methods are easy to understand in simple systems, however they become difficult to
apply to systems with complex geometries or with complex boundary conditions, an
example of this would be a system involving non-isotropic material properties.
The finite element method use integral formulations, rather than difference equations,
to create a system of algebraic equations.
Moreover, an approximate continuous
function is assumed to represent the solution for each element. The complete solution
is generated by connecting or assembling the individual solutions, allowing for
continuity at the inter-elemental boundaries.
________________________________________________________________________________________________
J. Stokes
95
4.4 STEPS IN THE FINITE ELEMENT METHOD

The basics steps involved in any finite element analysis consist of the following:
Preprocessor Phase
Create and discretize the solution domain into finite elements, that is the
system is sub-divided into elements and nodes
Assume a shape function to represent the physical behaviour of an element.
Develop equations for an element
Arrange and assemble the elements to present the entire system. Construct
the global stiffness matrix
Apply boundary conditions, initial conditions and loads
Solution Phase
Solve a set of linear or non-linear algebraic equations simultaneously to

obtain nodal results, such as displacement values at different nodes or
temperature values at different nodes in a heat transfer system
Postprocessor Phase
Obtain other important information including stress values, heat fluxes and
so on
________________________________________________________________________________________________
J. Stokes
96
4.5 NUMERICAL FORMULATION

The following examples illustrate the simplified one-dimensional direct formulation
steps used to analyse both heat transfer and residual stress systems experienced in the
current research, with reference to Fagan [105] and Moaveni [106].
4.5.1 Heat Transfer System

A sample consists of a stainless steel substrate, coated with aluminium (releasing
layer), on which a coating of tungsten carbide-cobalt is placed.
The following
describes the determination of the temperature distribution through the three materials,
when a heat source is placed near the tungsten carbide-cobalt surface.
Preprocessing Phase
Discretize the solution domain into finite elements.
The system is represented by six nodes (discrete points) and five elements (subsystems) as shown in figure 35.
Tflame
1
DJ
Gun
(1
Element
1
(2
Element
2
Substrate
Aluminium
WC-Co
(3
Element
3
(4
Element
4
Back of
Substrate
T
(5
6 air
Element
5
Figure 35, Finite Element model for temperature distribution through various
materials.
________________________________________________________________________________________________
J. Stokes
97
Assume a solution that approximates the behaviour of an element.

In this situation, there are two modes of heat transfer (conduction and convection),
both of which must be understood before formulation of the conductance matrix and
the thermal load matrix. Is must be noted that the solution ignores the fact that heat
loss will be experienced around the sides of the sample, hence an approximation of the
real solution will be found.
The steady state thermal behaviour of the elements (2), (3), and (4) may be modelled
using Fouriers Law. When there is a temperature gradient in a medium, conduction
heat transfer occurs, as shown in figure 36. The energy is transported from the hightemperature region to the low-temperature region by molecular activity.
The heat transfer rate is given by Fouriers Law:
qx = -kA
T
X
Equation 32
qx is the X-component of the heat transfer rate, k is the thermal conductivity of the
medium, A is the area, and
T
is the temperature gradient. The minus sign is due to
X
the fact that heat flows in the direction of decreasing temperature.
Ti
qx
Ti+1
X
Figure 36, Heat transfer by conduction through a medium.
________________________________________________________________________________________________
J. Stokes
98
Equation 32 can be written in terms of the spacing between the nodes (length of the
element) L and the respective temperatures of the nodes i and i+1, Ti and Ti+1 by the
following [107]:
kA(Ti Ti +1 )
L
q=
Equation 33
The steady state behaviour of elements (1) and (5) may be modelled using Newtons
Law of Cooling. Convection heat transfer occurs when a fluid in motion comes into
contact with a surface whose temperature differs from the moving fluid. The overall
heat transfer rate between the fluid and the surface is governed by Newtons Law of
Cooling [108], by the following equation:
q = hA(Tf Ts)
Equation 34
where h is the heat transfer coefficient (W/m2K), Ts is the surface temperature, and Tf
represents the temperature of the moving fluid (gas). According to Holman [108],
under steady state conduction, the application of energy balance to a surface requires
that the energy transferred to a surface by conduction must be equal to the energy
transferred by convection. Hence,
- kA
T
= hA[Tf Ts]
X
Equation 35
as shown in figure 37.
h (for convection) is given by:
h =
k
L
Equation 36
________________________________________________________________________________________________
J. Stokes
99

therefore:
kA
kA
T
=
(T)
X
L
Ti =Ts
qconvection
Equation 37
qconduction = -kA
T
X
qconvection = hA[Tf Ts]
qconduction
Ti+1
Tf , h
Figure 37, Energy balance at a surface with convective heat transfer.
Develop equations for an element.

The heat flow through nodes i and i+1 is depicted in figure 38. As each element in the
coated substrate System has two nodes, and has associated temperatures with each
node, two equations can be created for each element. These equations must involve
nodal temperatures and the elements thermal conductivity based on Fouriers law.
Under steady state conditions, the application of the conservation of energy requires
that the sum of qi and qi+1 into an element be zero, that is the energy flowing out of
node i+1 must be equal to the energy flowing into node i. There exists two situations
encountered at an elemental position. One, the element may experience a heat gain at
one side and a heat loss at the other (figure 38 a), or secondly the element may
experience a heat gain at both sides (figure 38 b). Note that the sum of qi and qi+1 is
zero regardless of which representation of figure 38, is selected. The following analysis
________________________________________________________________________________________________
J. Stokes
100

is based on the representation shown in figure 38 (b). The heat transfer rates, qi and
qi+1, and the nodal temperatures (discrete points), Ti and Ti+1, for an element (subsystem) are related according to the equations:
qi =
qi+1 =
kA
(Ti Ti+1)
L
Equation 38
kA
(Ti+1 Ti)
L
Equation 39
Ti
qi =
kA
(Ti Ti+1)
L
qi+1 =
kA
(Ti Ti+1)
L
Ti+1
X
Ti
qi =
kA
(Ti Ti+1)
L
qi+1 =
kA
(Ti+1 Ti)
L
Ti+1
X
Figure 38, Heat flow through nodes i and i+1 showing (a) heat gained at one
side and lost at the other and (b) heat gained at both sides.
________________________________________________________________________________________________
J. Stokes
101
The elemental description given by equations 38 and 39 may be expressed in matrix

form as follows:
qi kA 1 1 Ti
=

L 1 1 Ti +1
qi +1
Equation 40
The thermal conductance matrix for the element shown in figure 38 (b), is given as:
[K ](e ) =
kA
L
1 1
1 1
Equation 41
Assemble the elements to present the entire system.

Using equation 41 the position matrix for the first element is given by:
k1
[K ](1) = A Lk1
1
L1
k1
L1
k1
L1
Equation 42
and its position in the global conductance matrix (overall matrix to describe the total
system) is given by:
k1
L
1
k1
L
[K ](1G ) = A 0 1
0
0
k1
L1
k1
L1
0
0
0
0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
T1
T2
T3
T4
T5
T6
Equation 43
________________________________________________________________________________________________
J. Stokes
102
Similar to element 1, the elemental position matrices together with their position in the
global conductance matrix for elements 2, 3, 4 and 5 are given as follows:
k2
[K ](2 ) = A Lk2
2
L2
k2
L2
k2
L2
and
0
0
k2
0 L
2
k
2
[K ](2G ) = A 0 L2
0
0
0
0
0
0
0
k
2
L2
k2
L2
0
0
0
0 0 0
T1
0 0 0
T
2
0 0 0 T3
T
4
0 0 0
T5
0 0 0
T6
0 0 0
Equation 44
k3
[K ](3) = A Lk3
3
L3
k3
L3
k3
L3
and
0
0
0
( 3G )
[K ] = A
0
0
0
0
0
0
0
0
0
0
0
k3
L3
k
3
L3
0
0
0 0
T
0 0 1
T2
0 0
T3
0 0 T4
T5
0 0
T6
0 0
0
0
k
3
L3
k3
L3
0
0
Equation 45
k4
[K ](4 ) = A Lk4
4
L4
k
4
L4
k4
L4
and
0
0
0
(4G )
[K ] = A 0
0 0
0 0
0 0
0
k4
0 0
L4
k
0 0 4
L4
0 0
0
0
k
4
L4
k4
L4
0
0
T
0 1
T
0 2
T3
0
T
4
0 T5
T6
0
Equation 46
________________________________________________________________________________________________
J. Stokes
103
k5
[K ](5 ) = A Lk5
5
L5
k5
L5
k5
L5
0
0
[K ](5G ) = A 0
0
and
0 0 0
0 0 0
0 0 0
0 0 0
0
k5
0 0 0
L5
k
0 0 0 5
L5
0
0
0
0
k
5
L5
k5
L5
T1
T2
T3
T4
T5
T6
Equation 47
The overall global conductance matrix (matrix for the system) is given as the sum of
each elemental position:
[K](G) = [K](1G) + [K](2G) + [K](3G) + [K](4G) + [K](5G)
Equation 48
Hence the overall global conductance matrix for the coated substrate system shown in
figure 35 is given as follows:
k1
L
1
k1
L1
0
(G )
[K ] = A
0
k1
L1
k1 k 2
+
L1 L2
k
2
L2
0
k2
L2
k 2 k3
+
L2 L3
k
3
L3
k3
L3
k3 k4
+
L3 L4
k
4
L4
0
k4
L4
k 4 k5
+
L4 L5
k
5
L5
0
Equation 49
k5

L5
k5
L5
________________________________________________________________________________________________
J. Stokes
104
Apply Boundary Conditions and Thermal Loads

Finite element formulation of heat transfer systems, always lead to an equation of the
form [108]:
[K]{T} = {q}
Equation 50
[Conductance Matrix]{Temperature Matrix}={Heat Flow Matrix}
Equation 51
In this system, the WC-Co coating is exposed to a known flame temperature T1 =

500oC (first boundary condition), and the air temperature behind the substrate T6 =
50oC (second boundary condition).
In the case the overall matrix equation is
represented as follows:
k1
L
1
k1
L1
0
A
0
k1
L1
k1 k 2
+
L1 L2
k
2
L2
0
k2
L2
k 2 k3
+
L2 L3
k
3
L3
k3
L3
k3 k4
+
L3 L4
k
4
L4
0
k4
L4
k 4 k5
+
L4 L5
k
5
L5
k5

L5
k5
L5
T1
T
2
T3
=
T4
T5

T6
500 o C
0
0
0
0
o
50 C
Equation 52
As matrices are a representation of a series of simultaneous equations, the equation for

the first row and the final row should read as follows:
T1 = 500oC
Equation 53
T6 = 50oC
Equation 54
________________________________________________________________________________________________
J. Stokes
105

Therefore rows 1 and 6 are adjusted (by removing the
k
constants) in the global
L
conductance matrix, to yield the results shown equations 53 and 54 as follows:

1
A
k
1
L1
0
A
0
k1 k 2
+
L1 L2
k
2
L2
k2
L2
k 2 k3
+
L2 L3
k
3
L3
0
0
k3
L3
k3 k4
+
L3 L4
k
4
L4
k4
L4
k 4 k5
+
L4 L5
k
5
L5
1
T1
T
2
T3
=
T4
T5

T6
500 o C
0
0
0
0
o
50 C
Equation 55
Row two is as follows when multiplied out:
Ak1
Ak1
Ak2
Ak2
T1 +
T2 +
T2 T3 = 0
L1
L1
L2
L2
Equation 56
By equation 53:
Ak1
Ak2
Ak2
k
T2 +
T2 T3 = A 1 500
L1
L2
L2
L1
Equation 57
Similarly row five reduces to the following:
Ak4
Ak4
k
k
T4 +
T5 + 5 T5 = A 5 50
L5
L5
L4
L4
Equation 58
________________________________________________________________________________________________
J. Stokes
106
Ignoring the equations involved in row 1 and 6, and incorporating the known boundary
conditions into rows 2 and 5 of the heat transfer system, equation 55 is reduced to:
k1 k 2
L + L
2
1
k
2

L2
A
0
k2
L2
k 2 k3
+
L2 L3
k
3
L3
T
2
0
T3
.
=
k
4 T4
L4 T5
k 4 k5
+
L4 L5
0
k3
L3
k3 k4
+
L3 L4
k
4
L4
k1
L 500
1
0
k 5 50
L5
Equation 59
Substituting the conductivity k values, length L (taken as the heat effected zone
distance for the flame or the thickness of the material) and exposure area (A) values
(latter two taken from the experimental data in the current research, see figure 39), into
equation 59, results in the calculation of the various elemental temperatures (T2-5). The
data shown in table 7 represents a real heat transfer situation:
L3
L4
Stainless Steel
Aluminium
WC-Co
L1
Area A
L2
Figure 39, Schematic of the heat transfer system.

________________________________________________________________________________________________
J. Stokes
107
Table 7,
Heat transfer parameters found from experimental data.

Item
(1) Flame (500oC)
Conductivity k
Length L
k/L
Area A
(W/mK)
(m) x 10-3
(W/m2K)
(m2) x 10-3
0.058
20
3.0
1.6
45000
1.6
0.75
166666.7
1.6
4166.7
1.6
20
1.5
1.6
Ref. [108]
(2) WC-Co
90
Ref. [33]
(3) Aluminium
125
Ref. [109]
(4) Stainless Steel
25
Ref. [109]
(5) Air Behind

Substrate (50oC)
0.03
Ref. [108]
According to table 7;
45000
0
0
3 + 45000
45000 45000 + 166666.7
166666.7
0
.
A.
0
166666.7
166666.7 + 4166.7
4166.7
0
0
4166.7
4166.7 + 1.5
3(500)
0
0
1.5(50)
T2
T
3

T4
T5
Equation 60
Simplifying, the following result is obtained;

45000
0
0
45003
45000 211666.7 166666.7
A
0
166666.7 170833.4 4166.7
0
4166.7
4168.2
0
T2
T
3
=
T4
T5
1500
0
Equation 61
0
75
________________________________________________________________________________________________
J. Stokes
108

In the present research, convection from the flame at the front and convection with the
air at the back of the sample was ignored. Experimental data provided temperatures
both at the front and back of the sample, hence convection was not required. The
system use is shown in figure 40.
WC-Co
Tfront
(1
Element
1
Aluminium
(2
Back of
Substrate
Substrate
(3
Element
2
Element
3
4
Tback
Figure 40, Finite Element model used in the present research.
Equation 52 in this case is reduced to:

k1
L
1
k1
L
A 1
0
k1
L1
k1 k2
+
L1 L2
k
2
L2
k2
L2
k 2 k3
+
L2 L3
k
3
L3
0
T
1
0
T2
.
=
k3 T3

L3 T4
k3
L3
500
0

Equation 62
0
0
Where 500 is the only known front temperature. Substituting the conductivity k values,
length L (taken as the heat effected zone distance for the flame or the thickness of the
material) and exposure area (A) values (from table 8) into equation 62, results in the
calculation of the various elemental temperatures (T2-4).
________________________________________________________________________________________________
J. Stokes
109

Table 8,
Parameters used in the heat transfer finite element analysis.

Item
Conductivity k
Length L
k/L
Area A
(W/mK)
(m) x 10-3
(W/m2K)
(m2) x 10-3
90
45000
1.6
0.75
166666.7
1.6
4166.7
1.6
(1) WC-Co
Ref. [33]
(2) Aluminium
125
Ref. [109]
(3) Stainless Steel
25
Ref. [109]
45000
0
0
45000
T1
45000 211666.7 166666.7
T
0
. 2 =
A
0
166666.7 170833.4 4166.7 T3
0
4166.7
4166.7 T4
0
500
0

Equation 63
0
0
Solution Phase
Solve a system of algebraic equations simultaneously.

Solving the previous matrix (equation 63) yields the temperature distribution across the
sample:
T1
T
2
=
T3
T4
500
499.99
C
499
.
98
499.98
Equation 64
The results show that the spraying temperature (front temperature) of 500oC, is
reflected in a temperature of 357.8oC, across the substrate (back temperature).
Postprocessing Phase
Obtain other Information.

The solution (equation 64) may be represented as a temperature contour of the thermal
gradient through the sample.
________________________________________________________________________________________________
J. Stokes
110
4.5.2 Residual Stress Determination

A substrate material (stainless steel) is coated with WC-Co to a certain thickness.
During deposition, stresses (quenching of lamella and cooling of coating) generate,
generating a moment at the ends of the sample, causing the sample to deflect, as shown
in figure 41.
Heat
Coating
Heat
Moment
Substrate
Deflection
Figure 41, Moment in a sample generating a deflection.
As explained in an earlier section, the simulation of both the quenching and cooling
stresses in one system is quite difficult, hence the method of simulation used in the
present study relied on the deformation of the final sample post-spraying. To
demonstrate the numerical formulation for this system, two simple approximation
methods were analysed. One method was to apply various temperatures to a coated
sample of known thickness and then measure the stress through the tungsten carbidecobalt deposit when the resulting deflection measured in the finite element analysis,
equalled that found experimentally in the Clynes Method. In this case the applied
temperature in the finite element system generated a thermal load in the centre of the
beam, and caused the beam to deflect, similar to that used by Steffens et al. [102].
Appendix 7 shows the procedure used to simulate this system.
________________________________________________________________________________________________
J. Stokes
111
Similarly this system was simulated by applying a known deflection to the sample
(Appendix 7) and compared to the latter. The system was simulated by locking both
ends of the sample while subjecting the centre of the sample to a thermal load P (to
generate the deflection), as shown in figure 42. The load and in turn deflection creates
an internal stress in the deposit causing permanent bending in the sample. The elastic
modulus for each material is given as Es (substrate) and Ec (coating). The sample has a
width W, length L, and thicknesses of ts (substrate), and tc (coating). The numerical
formulation for the latter system will be described in the following section.
L
W
Coating
ts
Substrate
tc
P
Fixed Point
Figure 42, Coated sample subjected to a central displacement.
Preprocessing Phase
Discretize the solution domain into finite elements.

In order to highlight the basic steps in the finite element analysis, the coated sample is
discretize into simple parts. The sample has been divided into four individual segments
called elements, with each segment having a uniform cross-section as shown in figure
43. Each element is represented as an elastic beam with nodes top and bottom of the
element and in turn each element is joined to another element by connecting one node
to the next. The accuracy of the results may be increased, by increasing the number of
additional nodes and elements.
________________________________________________________________________________________________
J. Stokes
112
4
2
3
1
P
u5
5
Element 4
u4
4
Element 3
u3
3
Element 2
u2
Elastic
Beam
Element 1
u1
P
Figure 43, Sub-dividing the sample into elements and nodes.
Assume a solution that approximates the behaviour of an element.

In order to study the behaviour of a typical element, figure 44 shows a solid member
with a uniform cross-section A of a length l subjected to a force F.
________________________________________________________________________________________________
J. Stokes
113
kequivalent
x
Figure 44, A solid member of uniform cross-section subjected to a force F.
The average stress in the member is given by:

=
F
A
Equation 65
The average normal strain of the member is defined as the change in length l per
unit length l of the member:
=
l
l
Equation 66
Over the elastic region, the stress and strain are related by Hookes Law, according to
the following equation,
= E
Equation 67
where E is the modulus of elasticity of the material. Combining these equations results
in:
AE
F=
l
l
Equation 68
________________________________________________________________________________________________
J. Stokes
114

Equation 68 is similar to the equation for a linear spring, F = kx. Therefore, a centrally
loaded member of uniform cross-section may be modelled as a spring with equivalent
stiffness of
keq =
AE
l
Equation 69
The coated sample is modelled as a series of centrally loaded members with equal
cross-sections. Thus, the sample is represented by a model consisting of four elastic
springs (elements) in series, and the elastic behaviour of each element is modelled by
an equivalent linear spring according to the equation:
f = keq (ui+1 ui) =
AE
(ui+1 ui)
l
Equation 70
ui and ui+1 are the displacements of the member at nodes i and i+1 respectively, and l is
the length of the element.
Develop equations for an element.

As each element has two nodes, and with each node there is an associated
displacement, then the system requires two equations for each element.
These
equations must involve nodal displacements and the elements stiffness. The elemental
relationship between the internally transmitting forces (fi and fi+1), and the end
displacements (ui and ui+1) is shown in figure 45.
________________________________________________________________________________________________
J. Stokes
115
fi+1 = keq (ui+1 ui)
Node i+1
Node i+1
y
or
Node i
Node i
ui u
i+1
fi = keq (ui+1 ui)
fi = keq (ui ui+1)
ui
Figure 45, Internally transmitted forces through an arbitrary element.
Static equilibrium conditions require that the sum of fi and fi+1 be equal to zero. In
figure 45, there are two ways (tension or two forces acting in the same directions) to
represent the transmission of forces acting on a member. The representation shown in
figure 45 (b), will be observed, so that the directions of fi and fi+1 are in the same
direction as y. Thus the transmitted forces at nodes i and i+1 are as follows:
fi = keq (ui ui+1)
Equation 71
Equation 72
Equations 72 may be expressed in matrix form as follows:

f i keq
=
f i +1 keq
keq ui

keq ui +1
Equation 73
________________________________________________________________________________________________
J. Stokes
116
Assemble the elements to present the entire system.

Applying the elemental description given in equation 73, to all elements, and
assembling them will lead to the formation of an elemental matrix.
[K ](1) =
k1 k1
k
1 k1
Equation 74
and its position in the global stiffness matrix is given by:
k1 k1
k
1 k1
(1G )
[K ] = 0 0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Equation 75
The nodal displacement matrix is shown alongside the position of element 1 in the
global stiffness matrix to observe its contribution of a node to its neighbouring
elements. Similarly for elements 2, 3 and 4, the following equations apply:
[K ](2 ) =
k2
k
2
k2
and
k 2
0 0
0 k
2
[K ](2G ) = 0 k2
0 0
0 0
0
k2
k2
0
0
0
0
0
0
0
0
0
0
0
0
Equation 76
________________________________________________________________________________________________
J. Stokes
117
[K ](3) =
k3
k3
k3
k3
and
0
0
[K ](3G ) = 0
0
0
0 0
0 0
0 k3
0 k3
0 0
0
0
k3
k3
0
0
0
0
0
0
Equation 77
[K ]
(4 )
k
= 4
k4
k4
and
k 4
0
0
[K ](4G ) = 0
0
0
0 0
0 0
0 0
0 0
k4
0 0 k4
0
0
0
k4
k4
Equation 78
The final (overall) global stiffness matrix is obtained by assembling, or adding together
each elements position in the global stiffness matrix:
[K](G) = [K](1G) + [K](2G) + [K](3G) + [K](4G)
k1
k1
k k + k
1 1 2
(G )
[K ] = 0 k2
0
0
0
0
k2
k 2 + k3
0
k3
k3
k3 + k 4
k4
0
0
0
k4
k 4
Equation 79
Equation 80
Apply Boundary Conditions and Loads

In this system, the sample is held at both ends while a load P is applied to the centre of
the sample. Applying this condition (adopting equation 73) results in the following set
of linear equations:
________________________________________________________________________________________________
J. Stokes
118

k1
k1
k k + k
1 1 2
0
k2
0
0
0
0
k2
k 2 + k3
k3
k3
k3 + k 4
k4
0 u1
P
0
0 u2

0 . u3 = 0
0
k 4 u4

0
k 4 u5
Equation 81
In solid mechanics systems, the finite element formulation always lead to a general
equation of the form [106]:
[K = Stiffness matrix]{u = Displacement matrix} = {F= Load matrix}
Equation 82
The solution to this system cannot be calculated, as applying a force to a fixed point
will cause no displacement, as explained in figure 46. Where one element (that is
element 1 in figure 45 a) occupies length of the sample (as this is a one-dimensional
system), applying the boundary condition (u1 = 0) to this element prevents the force P
from impacting elements 2, 3, and 4. The numerical matrix formulation depicts a
simple one-dimensional analysis however this problem was performed as a twodimensional system (or higher orders) in the current research, hence the elemental
distribution would look more like that shown in figure 45 b. Hence the boundary
condition (u1 = 0) was applied to the lower corners elements of the beam. This
formulation of the system would allow the centre of the sample to deflect. This
highlights the inaccuracy found in results, when using a coarse (fewer elements) mesh.
________________________________________________________________________________________________
J. Stokes
119
Deflection
No Deflection
P
(a) Coarse Mesh
4
Elements
P
(b) Fine Mesh
=> Boundary Condition (u1 = 0)
Figure 46, Applying boundary conditions to coarse and fine meshed problems.
The residual stress determination was carried out as a two or in some cases a three
dimensional system, however expressing the numerical formulation for a 2-D or 3-D
system, would be too lengthy. Expressing the residual stress as an one-dimensional
system provides sufficient information as to the mechanism of stress determination in
deflected bimetallic beam. The analysis is based upon applying a load to one side of
the system (at node 1), while fixing the opposite side of the system (at node 5), as
shown in figure 47.
Using equation 81, adjusting its stiffness matrix to ensure that the equation in row 5
states:
u5 = 0
Equation 83
Therefore the overall matrix for this system is expressed in equation 85.
________________________________________________________________________________________________
J. Stokes
120
k1
k1
k k + k
1 1 2
0
k2
0
0
0
0
0
k2
k 2 + k3
k3
0
0 u1
P
0
0 u2

0 . u3 = 0
0
k 4 u4

0
1 u5
0
0
k3
k3 + k 4
0
Equation 84
u5 = 0
5
Node
Element 4
u4
4
Element 3
u3
3
Element 2
u2
2
Element 1
u1
1
Coating
Substrate
Figure 47, Schematic of the coated sample system used to demonstrate how the
finite element technique arrives at a solution.
Solution Phase
Solve a system of algebraic equations simultaneously.
In order to obtain numerical values of the nodal displacements, Es = 200 x 103 N/mm2
was used as the substrate stiffness [109] and Ec = 190 x 103 N/mm2 found in the current
research, and the sample width W = 20mm, length L = 80mm, and a thickness of ts =
________________________________________________________________________________________________
J. Stokes
121

1mm (substrate), and tc = 1mm (coating), all values used in the experimental section of
the research. The load applied will be assumed to be 5N.
The sample was discretized into four equal parts across its thickness, therefore the
cross-sectional area of each element in the y-direction can be expressed as:
A(y) = W.L = (20)(80) = 1600mm2
Equation 85
as ts = tc = t =1mm, therefore the length of each element, is as follows:
l=
t
= 0.5mm
2
Equation 86
The equivalent stiffness coefficient for each element is calculated from the following
equations:
keq =
A( y ) E
l
Equation 87
(1600)(200 x10 3 )
k1 =
= 6.40 x 108 N/mm
0.5
Equation 88
(1600)(200 x10 3 )
k2 =
= 6.40 x 108 N/mm
0.5
Equation 89
k3 =
(1600)(190 x10 3 )
= 6.08 x 108 N/mm
0.5
Equation 90
k4 =
(1600)(190 x10 3 )
= 6.08 x 108 N/mm
0.5
Equation 91
Putting these into the global stiffness matrix given by equation 84, to get:
________________________________________________________________________________________________
J. Stokes
122

0
0
0 u1
5
6.4 6.4
0
6.4 12.8 6.4
0
0 u2

108. 0
6.4 12.48 6.08
0 . u3 = 0 Equation 92
0
6.08 12.16 6.08 u4
0
0

0
0
0
0
0
1 u5
The displacement solution to this matrix is: u1 = 3.2 x 10-8mm, u2 = 2.5 x 10-8mm, u3 =
1.7 x 10-8mm, u4 = 0.8 x 10-8mm, and u5 = 0mm. The force of 5N caused the sample to
deflect by varying displacements from node 1 (highest displacement at point of force)
up to node 5 (no displacement). A contour line representation of this result is shown in
figure 48.
2 mm
0.5mm
u1 Deflection
u2 Deflection
u3 Deflection
u4 Deflection
Figure 48, Schematic of the graphical contour lines showing the displacements of
each node (solid lines) compared to pre-deformed shape (coloured lines).
Postprocessing Phase
Obtain other Information.
Additional information such as average normal stress in each element may be of

concern. These values can be determined from the equation:
keq (ui +1 ui )
f
=
=
=
Aavg
Aavg
Aavg E
(ui +1 ui )
Aavg
u u
= E i +1 i
l
Equation 93
________________________________________________________________________________________________
J. Stokes
123

u u
where i +1 i = , the strain in each element. The average normal stress through
l
each element is:

u u
2.5 3.2
-8
-5
2
(1) = Es 2 1 = (200 x 103)
x 10 = -280 x10 N/mm
l
0.5
Equation 94
u u2
3 1.7 2.5
-8
-5
2
(2) = Es 3
= (200 x 10 )
x 10 = -320 x10 N/mm
l
0.5
Equation 95
u u3
3 0.8 1.7
-8
-5
2
(3) = Ec 4
= (190 x 10 )
x 10 = -342 x10 N/mm
l
0.5
Equation 96
u u4
3 0 0.8
-8
-5
2
(4) = Ec 5
= (190 x 10 )
x 10 = -304 x10 N/mm
l
0.5
Equation 97
The compressive stress is experienced by both the substrate and the coating The
highest stress is found at the interface (bonding) point of the substrate and coating.
The stress is highest here due to the mismatch in stiffness. In the ANSYS program
these stresses would be represented by coloured contours, similar to that shown in
figure 49.
________________________________________________________________________________________________
J. Stokes
124
Node 4
Coating
-0.00342
Node 3
Node 2
Substrate
-0.0032
-0.00304
-0.0028
Node 1
N/mm2
Figure 49, Schematic of the nodal stress contours based on the numerical formulation
problem used in the present section.
The residual stress was simulation by applying a thermal load to the system (based on
the suggestion from the ANSYS user manual [110]) and in parallel by creating a
moment in the sample (applying a compressive force to the deposit while applying an
equal tensile force to the substrate), as shown in figure 49. Each of these simulations
should yield similar results to that found using Clynes Method and stress distributions
through the sample thickness, as all three extend from Beam Bending Theory. The
objective is to use the finite element analysis as a predictive method for determining
residual stress in deposits.
P
M
-P
Deposit
Substrate
-P
Generated moment generates an upward

deflection in the centre of the sample
Figure 50, Generated moment finite element analysis system.
________________________________________________________________________________________________
J. Stokes
125
4.6 Verification of Results

In recent years, the use of finite element analysis as a design tool has grown rapidly.
User friendly comprehensive packages such as ANSYS have become a common tool in
engineering design [110]. Unfortunately, without proper training or a solid
understanding of the underlying engineering principles on the part of the operator, the
finite element procedures can create errors or incorrect results. Sources of errors
include:
(a) Wrong input data, such as physical properties and dimensions
This mistake may be corrected by listing and verifying the physical properties and
co-ordinates of nodes or keypoints before proceeding any further with the analysis.
Units of the geometrical shape, properties and other influencing parameters must be
consistent.
(b) Selecting inappropriate types of elements
It is important to fully grasp the limitations of a given type of element and

understanding what element types belong to what system type, that is a two
dimensional element would be used in a two dimensional system and so on.
(c) Poor element shape and size after meshing
Inappropriate element shape and size will influence the accuracy of the results. It
is important to understand the difference between free meshing (using mixed-area
element shapes) and mapped meshing (using all quadrilateral area elements or
hexahedral volume elements) and the limitations associated with them.
(d) Applying wrong boundary conditions and loads
This step is usually the most difficult aspect of modelling [110]. It involves taking
an actual system and estimating the loading and the appropriate boundary
conditions for the finite element model.
Experimental testing (as carried out in this research) is the best way to check, verify
and predict finite element results. Where finite element results are not verified by
________________________________________________________________________________________________
J. Stokes
126

experimental results (due to the expense and time consumption in producing
experimental results), verification of finite element results may be found by other
methods, such as comparing results to simply analytical solutions.
Applying
equilibrium conditions and energy balance to different parts of the model, to ensure
that the physical laws are not violated is one method of verification. In static models,
the sum of forces acting on a body must equal zero. In heat transfer systems under
steady state conditions, the conservation of energies flowing in and out of nodes should
be balanced.
________________________________________________________________________________________________
J. Stokes
127
4.7 CAPABILITIES AND ORGANISATION OF ANSYS

The basic steps involved in creating and analysing a model with ANSYS are described
in this section. There are three processors used in ANSYS: (1) the Preprocessor
(PREP7), (2) the Processor (SOLUTION), and (3) the general Postprocessor
(POST1).
4.7.1 Creating A Model With ANSYS: Preprocessor

The preprocessor (PREP7) contains the commands needed to build a model, such as
creating a model geometry and defining element types and element real constants and
material properties, and defining meshing controls and to mesh the object created.
Creating a Model Geometry
There are two methods provided by ANSYS to construct a finite element models
geometry, direct or manual generation and the solid-modelling approach.
Direct
generation is a simple method where the user specifies the location of nodes and
manually defines which nodes makes up an element. This approach is generally
applied to simple systems that can be modelled with line elements, such as links or
beams, similar to that shown in figure 51 (a). While in the solid-modelling approach,
simple primitives (geometrical shapes such as rectangles, circles, polygons, blocks,
cylinders, and spheres) are used to construct the model, as shown in figure 51 (b).
Boolean operations (such as addition, subtraction, and intersection) are then used to
combine the primitives.
________________________________________________________________________________________________
J. Stokes
128
Node Line Element
Loaded Bridge
(a)
Problem
Finite Element Model

Add
Subtract
Problem
Boolean Operations
Elemental Model
(b)
Figure 51, Finite element model generation, (a) direct generation, (b) solid modelling
approach.
When creating primitives such as rectangles, the object is made up of entities including
keypoints, lines, and areas or volumes. Keypoints define the vertices of an object, lines
are used to represent the edges of the object, areas or volumes are used to represent
either two or three-dimensional solid objects respectively, as shown in figure 52.
These entities are automatically numbered by the ANSYS program, for data logging.
In general volumes are bounded by areas, areas are bounded by lines, and lines are
bounded by keypoints.
________________________________________________________________________________________________
J. Stokes
129
K1
L4
L1
K2
A1
L2
K = Keypoint
L = Line
A = Area
K2
K4
L3
Figure 52, The relationship between keypoints, lines and areas.
Element Type Options
ANSYS provides more than one hundred various element types, all of which may be
used to analyse systems.
Selecting the correct type of element type to use in a
particular system, is crucial in deriving a true solution. Elements may be of three

types: one, two and three-dimensional elements.
The structural and heat transfer systems analysed earlier in the report, were both
analysed using one-dimensional elements.
The element is represented by a line
bounded by two nodes, and acts in one direction. The line acts like a spring extending
or compressing depending on the change of its state, as shown in figure 53.
Where an object reacts in two directions due to some external condition, twodimensional elements are required to model the system. One-dimensional solutions are
approximated by line segments, whereas the two-dimensional solutions are represented
by plane segments. These two-dimensional plane functions may be in the form of
rectangular elements (4 nodes), quadratic quadrilateral elements (8 nodes), triangular
elements (3 nodes), or quadratic triangular elements (6 nodes), as shown in figure 54
[111].
________________________________________________________________________________________________
J. Stokes
130
Problem
Compressed
Beam
3
(1-Dimensional)
Element Model
Spring
Representation
Solution in Both
Cases
Figure 53, One-dimensional representation of a problem.
An example of a thin plate with a perforated hole, held at the top left hand corner
(boundary condition, x&y) and subjected to an applied load P1&2, is shown in figure 54.
This depicts the four possible types of two-dimensional elements that may be found.
Numerical equations apply in a similar way to two-dimensional elements as to onedimensional elements, however in figure 54, nodal displacements are calculated in both
x and y directions. A simple way to describe how an element behaves due to an
external condition, is to imagine that each line that bounds a two dimensional element,
reacts like springs, as shown in figure 55.
Just as a two-dimensional object reacts to external stimuli in two directions, a threedimensional object with react in three directions. Examples of three-dimensional
elements are brick elements (8 and 20 nodes) and tetrahedral (4 and 10 nodes) as
shown in figure 56 [111].
________________________________________________________________________________________________
J. Stokes
131
y = 0
Plate with Perforated Hole
x = 0
P1
P2
P = Load
= Deflection
Rectangular Element
Triangular Element
Quadratic
Quadrilateral Element
Quadratic Triangular
Element
Figure 54, Various types of two-dimensional elements [111].
________________________________________________________________________________________________
J. Stokes
132
(b)
(a)
(c)
kx
ky
ky
kx
Figure 55, Simple representation of a two-dimensional element (a), as a series of

springs (b) and subjected to a load (c).
8 Node
20 Node
z
x
y
4 Node
10 Node
Figure 56, Schematic of various three-dimensional elements.

________________________________________________________________________________________________
J. Stokes
133
ANSYS Element Types Used In Research
As described, there are many element types to choose from, and the selection of the
correct element for a specific system is important. The ANSYS user manual [110],
lists the correct element types to use for specific systems, hence this is how the
following element types were chosen. The detail of each element, is a synopsis of the
detail given in the ANSYS manual [111].
Heat Transfer:
PLANE55 is a four-node quadrilateral element used in two-dimensional heat
conduction systems. The element is defined by four nodes, with one degree of freedom
at each node, temperature. Output data include nodal temperatures and elemental data,
such as thermal gradient and thermal flux components.
PLANE77 is an eight-node quadrilateral element used in the modelling of twodimensional heat conduction systems. It is basically a higher order version of the
PLANE55 element.
The element is capable of modelling systems with curved
boundaries. At each node, the element has a single degree of freedom, temperature.
Output data of this element type include nodal temperatures and elemental data, such
as thermal gradient and thermal flux components.
SOLID70 is a three-dimensional brick element used to model conduction heat transfer
systems.
It has eight nodes having a single degree of freedom, temperature.
Convection or heat fluxes are applied to the elements surface. In addition heat
generation rates may be applied at the nodes. The element may be used to analyse
steady-state or transient systems. The solution output consists of nodal temperatures
and average face temperature, temperature-gradient components and the heat-flux
components.
________________________________________________________________________________________________
J. Stokes
134

Residual Stress:
PLANE13 has a 2-D magnetic, thermal, electrical, piezoelectric, and structural field
capability with limited coupling between the fields (thermal properties can be inputted
into the system, while stresses may be outputted). PLANE13 is defined by four nodes
with up to four degrees of freedom per node. The element has non-linear magnetic
capability for modelling B-H curves or permanent magnet demagnetization curves.
When used in a structural analysis PLANE13 has large deflection, large strain, and
stress stiffening capabilities.
PLANE42 is a four-node quadrilateral element used in two-dimensional solid
modelling systems. The element is defined by four nodes, with two degrees of freedom
at each node, the translation in the x and y-directions. The element input data can
include thickness if KEYOPTION 3 (plane stress with thickness input) is selected.
Surface pressure loads may be applied to the element edges. Output data include nodal
displacements and elemental data, such as directional stresses and principal stresses.
PLANE82 is an eight-node quadrilateral element used in two-dimensional solid
modelling systems. It is a higher order version of the two-dimensional PLANE42
element. This element offers more accuracy when modelling systems with curved
boundaries. At each node, there are two degree of freedom, the translation in the x and
y-directions. Just like PLANE42, the element input data can include thickness if
KEYOPTION 3 (plane stress with thickness input) is selected. Surface pressure loads
may be applied to the element edges. Output data include nodal displacements and
elemental data, such as directional stresses and principal stresses.
SOLID45 is an eight-node brick element used to model isotropic solid systems. Each
node has three translational degrees of freedom in the nodal x, y and z-directions.
Distributed surface loads may be applied to the element surfaces. This element is used
to analyse large deflections, large strain, plasticity and creep systems. The output
solution data consists of nodal displacements and elemental data, such as stresses in the
x, y, and z directions, shear stresses and principal stresses.
________________________________________________________________________________________________
J. Stokes
135

SOLID186 is a higher order 3-D 20-node structural solid element with a block
displacement behaviour (deforms in all three x, y and z-directions) and is well suited to
modelling complex shapes (such as those produced by various CAD/CAM systems).
The element has three degrees of freedom per node: translations in the nodal x, y, and z
directions. The element supports plasticity, hyperelasticity, creep, stress stiffening,
large deflection, and large strain capabilities
Finally, it may be worth noting that in general, better results and greater accuracy are
achieved with higher order elements, however these elements require more
computational time [111]. This time requirement is due to the numerical integration of
more complex elemental matrices involved in the calculation of the solution.
Element Real Constants
Element real constants are quantities that are specific to a particular element. For
example, a beam element requires a cross-sectional area, second moment of area and
so on, therefore these real constants vary from one element type to another,
furthermore not all elements require real constants. If the analyst has not used the real
constant, an ANSYS warning prompts as to the missing information.
Material Properties
Physical properties of the model must be defined, before the solution can be calculated.
For example, the modulus of elasticity, Poissons ratio or the density of the material
must be defined in solid structural systems, whereas in thermal systems, thermal
conductivity, specific heat, or the density of the material must be defined.
Define Meshing Controls
The finite element model geometry is divided into nodes and elements, a process
known as meshing. The ANSYS program automatically generates the nodes and
specified elements, once the element attributes and element size have been specified:
a) The element attributes include element type or types, real constants, and material
properties.
________________________________________________________________________________________________
J. Stokes
136

b) The element size controls the coarseness or fineness of the mesh. The larger the
element size, the coarser the mesh, conversely the smaller the element size, the
finer the mesh. If an element edge length of 10 units (unit will depend on the
dimensional unit used to create the model, like millimetres) is specified, then the
ANSYS program will generate a mesh of a series of elements with edge lengths no
larger than 10 units. Another way to control the mesh size is by specifying the
number of element divisions along a boundary line.
Meshing the Model
Once the meshing controls have been entered, meshing of the model may commence.
Clicking on the Meshing icon, meshing may be applied to areas or the entire volume.
The meshing process can take some time, depending on the model complexity and the
speed of the users computer. During the meshing, ANSYS writes a meshing status to
an output window.
There are two types of meshing options that can be used: free and mapped meshing.
Free meshing uses either mixed area element shapes (triangular elements mixed with
quadrilateral elements) or all triangular area elements. The mapped meshing options,
uses all quadrilateral area elements and all hexahedral (brick) volume elements.
Mapped area mesh requires the object to be analysed to have three or four sides, an
equal number of elements on opposite sides of a four sided object, and an even
numbers of elements on all sides of a three sided object. If an object is bound by more
than four lines, the concatenate command is used to combine some of the area edge
lines to form elements, to reduce the total number of lines.
4.7.2 Solution
The solution processor (SOLUTION) has the commands that allow you to apply
boundary conditions and loads.
Once all of the boundary information is made
available to the solution processor (SOLUTION), it solves for the nodal solutions.
________________________________________________________________________________________________
J. Stokes
137
4.7.3 General Postprocessor

The general postprocessor (POST1) contains the commands to read the results data
from the results file, read element results data, plot results and list results of an
analysis. There are other processors that perform additional tasks, such as the TimeHistory Processor (POST26), which contains the commands to review results over
time in a transient analysis at a certain point in the model [112].
The Design
Optimisation Processor (OPT) allows the user perform a design optimisation analysis.
________________________________________________________________________________________________
J. Stokes
138
4.8 ANSYS PROCEDURE

The procedure used in the current research, for both the heat transfer system and the
residual stress determination is contained in Appendix 7. The procedure describes the
following:
Describing the system
Building the geometry in the Preprocessor Phase
Defining material properties
Selecting the element type
Generating the mesh
Applying the boundary conditions in the Solution Phase
Solving the system
Obtaining the solution in the Postprocessor Phase
Saving and exiting the ANSYS program
________________________________________________________________________________________________
J. Stokes
139

Modelling

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

Modelling

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Modelling

Uploaded by

Copyright:

Available Formats

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

Residual stress is a major problem in thermal spraying where high

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

The current version of ANSYS has multiple windows incorporating a

Steps in the Finite Element Method

Numerical Formulation of systems specific to this research

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

4.2 ENGINEERING PROBLEMS

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

Physical properties characterising and disturbing various engineering

Parameters that Disturb

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

4.3 NUMERICAL METHODS

Moreover, an approximate continuous

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

4.4 STEPS IN THE FINITE ELEMENT METHOD

Assume a shape function to represent the physical behaviour of an element.

Develop equations for an element

Apply boundary conditions, initial conditions and loads

Solve a set of linear or non-linear algebraic equations simultaneously to

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

4.5 NUMERICAL FORMULATION

4.5.1 Heat Transfer System

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

Assume a solution that approximates the behaviour of an element.

the fact that heat flows in the direction of decreasing temperature.

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

as shown in figure 37.

h (for convection) is given by:

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

qconvection = hA[Tf Ts]

Figure 37, Energy balance at a surface with convective heat transfer.

Develop equations for an element.

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

The elemental description given by equations 38 and 39 may be expressed in matrix

Assemble the elements to present the entire system.

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

Apply Boundary Conditions and Thermal Loads

[Conductance Matrix]{Temperature Matrix}={Heat Flow Matrix}

In this system, the WC-Co coating is exposed to a known flame temperature T1 =

In the case the overall matrix equation is

As matrices are a representation of a series of simultaneous equations, the equation for

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

conductance matrix, to yield the results shown equations 53 and 54 as follows:

Row two is as follows when multiplied out:

Similarly row five reduces to the following:

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

Figure 39, Schematic of the heat transfer system.

CHAPTER 4: MODELLING OF SPRAY DEPOSIT

Heat transfer parameters found from experimental data.

(1) Flame (500oC)

(4) Stainless Steel

(5) Air Behind

45000 45000 + 166666.7

Simplifying, the following result is obtained;