Analysis of Fan Blade

ANALYSIS OF FAN BLADE
ATTACHMENT
Master Degree Project in Applied Mechanics

A-Level, 30 ECTS
Spring term 2014
Patrick Shingu
Miguel García Cabrera
Supervisor: Dr. Karl Mauritsson

Examiner: Dr. Thomas Carlberger
1
ABSTRACT
This thesis work was based on the analysis of a fan blade attachment whereby a complete 3D
model was presented by a partner company. The acceptability of a new design regarding to
the mechanical loads and consisting of dividing the hub into two parts instead of using a solid
hub is studied. From the model some critical parameters for the attachment of the blade with
respect to the stresses were chosen such as the rotational speed, fillet of the blade and the neck
of the blade. Parametric studies of these parameters were carried out in order to suggest the
new design. Bearing in mind that a safety factor of 2 was the prerequisite, based on the
analysis performed on ANSYS Workbench, it was suggested from the preliminary design that
the axial fan can operate in two specific scenarios consisting of a rotational speed of 1771 rpm
and a rotational speed of 1594 rpm. Using this set of parameters, a suggestion was drawn up
on the blade fillet which will give the lower stress. Blade fillet size of 30 to 35mm was
recommended while a size of 45mm was recommended on the neck of the blade. A modal
analysis was performed in order to find at what frequency will the model be vibrating and a
lowest and critical frequency of 16.8 Hz was obtained. Finally, a fatigue analysis of some
interesting areas was performed in order to determine the numbers of cycles before fatigue
failure occur. It was recommended to use the rotational speed mention previously since these
speeds have offered a High Cycle Fatigue results.
I
TABLE OF CONTENTS
1. INTRODUCTION ............................................................................................................... 1
2. BACKGROUND.................................................................................................................. 1
3. PROBLEM .......................................................................................................................... 4
4. GOAL AND PURPOSE ...................................................................................................... 6
5. LITERATURE STUDY OF THE LOADS ........................................................................ 6
5.1. AERODYNAMICAL LOADS .......................................................................................... 6
5.1.1. LIFT ................................................................................................................................ 7
5.1.2. DRAG ............................................................................................................................. 7
5.1.3. SKIN FRICTION FORCE ............................................................................................... 8
5.2. MECHANICAL LOADS .................................................................................................. 9
5.3. STRESS CONCENTRATION ....................................................................................... 10
6. METHODS........................................................................................................................ 10
6.1. PARAMETRIC STUDY ................................................................................................. 10
6.2. THE FINITE ELEMENT METHOD ........................................................................... 11
6.3. MODAL ANALYSIS ...................................................................................................... 13
6.4. CONVERGENCE ANALYSIS ...................................................................................... 13
6.5. FATIGUE ANALYSIS ................................................................................................... 15
7. DELIMITATIONS............................................................................................................ 18
II
8. FINITE ELEMENT MODEL .......................................................................................... 19
8.1. LOADS............................................................................................................................ 19
8.2. BOUNDARY CONDITIONS ......................................................................................... 21
8.3. MESH ............................................................................................................................. 22
8.4. CONTACT COUPLING BETWEEN PARTS .............................................................. 23
9. IMPLEMENTATION ....................................................................................................... 25
9.1 SOFTWARE ANALYSIS ................................................................................................. 25
9.1.1. ENGINEERING CONSIDERATIONS .......................................................................... 25
9.1.2. PARAMETRIC STUDY ................................................................................................ 26
9.1.3. MODAL ANALYSIS ...................................................................................................... 30
9.1.4. FATIGUE ANALYSIS ................................................................................................... 31
9.1 ANALYTICAL ANALYSIS .............................................................................................. 25
9.2.1. CONTACT PRESSURE ................................................................................................ 32
9.2.2. BENDING STRESS IN THE BLADE ........................................................................... 37
9.2.3. CENTRIFUGAL STRESS ............................................................................................. 38
9.2.4. STRESS CONCENTRATION AROUND THE WHOLE ............................................... 40
10. RESULTS .......................................................................................................................... 41
10.1. FILLET SIZE OF THE BLADE .................................................................................. 41
10.2. DIAMETER OF THE NECK OF THE BLADE .......................................................... 48
10.3. CONVERGENCE .......................................................................................................... 51
III
10.4. MODAL ANALYSIS RESULTS ................................................................................... 53
10.5. FATIGUE ANALYSIS ................................................................................................... 55
11. ANALYSIS ........................................................................................................................ 58
12. CONCLUSIONS ............................................................................................................... 64
13. FUTURE WORKS ............................................................................................................ 65
14. REFERENCES ................................................................................................................. 66
IV
TABLE OF FIGURES
Figure 1. Sketch of Impeller PFJ1 and components .................................................................. 2

Figure 2. Sketch of the blade and components........................................................................... 3
Figure 3. Old design using a solid hub ...................................................................................... 5
Figure 4. New design using a split hub ...................................................................................... 5
Figure 5. Airfoil shape (Zare, 2013) .......................................................................................... 7
Figure 6. The direction of the air force ...................................................................................... 8
Figure 7. Area affected by the skin friction force ...................................................................... 8
Figure 8. A solid of domain Ω (Dixit, 2009) ............................................................................ 12
Figure 9. Support conditions (Dixit, 2009) .............................................................................. 12
Figure 10. Straight line showing that error in is proportional to (Cook et al, 2002).... 15
Figure 11. S- N Curves for Alumium Alloy and Structural Steel (Fatemi, 2013) .................... 16
Figure 12. Constant Amplitudes (Fatemi, 2013)...................................................................... 17
Figure 13. Fictive centrifugal load applied to get equilibrium................................................ 19
Figure 14. Centrifugal force in terms of rotational speed ....................................................... 20
Figure 15. In red color, pressure load applied in a face ......................................................... 21
Figure 16. In yellow, boundary conditions applied to the model ............................................ 22
Figure 17. the quadratic (Ten node) tetrahedron using element with planar faces and side
nodes located at side midpoints ............................................................................................... 23
Figure 18. Mesh used in the analysis ....................................................................................... 23
Figure 19. Detail of the fillet on the blade in the initial design (mm) ..................................... 27
Figure 20. Detail of the blade fillet on the whole model in the initial design (mm) ................ 28
Figure 21. Magnification of the neck of the blade ................................................................... 29
Figure 22. Detail of the neck of the blade on the whole model................................................ 30
Figure 23. Sketch of the assembly between hub and boss ........................................................ 32
Figure 24. Boundary condition for ................................................................................ 35
Figure 25. Boundary Condition for ............................................................................... 35
Figure 26. Sketch of kinematical condition .............................................................................. 36
Figure 27. Sketch of the cantilever beam used to calculate the moment ................................. 38
Figure 28. Sketch to calculate the centrifugal stress ............................................................... 39
Figure 29. Stress analysis in the whole blade .......................................................................... 41
Figure 30. Maximum stress against blade fillet size for each rotational speed....................... 45
Figure 31. Stress in the fillet against blade fillet size for each rotational speed ..................... 46
V
Figure 32. Stress in the neck of the blade against blade fillet size for each rotational speed . 46
Figure 33. Maximum stress against rotational speed for each blade fillet size....................... 47
Figure 34. Stress in the fillet against rotational speed for each blade fillet size ..................... 47
Figure 35. Stress in the neck of the blade against rotational speed for each blade fillet size . 48
Figure 36. Maximum stress against rotational speed .............................................................. 49
Figure 37. Stress at the blade fillet against rotational speed .................................................. 50
Figure 38. Stress at the neck of the blade against rotational speed ........................................ 50
Figure 39. Convergence on the fillet of the blade .................................................................... 51
Figure 40. Convergence on the neck of the blade .................................................................... 52
Figure 41. Mode number against natural frequencies and coarse mesh against fine mesh .... 53
Figure 42. Picture of the fine mesh on the whole blade ........................................................... 54
Figure 43. Detail of the fine mesh on the blade ....................................................................... 54
Figure 44. Picture of the coarse mesh of the blade ................................................................. 55
Figure 45. Constant amplitude load obtained from ANSYS analyses (Fatemi 2013) .............. 56
Figure 46. Rotational speed against Number of cycles, where N stands for the number of
cycle .......................................................................................................................................... 57
Figure 47. Fatigue Analysis on the Fillet of blade at 1594 rpm .............................................. 57
Figure 48. Fatigue Analysis on the neck of the blade at 1594 rpm ......................................... 58
Figure 49. Maximum stress on the fillet of the boss................................................................. 59
Figure 50. Detail of maximum Stress on the fillet of the boss ................................................. 59
Figure 51. Maximum stress in the fillet of the blade ................................................................ 59
Figure 52. Approximation of FE solution of the bending stress (Pa) ...................................... 63
Figure 53. Bending stress analytical solution (Pa).................................................................. 63
VI
LIST OF SYMBOLS
ρ = Density
µ = Dynamic Viscosity
w = Weight
A = Area
CL = Lift Coefficient
CD = Drag Coefficient
Cf = Skin friction coefficient
V = Velocity
FD = Force due to drag
Fs = Force due to skin friction
Re = Reynolds Number
KI = Stress intensity factor
α = Angle of attack
LE = Leading edge
TE= Trailing edge
σcen = Centrifugal stress
W = Rotational speed
[K] = Stiffness matrix
[M] = Mass matrix
w = Natural frequency
h = Element size
P = Pressure
C, M = Fatigue Parameters (materials constants)
= Crack growth rate
∆K = range of the stress intensity factor
a = crack length
∆ = Diametric grip
VII
υ = Poisson’s ration
I = Moment of inertia
VIII
LIST OF SPECIAL TERMS AND ABBREVIATIONS
FEA-Finite Element Analysis
FEM- Finite Element Method
UTS- Ultimate tensile strength
SF- Safety Factor
HCF- High Cycle Fatigue
LCF- Low Cycle Fatigue
Leading Edge (LE): first part of solid surface in contact with fluid. Alternatively upstream or front
Trailing Edge (TE): last part of solid surface in contact with fluid. Alternatively downstream side or
back
Angle of attack: angle of attack is the angle between the body's reference line and the oncoming flow
Neck of the blade: the neck of the blade is the small piece of shaft which support the fan blade on to
the hub
Blade fillet: the blade fillet is the frontal fillet on the blade
Convergence: convergence is a way of confirming if the results obtained are corrected
Stress concentration: stress concentration is also known as stress raiser which occurs due to sharp
changes of geometry
Contact Pressure: contact pressure occur when there is contact between two bodies
IX
1. INTRODUCTION
For all the students registered for the Applied Mechanics program at the University of
Skövde, it is required to do a project in order to complete the program. The project runs over a
period of six months. The project also requires students to work in pair. This project was
conducted in partnership with Akron. Akron is a company which specialized in the design and
manufacturing of grain handling, fans and bioenergy components. This report covers the
analysis of a fan blade attachment. The company was planning to change the design in order
to reduce the cost and the fabrication of a component called the axial fan hub.
An axial fan is a type of compressor which moves air or gas parallel to the axis of rotation
unlike the centrifugal or radial flow fans which moves air or gas perpendicular to the axis of
rotation (Anon, 2002).
This report focuses on analyzing the attachment of the fan blade, according to mechanical
loads. The stresses acting on both the blades and the hub are analyzed, a modal analysis is
performed and finally, a fatigue analysis is carried out using the stress life approach.
2. BACKGROUND
Throughout history, the use of fans and ventilation systems has been increasing parallel to the
industrial developments as well as the requirements of the situations. Although Leonardo da
Vinci in the end of the 15th century described fans in his designs, the spreading of them
became more important with the appearance and exploitation of the mining industry. Thus,
along the 16th century, due to the extraction of coal, metal ores and other substances the
ventilation became basic in order to avoid flow gas which might either asphyxiate the miners
or explode with disastrous results. Indeed, the first mine ventilation system was published by
Georgious Agricola (1912). Many of the sophisticated airflow systems used nowadays were
embodied in his book. However, due to the lack of knowledge of the air properties, safety and
health measures, non-important advances appeared until the end of 17th century thanks to the
investigation of some important scientists like Galileo, Torricelli, Pascal, Boyle and Newton.
Along the 18th and the beginning of the 19th century, until the First World War, the main
researches were developed in United Kingdom, both in the seat of the British Government
1
and the mine industry. The present day applications
applications of fans are far too numerous to list owe to
the improvement achieved, highlighting the aviation industry (Cory, 2005)..
An axial flow fan is a machine which creates flow within a fluid (gas) in such a way that air
flows linearly along the axis of it. It is composed of blades that force air to move parallel to
the shaft about the axis at which the blades rotate, which are rigidly secured. Industrial axial
are used in extremely conditions such as high temperature, large vibrations
vibrations amplitudes or high
corrosion, which cause large strains. Due to this, both the hub and the fan blades are made out
of casting aluminum, which is capable to resist aggressive environments, have light weight
and be easily malleable to the conditions required. However, nowadays, Akron is using a
solid model, which means an increase of the material used and consequently, the cost of
production is higher. A sketch of an impeller, its components and a detailed sketch of the fan
blade are shown in Figure 1 and Figure 2.
Figure 1.. Sketch of Impeller PFJ1 and components
2
Figure 2. Sketch of the blade and components
Due to the importance of the fans and ventilation systems in the industry, as explaining above,
several researches have been carried out using different analysis and investigations in order to
know the behavior both the stresses and the fracture mechanics. Most of them have been
performed by using of the Finite Element Methods (Cory, 2005).
With respect to the analysis of the stresses, Abdullah and Schlattmann (2012)
2012) performed an
analysis taking into account the effect of centrifugal and aero-dynamical
aero dynamical loads and varying
geometrical parameters, concluding
luding that in case of solid hub, maximum stresses are induced
at the root of the blade, in such a way that if the thickness increases reducing the stresses at
the root, natural frequency decreases. Furthermore, they suggested that the increment in the
diskk radius implies higher stresses and deformations. The above statements were supported
by Arewar and Bhope (2013)
2013).. According to Zare and his analysis of axial fan blade (2013),
the relation between the safety factor, chord length and pressure load is such that increasing
the constant pressure load, remaining
remain the chord length constant, von
on Mises stress increases.
increases
3
With respect to the fracture mechanics analysis, based on Tsai (2004) and his study about the
rotating vibration behavior of the turbine blades, dynamic behavior and resonant frequencies
varies widely regarding to the position of the fan blades. He also suggested that highest
stresses are located at the first neck of the blade root. According to Amoo (2013), vibration
loads acting on the fan blades can lead to fatigue especially if the blades are not designed
properly. There are two main types of fatigue which affect the fan blades: low cycle fatigue
(LCF), which is less than 100000 cycles or within the range of 0.1-5 Hz, and high cycle
fatigue (HCF) which has values over 100000 cycles or within the range of 17-50 Hz (Totten
and MacKenzie, 2003). The HCF is known for having a high frequency which is subjected to
cyclic bending and the LCF is known for higher amplitudes and a low frequency (Amoo,
2013).
Barlow and Chandra (2005) carried out a fatigue crack propagation simulation under
centrifugal and aero-dynamical loads, suggesting that stress intensity factors indicate a strong
Mode I (KI) and Mode III (KIII) at the edge of contact, while only crack opening condition
(KI) is observed in free surface. Finally, Poursaeidi and Salavation (2007) performed an
analysis of the failure of a generator rotor fan blades, stating that failure was caused by aero-
dynamical disturbances that created a resonant condition of vibration which induced to stress
intensity factors superior to the critical stress intensity factor. However, despite all the
analysis mentioned in the previous paragraphs there is no work done on an axial flow fan with
a split hub.
3. PROBLEM
As it has been explained in the background, a new innovative concept design has been
modeled by Akron in order to overcome the material wasted and consequently the costs when
manufacturing the fans using solid hubs (Figure 3). The design consists of dividing the hub
into two different parts, in such a way that both parts are joined to the base of the fan blades
by bolting each other (Figure 4). Therefore, stress concentrations are expected in the
attachment points. A thorough analysis of the mechanical loads is required in order to predict
the behavior of the model and the mechanical viability of the new design.
4
Figure 3. Old design using a solid hub
Figure 4. New design using a split hub
5
4. GOAL AND PURPOSE
The goal of the project is a suggestion of an acceptable design according to the stresses acting
on the fan blade attachment points. That is, the stresses both on the fillet of the blade and the
neck of the blade are analyzed in such a way that a safety factor ≥ 2 is required. The
suggestion will be based on the lowest stresses considering the prerequisite of the safety
factor varying the rotational speed.
The purpose of obtaining this goal is to reduced production costs due to a reduction of the
material used.
5. LITERATURE STUDY OF THE LOADS
As it has been explained in previous sections, an axial fan is loaded by both aerodynamical
and mechanical loads. A literature study was performed about these loads in order to gather
information.
5.1. AERODYNAMICAL LOADS

In order to find the aerodynamical load, research was based on the design load and the profile
of the fan blade. It was noted that the blade is profiled in a streamline shape of cross section
called airfoil (Zare, 2013). This shape plays a vital role on the performance of the blade
(Figure 5). The airfoil can increase the efficiency of the blade and can also reduce the flow
turbulence (Zare, 2013). A higher efficiency is achieved if the rotor blades are profiled in an
appropriate way and have been twisted as well (Schildhauer and Minges, 2010). The airfoil
section is being used in various areas such as airplane wings, wind turbine, gas turbine, fan
blade, compressor, etc… (Zare, 2013)
6
Figure 5. Airfoil shape (Zare, 2013)
As seen in Figure 5, the chord length is the distance between the leading edge (LE) and the
trailing edge (TE). The angle of attack (α) is between the chord length and the relative air
velocity and where L stands for the lift.
5.1.1. LIFT
Lift can be defined as the part of the aerodynamic force which is perpendicular to the relative
airflow and F is the resultant force.
The lift force is given by equation (1):
F = 0,5 ∙ C ∙ ρ ∙ A ∙ v (1)
Where FL is the lifting force, ρ is the density of air, v is the relative velocity of the airflow, A
is the area of the airfoil as viewed from an overhead perspective and CL is the lift coefficient
(Sullivan, 2006).
Further research was done in order to have a better understanding about the lift theory. The
theory used is Bernouilli’s principle, which states that a gas or liquid, which has a high
relative velocity, will create a lower pressure and the area with a higher pressure will have a
lower relative velocity (Sullivan, 2006). This means that the airfoil is shaped in such a way
that the upper surface is larger than the lower surface. Therefore, there is a separation as the
air separates from the leading edge (Sullivan, 2006). The air which is going through the upper
section has a greater path to travel.
5.1.2. DRAG
The drag force is the resistance created by a medium flowing over a body. This force acts in
the direction of the object as it is shown in Figure 6 (Sullivan, 2006).
7
Figure 6. The direction of the air force
The drag force helps to choose what strength of driving system is required to propel the fan
blades. The drag force equation is given by equation (2):
F = 0,5 ∙ C ∙ p ∙ A ∙ v (2)
The formula is very similar to the lift force equation; the CL is replaced by the CD which is the
coefficient of drag which can be obtained by the use of a graph or various formulas depending
on magnitude (Sullivan, 2006).
5.1.3. SKIN FRICTION FORCE

The skin friction force is the force created by the actual parallel perimeter of the object
(Figure 7). The skin friction depends on what type of material is used as a skin of the fan
blades; the rougher the skin, the more the drag. The skin friction can also be created by the
smooth surface but the result is by far less compare to the rough surface.
Figure 7. Area affected by the skin friction force
The Reynolds number in this case is characterized by the formula = ∙ ! ∙ "⁄#. Here x is
the horizontal surface length of the above cross section (distance from the leading edge to the
trailing edge). The larger x is the more surface length there is to create skin friction drag.
The skin friction force is given by (3)
F% = 0,5 ∙ C% ∙ p ∙ A ∙ v (3)
8
where Cs is the skin friction coefficient which is calculated in various ways depending on the
scenario. If the roughness of the surface is specified then one would calculate the relative
roughness (L/e) where e represents the roughness ratio, the Reynolds number and get the
specific skin friction coefficient of a graph (Sullivan, 2006).
In general most of the airfoil sections are obtained from the catalogue mostly provided by
NACA. These sections are mainly used in the turbine industry, aviation industry and fans
industry (Zare, 2013). Note that NACA stands for National Advisory Committee for
Aeronautics and has been producing the airfoil section since 1930 (Zare, 2013). The first step
to take when designing an axial fan is to obtain the tip diameter and the hub diameter, after
that select an airfoil profile, then using NACA catalogue, the blade chord will be provided.
5.2. MECHANICAL LOADS

As part of the project brief, the mechanical load acting on the fan blades and the hub had to be
determined. Research was done in order to find these loads. It was found that the fan blades
are subjected to continuous steady and vibratory loads during operation (Amoo, 2013). The
vibratory loads will be discussed at a later stage. The steady loads consist of centrifugal pull
load and the torque (Amoo, 2013). Theses loads can be analysed, assessed and determined. It
was also noted that the centrifugal pull loads are dependent on the span- wise mass
distribution of the blade while the torque load is dependent on the span-wise mass
distribution, chord mass distribution and the twist of the blade (Amoo, 2013).
In general it is considered that the centrifugal load is by far the greatest contributor taking into
account both the mechanical and the aerodynamical loads acting on the fan blades and hub. It
was also found that the blades are also subjected to bending stress but usually it is low so that
most of the time it is being neglected (Poursaeidi & Salavatia, 2007).
The centrifugal stress equation is given by (4):
M∙V M ∙ r) ∙ ω
σ)*+ = = (4)
A ∙ r) A
where M is the mass of the airfoil, V the surface Velocity, rc the radius of the mass from the
rotation axes, A the area of the cross section, and 1 is the rotor speed in rpm.
9
5.3. STRESS CONCENTRATION
The literature on the stress concentration was done based on the geometry of the model. Stress
concentration is also called stress raiser, where the stress is concentrated due to a geometry
changes (Budynas et al, 2006). This may be caused by:
- The geometry discontinuities such as holes, notches, sharp corners and fillets
- The material not being homogeneous during the molding and casting.
- The irregularities in the surface such cracks and marks created during the machining
operations.
It can be summarized that the stress concentration is affected by both the geometry and the
loading. Different methods can estimate the stress concentration factor and each of them offer
advantages and disadvantages (Budynas et al, 2006). The stress concentration may be
estimated using the catalogue, FEM and theoretical values calculated.
6. METHODS
Various text books, journals articles and internet websites were consulted in order to gather
enough information about the different possible methods used to perform the analysis in order
to achieve the goal and the purpose of this dissertation. The methods were being split up into
five parts, which are parametric study, the Finite Element Analysis, modal analysis,
convergence analysis and fatigue analysis.
6.1. PARAMETRIC STUDY

A parametric study is a method that consists of a description, evaluation and examination of
relationships between different parameters (Caicedo, 2007). The design step of an axial fan is
an iterative process, since product engineers use to modify this process in such a way that
acceptance criteria defined by safety, cost, performance, convenience and shape are found out
(Singh, et al., 2011). In order to optimize product development, a parametric study will
provide enough data to automate these iterations and find the best design. The study compares
the results of different scenarios in such a way that the best alternative may be found easily.
(Singh, et al., 2011).
10
The goal of this thesis will be reached by performing a parametric study in order to suggest an
acceptable design (considering stresses). In this case, the method is applied in the following
parameters since all of them are expected to affect the stresses: the fillet of the blade, the neck
of the blade and the rotational speed.
6.2. THE FINITE ELEMENT METHOD
The Finite Element Methods (FEM) is a method to get an approximated solution to

differential equations (Zare, 2013). The FEM is a practical method for analysing structure
with many degrees of freedoms (Zare, 2013). It is always recommended to use some
engineering software in order to carry out the analyses. The software is more useful when the
geometry, the loadings and material properties are complicated and the analytical approach
would not be suitable in order to obtain solutions. The tools are not only used for saving time
but also it prevents specialists from solving problems manually (Zare, 2013).
Before performing any finite element method, some basics terms used needs to be known.
Since the FEM method is an approximation solution to differential equation, the differential
equation will be subjected to boundary conditions (Dixit, 2009). The boundary condition can
be defined as the value of the field variables where the field variables are the variables of
interest used by the differential equation (Dixit, 2009). A node can be defined as a point in the
finite element where the field variable is required to be calculated explicitly (Dixit, 2009).
The shape functions can be defined as the interpolation functions of the nodal values. This can
be explained by the following equation:
2(3,4) = 56(3,4) ∙ 26 + 5 (3,4) ∙ 2 + 58(3,4) ∙ 28 (5)
where 26 , 2 and 28 are the field variable and 56 , 5 and 58 are known as the shape
functions.
A formulation of FEM for a linear differential equation can also be shown as
9 ∙ # + : = 0 (6)
where u can be considered as the functions of the coordinates, L is the differential operator
and q is the vector of known functions (Dixit, 2009). From the above equation u can be
considered to be the boundary condition.
11
STRESS FORMULATION
The formulation of the stress is important for this thesis. Consider the figure below where
there is a uniform thickness bounded by two parallel planes and any closed boundary Γ, as
shown in the Figure 8. The significance of the boundary condition is shown in Figure 9.
In the instance of the FEM formulation for the stress, the constitutive matrices are added.
Therefore, the stress formula becomes (7):
<=3 <=34 < #

+ + ?3 = @ ∙ (7)
<" <> <A
<=34 <=4 < !

+ + ?4 = @ ∙
<" <> <A
where fx and fy are the body force per unit volume on the x and y directions, ρ is the density of
the material, σx and σy are the normal stresses, u and v are the displacement in both the x and y
axis respectively and σxy is the shear stress in the xz and yz planes (Dixit, 2009).
Figure 8. A solid of domain Ω (Dixit, 2009)
Figure 9. Support conditions (Dixit, 2009)
12
6.3. MODAL ANALYSIS
A modal analysis is a study of the dynamic properties of structures under vibrational

excitation (Cook et al, 2002). A modal analysis is performed in structural mechanics in order
to determine the natural mode shapes and frequencies of an object or structure during free
vibration. The vibratory loads are loads which may include aerodynamics and installation
dynamics (include inlet flow distortion, inlet vortex, altitudes effects and thrust reverser)
(Amoo, 2013). FEM is used to determine the magnitudes of both the mechanical and
vibratory loads. This is done by using the concept of modal amplitudes (eigenvectors) and
frequencies (eigenvalues) of the blades (Amoo, 2013). The equation used by most engineering
software to find the eigenvalues and eigenvectors is given by equation (8):
(CKE − ω ∙ CME) ∙ GDI=0 (8)
where [K] is the stiffness matrix, [M] is the mass matrix, ω2 is an eigenvalue, ω is the natural
frequency and D is the eigenvector. From this equation, it can be said that a lower mass and /
or a stiffer beam increases the natural frequency where else the higher the mass and / or softer
beam will decrease the natural frequency (Cook et al, 2002).
A natural frequency can be defined as a frequency at which a free object vibrates once it is in
motion (Anon, 2014). It was found that the stress induced through vibrations is critical when
showing resonant in a structure, as this can lead to a catastrophic failure of the blades (Amoo,
2013).
According to Cook et al. (2002), the use of FEM in order to carry out the modal analysis is
profitable since the object to be analyzed may have arbitrary shape and the results obtained
are acceptable. Throughout the Experimental of the Modal Analysis, once the frequencies and
the shape modes have been determined, the physical test can calibrate if the assumptions
taken into account in the finite element model are correct (Ohman and Singhal, 1993).
6.4. CONVERGENCE ANALYSIS
In order to achieve trustworthy results, it is very important to check whether the solution has
converged. As the mesh is refined, an acceptable FE formulation has to converge to the exact
solution of the mathematical model. Furthermore a satisfactory rate of convergence is also
13
important, so that acceptable accuracy can be reached. The rate of convergence of a particular
type of element can be obtained by analysis, or by study of results provided by a sequence of
successively refined meshes (Cook et al, 2002).
H-method was chosen in order to perform the convergence in the software. This method uses
simple shape functions and many small elements (Cook et al, 2002). The method consists of
running the analysis using a coarse mesh. The results obtained from this method are compared
with a second running analysis in which the mesh used is finer than in the previous analysis
(Cook et al, 2002). The output from the two runs is compared and so on and so forth until the
percentage of change between solutions is less than 2%. This percentage represents the
discretization error, that is, the difference between the mathematical model and its discretized
(finite element) model (Cook et al, 2002). . Therefore, the error is calculated according to (9).
=MN6 − =M
JKKLK = ∙ 100 (9)
=M
where i represents the number of analysis.
Then, convergence may be represented by means of Richard´s Extrapolation, according to

Equation 10 (Cook et al, 2002).
26 ∙ ℎR − 2 ∙ ℎ6R
2P = R R (10)
ℎ − ℎ6
where 2P corresponds to element size h=0, ℎR is the order of error of 2, which is a quantity
of interest calculated at some location in a FE mesh.
According to Equation 10, convergence is represented by a straight line in a plot of 2 versus

ℎR , as it is shown in Figure 10.
14
Figure 10. Straight line showing that error in is proportional to (Cook et al, 2002)
6.5. FATIGUE ANALYSIS
Many parts on the axial fan may work well initially. In any point these parts may fail due to
fatigue failure caused by repeating the cyclic loading. The fatigue analysis is a study which is
capable of finding if material can survive the many cyclic components. There are at least three
ways of performing a fatigue analysis which are the strain life, the stress life and the fracture
mechanics.
The strain life has an advantage in measuring quantity which has a low cycle fatigue. This
method is suitable when there is also crack initiation, since the fatigue failure begins most of
the time in the notch, crack or the area where higher stress concentration are located
(Budynas, 2006). Therefore, when the stress exceeds the elastic limit, automatically plastic
strain would take over. If fatigue failure would occur, there is a strong possibility of being
plastic strain (Budynas, 2006). The strain life equation is shown in the equation (11)
∆T =UV Z
= ∙ (2 ∙ 5U )W + TUV ∙ X2 ∙ 5U Y (11)
2 J
[\
where is the total strain amplitude, E is the Young´s Modulus, Nf is the number of cycles to
failure, σf is the fatigue strength coefficient, b is the fatigue strength exponent, c is the fatigue
ductility exponent and εf is the fatigue ductility coefficient.
15
With respect to the stress life method, in order to find the materials strength when it
undergoes the action of fatigue loads, the materials undergo repeated or varying forces of
certain magnitudes under while the cycle are counted to failure (Budynas, 2006). This method
is based on S-N curves (Stress- Cycle curves) or Wöhler curve. The graphical representation
of S-N curves is shown below for Aluminium Alloy and Structural Steel.
Figure 11. S- N Curves for Alumium Alloy and Structural Steel (Fatemi, 2013)
The stress life method is applied to the total life and does not differ from the initiation and the
propagation of the crack (ANSYS, 2004). The method is mainly used for a high number of
cycles which is usually more than 105, this method deals mainly for HCF.
Regarding the stress method, it is important to describe the type of loading. There are two
different types of loading: constant amplitudes with proportional loading and constant
amplitudes with non-proportional loading,
The constant amplitudes with proportional loading is used since the loading step is constant
and only one set of FE stress results is required to calculate the alternating and mean values
(ANSYS, 2004). Figure 12 shows how to deal with constant amplitudes.
16
Figure 12. Constant Amplitudes (Fatemi, 2013)
From the above figure the following equation can be derived (Equation 12, 13, 14 and 15):
∆] ]^ 3 − ]^M_
] = = (12)
2 2
]^ 3 + ]^M_
]^ = (13)
2
]^ 3 = ]^ + ] (14)
]^M_ = ]^ − ] (15)
In order to obtain the fatigue properties, there are two common reference test condition used
(Fatemi,2013). The first one is = −1 which is called fully reversed condition. This implies
that ]^M_ = −]^ 3 . The second one is = 0, which means that ]^M_ = 0, that is, pulsating
tension. As it can be noticed from the above diagram with constant amplitudes loading type,
once cycle is equal to two reverses (Fatemi, 2013). From Figure 12, a cycle can be defined as
the smallest segment of the stress versus time history which is repeated periodically (Fatemi,
2013)
17
However, the constant amplitude with non-proportional loading is mostly suitable for
nonlinear contact, compression only or bolt load (ANSYS, 2004). Therefore, it is not relevant
for this dissertation.
The fracture mechanics approach is done by initially assuming the initial crack in order to
determine the crack’s growth. This method is called “crack life”. It is used to determine
inspection intervals. From the initial crack, the critical crack can be obtained from the crack
growth. According to the ANSYS manual (2004), the crack initiation is determined by the
strain life method while the fracture mechanics approach determines the crack life.
`LAab 9c? = dKaef cgcAcaAcLg + dKaef 9c? (16)
7. DELIMITATIONS
As part of the thesis, some limitations have to be considered in order to restrict the analysis
carried out. Since computational fluid dynamics (CFD) is not our area of expertise and due to
the lack of information, a CFD analysis will not be performed. Therefore, the aerodynamical
loads which include lift force and drag force are not included. Furthermore, according to
analytical calculations, the values of those forces are minimum with respect to the centrifugal
load and the pressure. Thus, they are neglected along the thesis.
The model provided by the company consisted of two split hub, cast out of aluminum. An
analysis of an alternative material is not carried out.
Due to the time constraint, vibrations coming from waves spread to the fan due to proximity
of other fans are not performed.
The environment impact is another area to look at. As it has been explained in the
background, the axial fan is subjected to extreme conditions of corrosion and temperature.
The materials used in the design of the model are Aluminum Alloy and Structural Steel. Both
materials have been design in room temperature of 22 °C. This means that running the fan at a
higher and lower temperature might affect the performance of the axial fan. When
manufacturing the model, in order to protect the axial fan against corrosion, the axial fan is
painted. Although these factors can affect the analysis, it is assumed that the axial fan is
18
protected against these environmental conditions. Thus, they have not been taken into account
when running the analysis.
Finally despite the parametric study, the general shape of the model (split hub joined together
by bolting them) did not change, since it was supposed to be the expected design.
8. FINITE ELEMENT MODEL
In order to perform the analysis, all loads, boundary conditions, mesh and contact relations
have to be defined. Along this chapter an explanation of how the model is defined in ANSYS
Workbench with respect to these terms is presented.
8.1. LOADS
Two types of load were considered when performing the analysis. The first load was the
centrifugal force. The centrifugal force is applied as a fictive force in order to achieve the
equilibrium. Although it is a virtual force, it is defined like a volume; otherwise the results
could not be trusted since the model would be rotating (Figure 13).
Figure 13. Fictive centrifugal load applied to get equilibrium
19
This load was applied in terms of the rotational speed. The software automatically generates
the centrifugal force once the rotational speed is applied. Since the rotational speed is constant
that means that it can be applied throughout the blade, as it is shown in Figure 14.
Figure 14. Centrifugal force in terms of rotational speed
Pressure load is the second type of load that the axial fan is subjected to. This is the pressure
which goes on to the inner and outer of the fan. The load is applied in a face in the form of
pressure load, as it is represented in Figure 15.
20
Figure 15. In red color, pressure load applied in a face
8.2. BOUNDARY CONDITIONS
Based on the concepts of strength of materials and taking into account that the model has been
defined as a solid, there are three degrees of freedom. These degrees of freedom are the
displacement in each axis and they have been fixed (Table 1). Furthermore, according to
Dynamics, the rotations may not be considered. Otherwise, if a single point is considered, as
the displacements are fixed, the model would rotate indefinitely without equilibrium. The
boundary conditions applied to the model are represented in Figure 16.
CONDITION
Displacement X Fixed
Displacement Y Fixed
Displacement Z Fixed
Table 1. Boundary conditions applied in the boss
21
Figure 16. In yellow, boundary conditions applied to the model
8.3. MESH
An important aspect when performing an FE analysis is to carry out the mesh. The axial fan
was modelled using a 10 node tetrahedral element. This element was part of the software
package used and the software uses the automatic meshing in order to create the mesh. The
element is known in performing better when it comes to stress analysis in structures and solid
mechanics (ANSYS, 2004). The element is also known for having four corner nodes, three
degrees of freedom at each node (ANSYS, 2004). Figure 17 shows a typical 10 Node element.
Another concept pops up when performing the mesh is the difference between coarse and fine
mesh. The literature shows that the coarse mesh cannot be trusted. In order to get accurate
results, the mesh has to be finer when certain geometry changes such as fillets, holes, notches
and sharp edges appear (Figure 18)(Cook et al, 2002). Differences between coarse and fine
mesh are compared in future chapters.
22
Figure 17. the quadratic (Ten node) tetrahedron using element with planar faces and side
nodes located at side midpoints
Figure 18. Mesh used in the analysis
8.4. CONTACT COUPLING BETWEEN PARTS
Another important concept when performing analysis is to define the contact conditions.
There are five types of contact: bonded, no separation, rough, frictionless and frictional.
ANSYS Workbench treats the last three type of contact as Non Linear (Rough, Frictionless
and Frictional) (Save, 2013).
23
BONDED
The bonded surfaces can be defined as surfaces which are rigidly fixed or glued together. The
surfaces are not allowed to separate or to slide. This means that the surface will be mated
without taking into account the penetration, gap, loading and behavior of other parts and
contact. ANSYS uses this setting as a default and it is suitable for linear type of contact (Save,
2013).
NO SEPARATION
No separation type of contact is almost similar to the bonded contact with the only difference
that the parts are allowed to slide slightly. This setting is used when knowing that the parts
will not be separated and the sliding will be always frictionless. However the sliding is very
limited (Save, 2013).
ROUGH
In the case of the rough contact type, the parts are not allowed to slide. Nevertheless, the part
can be separated depending on the loading (Save, 2013).
FRICTIONLESS
The parts can slide freely and the contact can be opened and close depending on the loading.
This type of loading can give rigid body error under constraint. This effect can happen on any
nonlinear contacts which are allowed to separate (Save, 2013).
FRICTIONAL
The parts can be slide if the user specifies the coefficient of friction and the parts can open
and close as well. This case can also be explained by having two parts sliding against one
another (Save, 2013).
When performing the analysis the bonded type of contact was used. This type of contact was
most suitable due to the fact that there was not sliding and that the contact type was under the
linear type of contact. This approach was used in order to simplify the problem.
24
9. IMPLEMENTATION
In this chapter, the procedure and the main considerations used in order to perform the
analysis are explained. It is divided into two main parts, according to the type of analysis
performed. The first part is called Software Analysis, while the second one is named
Analytical Analysis.
9.1. SOFTWARE ANALYSIS

Different software has been used to perform the software analysis as it follows. The company
provided the model in SolidWorks format. Using this software, the pertinent modifications
(parametric study) has been done in order to get the final required model. Then, the model
was saved as a step file and finally imported to ANSYS Workbench in order to carry out the
respective finite element analysis.
The software analysis consists of the five following main parts: engineering considerations, a
parametric study, a modal analysis, convergence analysis and fatigue analysis.
9.1.1. ENGINEERING CONSIDERATIONS
To carry out the software analysis, finite element analysis was performed in order to obtain
the stresses using ANSYS Workbench software. Some considerations had to be taken into
account before running the analysis.
• MATERIAL PROPERTIES
It is necessary in order to perform the analysis in Finite Element Software to specify the
material properties of the model. The whole axial fan was modeled using Aluminum Alloy
except for the fasteners (screw and washers), which were modeled using Structural Steel. A
linearly elastic material model was used to perform the analysis. In Table 3, the main material
properties of both elements are shown.
25
ALUMINUM
ALUMINUM ALLOY STRUCTURAL STEEL
Density (xy⁄z{ ) 2770 7780
Young´s Modulus (Pa) 7 ∙ 106… 2.06 ∙ 1066
Poisson´s Ratio 0.3 0.3
Shear Modulus (Pa) 2.7 ∙ 106… 7.7 ∙ 106…
Tensile Yield Strength (Pa) 2.8 ∙ 10ˆ 2.5 ∙ 10ˆ
Tensile Ultimate Strength
3.1 ∙ 10ˆ 4.6 ∙ 10ˆ
(Pa)
Table 3. Material properties of Aluminum Alloy and Structural Steel
9.1.2. PARAMETRIC STUDY

As it was explained in Chapter 6.1, the parametric study can be defined as an iterative method
developed in order to obtain acceptance criteria both for safety, cost, performance, shape and
convenience. This iteration describes and provides certain relationships between different
parameters. By means of the study and the comparison of the results obtained, different
scenarios are created, suggesting an idea of an acceptable design for the model. To perform
the parametric study, ANSYS Workbench was used. Three different parameters, whose
modifications are expected to affect the concentration of the stress in the attachment point
between the fan blades and the hub, are studied. These parameters are: rotational speed, fillet
size of the blade and diameter of the blade neck.
ROTATIONAL SPEED
The literature shows that the rotational speed and the pressure load are dependent of one
another. Therefore the purpose of this section was to establish the required relationship. It is
expected that the increase in one of them, will entail an increase of the stresses. However,
both the rotational speed and the pressure can be studied at the same time, since fans operate
under a predictable set of laws concerning them. According to Sullivan (2006), a change in
rotational speed of any fan will predictably change the pressure rise and power necessary in
order to operate the fan at these new values.
The relation between the rotational speed and the pressure can be stated according to Equation
(17)
Š‹
ŠŒ
= • ‹Ž (17)
Œ
26
where P is referred to the pressure and N to the rotational speed.
In order to perform the parametric analysis, seven different values of the rotational speed and
the pressure had been taking into account (Table 4). Both parameters starts at the maximum
values allowed, and it is decreasing as it has been explained above.
ROTATION SPEED (rpm) PRESSURE (Pa)

3000 9000
2700 7290
2430 5905
2187 4783
1968 3874
1771 3138
1594 2542
Table 4. Values of rotational speed and static pressure used
FILLET SIZE OF THE BLADE

In Figure 19 and Figure 20, the green color represents the fillet of the blade which would need
to be modified. It is expected that, owe to both the air pressure and the centrifugal load, the
increase of the fillet size will result in the decrease of the stress.
Figure 19. Detail of the fillet on the blade in the initial design (mm)
27
Figure 20. Detail of the blade fillet on the whole model in the initial design (mm)
The fillet size has been modified in such a way that it has been taken three different oversized
values with respect to the original model provided by the company, and three different more
undersized, apart from the original one. Each of them has been analyzed for each value of the
rotational speed. Table 5 is represented in order to show combination used to perform the
analysis.
10 15 20 26.5 30 35 40 Fillet Size (mm)

Rotational
3000 2700 2430 2187 1968 1771 1594
speed (rpm)
Table 5: Values to proceed with the fillet size analysis (All combinations were used to
perform the analysis)
DIAMETER OF THE NECK OF THE BLADE
A parametric study of the neck of the blade was performed (Figure 21 and Figure 22). This is
a crucial part due to the geometry, since the neck of the blade is the support of the blade, that
is, the base of the blade. Then, it is expected to give high stress concentration. Crack would
start propagating either from the top of the neck of the blade or from the bottom. For this
reason, a parametric study was required in order to examine what will be the outcome if the
28
neck of the blade had increased while rotating at the maximum speed. Therefore, three
parameters were chosen and analyzed at different rotational speed. A neck diameter of 45 mm
had been used on the preliminary design but this value was increased in such a way that the
stress around the blade would decrease or increase when it would be subjected to certain
rotational speed parameters. The values chosen were 47, 49 and 51 mm. The preliminary
design was the design provided by the company with no changes on any of the geometrical
parameters.
The first couple of analyses were performed using the preliminary design subjected into
different rotational speed and pressure load. The remaining analyses were performed using the
new parameters. The new parameters were analyzed using the worst condition and the lowest
condition. That means a diameter of 51 mm was analyzed using the rotational speed of 3000
rpm. By contrast, an analysis was performed using the lowest rotational speed of 1594 rpm.
Figure 21. Magnification of the neck of the blade
29
Figure 22. Detail of the neck of the blade on the whole model
9.1.3. MODAL ANALYSIS
The modal analysis was performed using the modal toolbox on ANSYS Workbench in order
to determine the undamped free vibration and the modes shapes of the model. The modal
toolbox is very similar to the structural analysis toolbox, with the only difference that it is not
possible to apply any type of loading when performing the modal analysis. This is realistic
since the undamped free vibration only depends on the mass and the stiffness of the structure.
Usually there are four steps when performing the modal analysis on ANSYS Workbench. The
first is to build the model, then choose analysis type and option, after that apply boundary
condition and solve and finally, review results.
In this case the model was already built which meant that choose the analysis type and options
were the second task to perform. In the analysis type, the material of the structure was
assigned, the model was meshed using a 10-node element and the software chose the mode
extraction methods. For this case, Block Lanczos setting was chosen as a default setting since
the Block Lanczos setting is recommended for most applications. This method consists of
substituting block algorithms for matrix block multiplies and block solvers for matrix-vector
products and simple solvers in unblocked algorithms. This setting was suitable since it has the
30
possibility of extracting larger number of modes in models, works well in rigid body, and it
can be used when the model is complex with a mixture of shell, solids and beam (ANSYS,
2004).
To apply the boundary condition was the third step. The boundary condition was applied as it
has been explained in section 8.2. The boundary condition was applied over the whole model
and not using the symmetry is because a symmetry boundary condition would result in the
shaped modes being symmetrical and consequently, some modes would be missing. After the
boundary conditions, the next step was to solve the analysis and review the results.
9.1.4. FATIGUE ANALYSIS
The fatigue life gives an indication of the remaining life for a specific fatigue analysis. Since
the loading was of constant amplitudes, the results would represent the remaining cycle until
the parts fail due to fatigue. Fatigue loading can be defined as the type of loading which
results in the cyclic variations in the applied stress on a component (Gopinath and Mayuram,
2014). The fatigue load will have different frequency depending on the rotational speed. For
instance running the fan at 3000 rpm, a frequency of 50 Hz will be obtained, for 2187 rpm
will give a frequency of 36.5 Hz, 1771 rpm will give a frequency of 29.5 Hz and finally 1594
gives a frequency of 26.6 Hz. When performing the analysis, the stress life method was used
in order to estimate the remaining life. Several analyses were performed using different
rotational speed (Fatemi, 2013). The life was estimated at certain point of interested which
include the whole blade, the neck of the blade and the fillet in front of the blade. When
performing the analysis, the stress life approach was chosen in order to estimate the remaining
life, obtaining the stresses which are required to use the S-N curves.
The analysis was performed in the same way that the previous analyses were done in order to
find the von Mises stress. However, in this case, there was an extra tool provided on the
software package called Fatigue tool. From the tool the life estimation of the model could be
estimated using the von Mises stresses obtained in the analysis. After that finding out the
loading type was the next step in order to perform the life estimation.
In this case, constant amplitude with proportional loading is the most suitable due to the fact
that this kind of loading has constant amplitude. Therefore, the loading helps to find out if the
load has a constant maximum value or values which change continually with time (ANSYS,
31
2004). The loading is said to be proportional also because of the fact that only one set of FE
results are needed.
9.2. ANALITICAL ANALYSIS
The analytical analysis was performed in order to get an estimated value of the different
analysis through the theory. Different theory models were used and some simplifications were
taken into account in order to get the results. Due to this, it is important to mention that the
results were an approximation. Moreover, some of the parameters and material properties
required were taken from the software SolidWorks and material catalogue.
This step is divided into five main parts: contact pressure, bending stress in the blade,
centrifugal stress and stress concentration around the hole.
9.2.1. CONTACT PRESSURE

In this case, the boss and the shaft are considered as cylinders which were rotating, with inner
pressure and massive axles with outer pressure. It was necessary to consider equations of
strength of material to get the results since they were influenced by multiaxial state of stress
and strain. That is, equilibrium, compatibility and the constitutive law equations were applied
to calculate the contact pressure. Figure 17 shows a sketch with a simplification of the model.
Steel Shaft
a
b=2a
Aluminium Hub
Figure 23. Sketch of the assembly between hub and boss
32
It was important to take into account that two different parts were required in order to get the
complete solution of each cylinder (Stigh, 2012). The first part was the solution of the
homogeneous differential equation, while the second one was the solution of the particular
solution. Finally, throughout the application of boundary conditions on r=a and r = b, the
four constants of integration could be determined (Stigh, 2012).
In Table 6 the material properties of the hub and the aluminum are shown. Here are some of
the material properties:
ALUMINUM HUB STEEL SHAFT
Young´s Modulus E1 = 70 GPa E2 = 206 GPa
Poisson´s Ratio ν1 = 0.3 ν2 = 0.3
Density ρ1 = 2770 kg/m3 ρ2 = 7780 kg/m3
Radius a = 55 mm b = 110 mm
Table 6. Material properties to take into account in the analysis
Regarding to the maximum and minimum diametric grip, it was calculated according to the
tolerances in order to obtain the value for the worst scenarios. The tolerance represents the
permissible limit or limits of variation in a measured value when the fan is manufactured. The
tolerances of the hub are class G7, which means minimum tolerance of +0.01 mm and
maximum tolerance of +0.04 mm. The tolerances of the shaft are class M6, which means a
minimum tolerance of +0.011 and a maximum tolerance of +0.03 mm. Therefore, based on
these tolerances, the maximum diametric grip is 3 µm, while the minimum is 1.1 µm.
ALUMINIUM HUB
On one hand, the radial displacement of the hub was determined by Equation (18) (Stigh,
2012):
K K @6 ∙ 1
#6 = #•6 + #•6 = ‘6 ∙ + ‘ ∙ − (1 − ν6 ) ∙ ∙ K 8 (18)
a a 8 ∙ J6
On the other hand, the radial stress was defined by Equation (19) (Stigh, 2012).
=’6 = =•’6 + =•’6 =

J6 ‘6 ‘ ∙a @6 ∙ 1
= ∙ “(1 + ν6 ) ∙ − (1 − ν6 ) ∙ ” − (3 + ν6 ) ∙ ∙ K (19)
1 − ν6 a K 8
33
Where w is the rotation speed, and the rest of parameters are defined in Table 6.
STEEL SHAFT
In a similar way as in the previous case, the displacement of the shaft is determined by
equation (20) (Stigh, 2012).
K K @• ∙ 1
# = #• + #• = ‘6 ∙ + ‘ ∙ − (1 − ν ) ∙ ∙ K 8 (20)
a a 8∙J
And the radial stress by Equation (21) (Stigh, 2012).
=’ = =•’ + =•’ =
J –6 – ∙a @ ∙1
= ∙ “(1 + ν ) ∙ − (1 − ν ) ∙ ” − (3 + ν ) ∙ ∙ K (21)
1−ν a K 8
BOUNDARY CONDITIONS
1. K = 0 → #6 (0) = 0
Inserting this boundary condition in equation (18) it is obtained the following term:
‘ ∙ ’ = 0 since it tends to ∞ when r tends to 0. Therefore:
˜ =™
2. K = š → =’ (š) = 0 (Free surface)
Replacing in equation (21).
J ‘6 ‘ ∙a @ ∙1
∙ “(1 + ν ) ∙ − (1 − ν ) ∙ ” − (3 + ν ) ∙ ∙ K = 0 (22)
1−ν a K 8
3. K = a
A representation of this boundary condition is shown in Figure 24. According to it:
=’6 (a) = −œ
Replacing in terms of rotational speed and pressure (20):
34
J6 ‘6 ‘ ∙a @6 ∙ 1
−œ = ∙ “(1 + ν6 ) ∙ − (1 − ν6 ) ∙ ” − (3 + ν6 ) ∙ ∙a
1 − ν6 a a 8
−• + {. žŸ ∙
˜ =
.¡ ∙ ™
Figure 24. Boundary condition for ( )
4. K = a
Similarly and based on the sketch of Figure 25:
=’ (a) = −œ
Figure 25. Boundary Condition for ( )
Replacing in terms of rotational speed and pressure (21):
35
J –6 – ∙a @ ∙1
−œ = ∙ “(1 + ν ) ∙ − (1 − ν ) ∙ ” − (3 + ν ) ∙ ∙a
1−ν a a 8
−œ = 5.3506459 ∙ 106 ∙ –6 − 2.8811 ∙ 106 ∙ – − 9.707981 ∙ 1 (23)
• Operating J:#aAcLg(22) − J:#aAcLg(23)
• + £. ∙
¢ =
. ∙ ™
• Operating š ∙ J:#aAcLg(22) − a ∙ J:#aAcLg(23):
™. ™™{ ∙ • + ™. ž ∙
¢ =
ž¡. Ÿ ∙ ™£
KINEMATICAL CONDITION
Finally, applying kinematical condition at K = a (Figure 26):
∆
= # (a) − #6 (a) (24)
2
Figure 26. Sketch of kinematical condition
The results of the contact pressure analysis for each rotational speed and worst scenarios are
shown in Table 7.
36
∆= { ¤z ∆= . ¤z
{™™™ ¥¦z 12 MPa 3.2 MPa
§™™ ¥¦z 12.4 MPa 3.5 MPa
ž{™ ¥¦z 12.7 MPa 3.8 MPa
¡§ ¥¦z 12.9 MPa 4.1 MPa
£Ÿ¡ ¥¦z 13.1 MPa 4.3 MPa
§§ ¥¦z 13.2 MPa 4.4 MPa
¨£ž ¥¦z 13.4 MPa 4.6 MPa

Table 7. Table of results of contact analysis
9.2.2. BENDING STRESS IN THE BLADE
The bending stress in the blade was calculated using Equation 25:
© ©
== = ∙ |>|^ 3 (25)
ªW «
where I is the moment of inertia, y is the distance from the centroid to the outer part of the
blade and M is the moment.
Some estimation was done to simplify the model and to calculate the moment. The moment of
inertia is taken from SolidWorks. Furthermore, the bending stress was calculated using the
pressure load. Although the pressure load was given in Pa, the load was converted into
Newton by using the following formula (Equation 26):
-
œ= (26)
‘
where F was the applied load and A the cross section area.
Finally, knowing that the force was applied as a uniformly distributed load in the blade
(Figure 27) and simplifying the blade as a cantilever beam, the moment could be calculated
by applying equilibrium. Therefore, the moment was given by Equation (27):
Œ
©=-∙ (27)
37
Figure 27. Sketch of the cantilever beam used to calculate the moment
Both the parameters required to calculate the bending stress in the lower of the blade and the
final result are shown in Table 8.
PARAMETERS VALUES
Force (N) 966.7
Moment (N ∙ m) 138.03
Moment of Inertia (m° ) 6.4 ∙ 10±²
Bending Stress Result (MPa) 5.8
Table 8. Table of results of bending stress analysis in the blade
9.2.3. CENTRIFUGAL STRESS
The centrifugal stress at the lower of the blade was calculated in order to compare the
analytical results with the FE solutions. The equation of the centrifugal stress was derived
from the basic stress formula shown in Equation (28).
-
== (28)
‘
where F is the reaction force and A is the cross section area at the bottom of the blade.
When expanding F using Newton´s Second Law (Equation 29) which is in the normal
direction
´ = µ × a_ (29)
38
Then, the linear acceleration could be converted into rotational acceleration by using Equation
(30)
a_ = K ∙ 1 (30)
where 1 is the rotational speed.
Therefore, replacing, the final centrifugal stress could be determined by Equation (31)
© ∙ KZ∙ 1
== (31)
‘
where M is the mass of the blade, rc is the distance from the center of mass to the outer part of
the blade, w is the rotational speed and A is the cross section area. Figure 28 shows a sketch
of the fan blade.
Figure 28. Sketch to calculate the centrifugal stress
In a similar way than in previous chapter, the parameters to calculate the centrifugal stress as
well as the centrifugal stress result are represented in Table 9.
39
PARAMETERS VALUES
Cross Section Area (m ) 2978 ∙ 10±²
Mass (kg) 2.4
Radius (m) 0.23
Rotational Speed (rpm) 3000
Centrifugal Stress (MPa) 79
Table 9. Table of results of the centrifugal analysis
9.2.4. STRESS CONCENTRATION AROUND THE WHOLE
Linear Elastic Fracture Mechanics was used in order to evaluate where the crack start to
propagate around the hole. The maximum stress around the hole was equal to 94 MPa and
knowing that the stress intensity was given by Equation 32:
¸¹ = =… ∙ √» ∙ a ∙ ?8 (¼) (32)
where ¼ = ’N .
From the above equation, the literature suggested an initial crack size value of 20 µm
(Chattopdhyay, 2008). Some basic assumption was made in which a 2D case was chosen over
3D. By using this assumption, automatically the shear stress was not neglected.
From this, s was calculated to be equal to 0.003 which made it valid since the value is
between the ranges proposed by Tada Data provided by Anderson (2005). From the theory,
based on Anderson (2005), the crack will start propagating when the stress intensity is equal
to the fracture toughness, which means that KI=KIC.
In Table 10, the parameters to calculate where the crack start to propagate are represented
PARAMETER VALUE
Initial Crack (μm) 20
f8 (s) 0.003
Fracture toughness (MPa ∙ √m) 36
Crack Start Propagating at (GPa) .ž
Table 10. Table of results of crack size around the hole analysis
40
10. RESULTS
The results obtained throughout the software analyses are shown in the next chapters.
10.1. FILLET SIZE OF THE BLADE
From Table 11 to Table 17, the results with tabulated format of the different variation of the
blade fillet size are shown according to Abdullah and Schlattmann (2012). For each rotational
speed and pressure, the parameters to take into account are: the maximum stress, the stress in
the fillet of the blade and the stress in the neck of the blade. Also, in Figure 29 an analysis of
the whole blade is shown. All the stresses obtained in the analyses were von Mises stresses.
Figure 29. Stress analysis in the whole blade
41
Max. Stress Neck of the
Rotational Blade Fillet
(MPa) Blade
speed Stress
(Inside the Stress
(rpm) (MPa)
boss) (MPa)
3000 356 31.9 27.0
2700 302 27.6 23.3

Blade Fillet Size
2430 240 21.3 17.6

40 mm
2187 197 15.4 14.3
1968 157 10.9 11.1
1771 121 9.6 9.8
1594 110 8.1 7.6
Table 11. Table of results for a blade fillet size of 40 mm

(MPa) Blade
speed Stress
(Inside the Stress
(rpm) (MPa)
boss) (MPa)
3000 368 31.8 26.4
2700 301 27.3 22.5

Blade Fillet Size
2430 241 26.3 17.0

35 mm
2187 192 15 12.2
1968 158 11.1 10.2
1771 121 10.6 8.6
1594 104 7.2 7.4
42
(MPa) Blade
speed Stress
(Inside the Stress
(rpm) (MPa)
boss) (MPa)
3000 351 27.6 27.2
2700 296 21.1 22.6

Blade Fillet Size
2430 230 18.6 17.12

30 mm
2187 191 14.1 12.8
1968 151 11.5 8.9
1771 116 7.8 8.8
1594 99.1 7.1 7.4
Table 13.Table of results for a blade fillet size of 30 mm

(MPa) Blade
speed Stress
(Inside the Stress
(rpm) (MPa)
boss) (MPa)
3000 367 25.7 26.5
2700 308 28.5 23.3

Blade Fillet Size
2430 241 24.7 17.4

26.5 mm
2187 201 17.2 13.5
1968 158 12.9 11.8
1771 121 8.4 9.5
1594 104 7.2 7.5
Table 14. Table of results for a blade fillet size of 26.5 mm
43
(MPa) Blade
speed Stress
(Inside the Stress
(rpm) (MPa)
boss) (MPa)
3000 367 28.1 26.9
2700 316 27 24.5

Blade Fillet Size
2430 241 25.7 17.2

20 mm
2187 206 20.0 14.5
1968 158 11.9 11.5
1771 121 8.7 9.3
1594 104 7.7 7.8

(MPa) Blade
speed Stress
(Inside the Stress
(rpm) (MPa)
boss) (MPa)
3000 366 31 31
2700 319 28.8 26.1

Blade Fillet Size
2430 240 21.1 17.4

15 mm
2187 242 18.9 15.7
1968 158 16.9 13.6
1771 121 9.4 8.8
1594 103 7.5 7.5
44
(MPa) Blade
speed Stress
(Inside the Stress
(rpm) (MPa)
boss) (MPa)
3000 464 30 27.6
2700 360 29.6 24.5

Blade Fillet Size
2430 304 22.7 17.8

10 mm
2187 258 20.5 14.9
1968 200 17.5 11.8
1771 124 14.6 9.5
1594 114 11.3 8.5
After these results, in order to have both a better understanding and easier way to compare the
results, from Figure 30 to Figure 35 graphs comparing the four parameters against the blade
fillet size and each rotational speed (horizontal axis) are plotted. It is important to notice that,
from the results obtained in the analysis, the maximum stress occurred at the same point (the
boss) while using different blade fillet size. Moreover, all the values are converged.
Figure 30. Maximum stress against blade fillet size for each rotational speed
45
s
Figure 31. Stress in the fillet against blade fillet size for each rotational speed
Figure 32. Stress in the neck of the blade against blade fillet size for each rotational speed
46
Figure 33. Maximum stress against rotational speed for each blade fillet size
Figure 34. Stress in the fillet against rotational speed for each blade fillet size
47
Figure 35. Stress in the neck of the blade against rotational speed for each blade fillet size
10.2. DIAMETER OF THE NECK OF THE BLADE
In Table 18, the results for each parameter of interest are given in a tabulated and graph
format (Abdullah and Schlattmann, 2012).
Max. Stress
Stress Stress Neck of
Analyses Ω (MPa)
Blade fillet the Blade
Number (rpm) (Inside the
(MPa) (MPa)
boss)
1 3000 472 84 32.5
2 2700 383 68 26
3 2430 309 53 21
4 2187 251 44 19
5 1968 203 35.3 14
6 1771 166 29 10.8
7 1594 133 23 9.10
8 (49 mm) 2187 218 43.0 19
9 (51 mm) 3000 407 82 35
10 (47 mm) 1594 155 23 10.7
Table 18. Table of results of the parametric study in the neck of the blade
48
After presenting the result in a table, in a similar way as in the previous chapter, from Figure
36 to Figure 38 graphs are plotted for each scenario. When plotting the graph, the horizontal
axis was labeled as rotational speed while the vertical axis was labeled as the von Mises stress
in different points. The legend specifies that there are two types of graphs where the first one
represents the stress as per the initial design, where initial design means the preliminary
design which was provided by the company. The other line represents the change of the
parameters. Table 18 shows all the tabulated values of each parameter used in the neck of the
blade and the value of the preliminary design (26.5 mm) as per the company’s drawings.
8 Rotational speed vs Maximum stress

x 10
5.5
Maximum stress as per initial design
5 Maximum stress due to the change of the neck of the blade
4.5
4
Maximum stress (Pa)
3.5
2.5
1.5
1
1500 2000 2500 3000
Rotational speed (rpm)
Figure 36. Maximum stress against rotational speed
49
7 Rotational speed vs stress in the fillet
x 10
9
stress at the (BF) as per initial design
stress at the (BF) due to the change of the neck of the blade
8
Stress at the blade fillet (BF) (Pa)
2
1500 2000 2500 3000
Figure 37. Stress at the blade fillet against rotational speed
7 Rotational speed vs Neck of the blade

x 10
3.5
3
Neck of the blade (NB) (Pa)
2.5
1.5
1
stress at the (NB) as per initial design
stress at the (NB) due to the change of the neck of the blade
0.5
1500 2000 2500 3000
Figure 38. Stress at the neck of the blade against rotational speed
50
10.3. CONVERGENCE
Convergence both on the fillet of the blade and the neck of the blade is plotted in Figure 39
and Figure 40. Furthermore, in Table 19 and Table 20, either percentage of change, or number
of elements, or number of nodes and element size taken into account are shown respectively.
Figure 39. Convergence on the fillet of the blade
VON MISES ELEMENT

CHANGE (%)
STRESS (Pa) SIZE (m)
1 . ž¡ ∙ ™Ÿ - 0.01018
2 . £Ÿ ∙ ™Ÿ £. {¨¨¡ 0.001002
3 . £¡ ∙ ™Ÿ ™. Ÿ§¨§ 0.000157
Table 19. Table of results of the convergence analysis on the blade fillet
51
Figure 40. Convergence on the neck of the blade
VON MISES ELEMENT

CHANGE (%)
STRESS (Pa) SIZE
1 ¨. { ∙ ™§ - 0.0250
2 ¨. žž ∙ ™§ 2.47 0.0025
3 ¨. ¨™™¨ ∙ ™§ . 0.000248
Table 20. Table of results of the convergence analysis on the neck of the blade
52
10.4. MODAL ANALYSIS RESULTS
The results of the natural frequencies obtained through the modal analysis are shown in Table
21. Furthermore, in Figure 41 is represented a graph plotting the mode numbers against the
Eigen frequencies. Also, it is compared the coarse mesh against the fine mesh (Figure 42 to
Figure 44).
FREQUENCY (Hz) FREQUENCY (Hz)

MODE
(Fine mesh) (Coarse mesh)
1 16.8 17
2 63.2 64.9
3 63.3 65
4 67.3 68.8
5 67.5 69.6
6 67.9 69.6
Table 21. Table of results of modal analysis
Mode number vs Eigenfrequency

70
60
50
Eigenfrequency (Hz)
40
30
Modal analysis using coarse mesh
Modal analysis using fine mesh
20
10
1 2 3 4 5 6
Mode number
Figure 41. Mode number against natural frequencies and coarse mesh against fine mesh
53
Figure 42. Picture of the fine mesh on the whole blade
Figure 43. Detail of the fine mesh on the blade
54
Figure 44. Picture of the coarse mesh of the blade
10.5. FATIGUE ANALYSIS
In a similar way than in previous analyses, the constant amplitude load is represented in
Figure 45. The fatigue loading cycle is caused by the variable stress at a point with respect to
the time (Gopinath and Mayuram, 2014). The stress at a point is varied due to the rotation of
the fan blades. The origin of the fatigue can be due to the geometry change where stress
concentration may occur. This will result in a fatigue crack which will occur at the
discontinuities in the material (Budynas, 2006). The other origin of fatigue may also be in the
microstructure of the material which may contain voids (Findlay and Harrison, 2002).
Furthermore, the relation between the rotational speeds against the number of cycles is
represented in Figure 46 and finally pictures of the fillet of the blade and the neck of the blade
are represented in Figure 47 and Figure 48 where it is shown the remaining cycle before the
part fails due to fatigue. Table 22 shows the results of the fatigue analysis where N represents
the number of cycles where cycle represents one complete turn of the axial fan.
55
ROTATIONAL BLADE NECK OF THE FILLET OF
SPEED (rpm) (N) BLADE (N) THE BLADE (N)
3000 10000 4704 67972
2187 1.2 ∙ 10² 2.9 ∙ 10² 2.9 ∙ 10Ã
1771 6.8 ∙ 10Ã 5.6 ∙ 10Ã 1 ∙ 10ˆ
1594 1 ∙ 10ˆ 1 ∙ 10ˆ 1 ∙ 10ˆ
Table 22. Table of results of fatigue analysis
Figure 45. Constant amplitude load obtained from ANSYS analyses (Fatemi 2013)
56
Rotational speed vs Number of cycles
3000
Number of cycles in the blade
Number of cycle in the neck of the blade
Number of cycle inthe fillet of the blade
Rotational speed (RPM)
2187
1771
1594
2 4 6 8 10
Number of cycle 7
x 10
Figure 46. Rotational speed against Number of cycles, where N stands for the number of
cycle
Figure 47. Fatigue Analysis on the Fillet of blade at 1594 rpm
57
Figure 48. Fatigue Analysis on the neck of the blade at 1594 rpm
11. ANALYSIS
MAXIMUM STRESS
When analyzing the results obtained in the FE analysis, the area where the maximum stress
occurs was studied. This very small area was located in the fillet of the blade and the fillet of
the boss (Figure 49, Figure 50 and Figure 51). Therefore, it can be stated that this area was
affected by stress singularities. As mentioned in the literature review, these singularities are
caused by the discretization error during the mesh generation.
58
Figure 49. Maximum stress on the fillet of the boss
Figure 50. Detail of maximum Stress on the fillet of the boss
Figure 51. Maximum stress in the fillet of the blade
59
As the previous researches suggest, these stresses can be ignored and deleted, changing the
measure scale. This thesis took into account these considerations. The results of the maximum
stresses were:
Evaluating the initial design (neck of the blade size equal to 45 mm and blade fillet size equal
to 26.5 mm), using a rotational speed higher than 1771 rpm with its respective pressure loads,
the maximum stress was found to be higher than 203 MPa. Therefore, by using the safety
factor equation and reminding that the ultimate tensile strength is equal to 310 MPa, it can be
checked that SF < 2. This safety factor did not take into consideration fatigue and the analysis
was performed using a linearly elastic material. Thus, these scenarios will not be acceptable.
Then, just the analysis 6, which corresponds to a rotational speed of 1771 rpm and Analysis 7,
corresponding to a rotational speed of 1594 of pressure load, are acceptable, since the SF ≥ 2.
On the neck of the blade, similar conclusions can be stated. Values for rotational speed higher
than 1771 rpm are out of range based on the SF. Therefore, these analyses do not meet the
requirements. Then, considering only acceptable analyses, and checking Figure 36, where a
comparison between preliminary design and modifications are plotted, it may be stated that
increment in the diameter of the neck of the blade will provoke an increment in the maximum
stress. Therefore, it can be suggested that the initial design is the optimal design for a
rotational speed of 1771 rpm or lower values according to maximum stress.
Similar conclusions are obtained from the parametric study of the blade fillet. Remaining
constant the diameter of the neck of the blade (initial design), values higher than 1771 rpm are
out of range with respect to the SF. Furthermore, comparing Figure 30, it can be noticed that a
size of 30 mm in the blade fillet, will cause the lowest value for the maximum stress.
STRESS IN THE FILLET OF THE BLADE
Although all the analyses are acceptable since the SF requirement is fulfilled for the rotational
speed and pressure load (Figure 34 and Figure 37), slightly higher stresses are found
comparing to the initial design, increasing proportionally to the increase of the neck, as it is
represented in Figure 37. Hence, initial design for the neck of the blade is the most
appropriated design (45 mm).
60
Regarding the modifications in the fillet of the blade, according to Figure 31, lower stresses
are caused when a fillet size of 30 mm is used, which suggests that optimal design appears,
again, when this size is utilized.
STRESS IN THE NECK OF THE BLADE
Finally, the stresses obtained in the neck of the blade achieve the requirement of the SF for all
the rotational speed and the pressure load (Figure 35 and Figure 38). Similarly than in
previous cases, the increase of the diameter of the neck will cause an increase of the stresses.
Therefore, the initial design with 45 mm is the appropriated according to Figure 38.
However, the variation of the size of the fillet blade is different than in previous cases.
According to Figure 32, the size of 35 mm has the lower stresses, followed by the 30 mm
size.
It may be noticed that the parameters of the neck of the blade chosen offer a slightly higher
stresses compared to the preliminary design. This can be explained by the fact that a
continuous increased of the diameter in the neck of the blade will generate a higher stress in
the neck of the blade, since the mounting surface on the hub would decrease.
CONVERGENCE ANALYSIS
According to Figure 39 and Figure 40, convergence is achieved both in the fillet of the blade
and in the neck of the blade, since the percentage of error is less than 2% (Table 20 and Table
21). That implies the results can be trusted.
MODAL ANALYSIS
From the results obtained the operating rotational speeds are 1500 to 2000 rpm which lies
between the first and the second eigenfrequency. The operating rotational speeds are not close
to any of the first two eigenfrequencies. However there may be problems during start up when
passing the first natural frequency. Figure 41 shows all the natural frequency and their mode.
This natural frequency may change if other factors are taken into account, for instance,
aerodynamics, damping and a couple of axial fan operating next to one another.
61
FATIGUE ANALYSIS
From the results obtained, it could be noticed that the higher the rotational speed, the less is
the number of cycle (N), as shown in Figure 46, since the rotational speed increases the stress
and this will result in reducing the life of the fan. When plotting the graph it could be noticed
that the graph were also following the trend of the S- N graph for aluminium.
Based on the results obtained from both the stress analysis and the life estimation, it is not
advisable to run the fan at full speed of 3000 rpm. This will result in early fatigue failure of
the components in the axial fan since the number of cycle fall under Low Cycle Fatigue.
Although running the fan at 2187 rpm could also be ideal since the number of cycle falls
under the high cycle fatigue, the rotational speed does not meet the safety factor requirement.
COMPARISON OF THE ANALYSTICAL AND THE EMPIRICAL SOLUTION
The first comparison is performed on the bending stress. The analytical calculations were
performed in order to compare with the result obtained from the FE-solutions. The first
comparison was performed in one quarter of the geometry where the structure was subjected
only to pressure load of 9000 Pa. The maximum bending stress at the bottom of the blade was
equal to 18 MPa where a stress of 5.8 MPa was obtained from the analytical calculation.
The second comparison was performed on the centrifugal force. The analytical calculation
was performed in order to compare with the FE solutions. When performing the analytical
calculation, a centrifugal stress of 79 MPa was found. An analysis was performed taking into
account the centrifugal load. A stress of 90 MPa was found at the lower bottom of the blade.
The stress comparison was performed between the FEM analysis and the analytical solution.
This comparison was performed in the stress distribution in the blades. From the analytical
analysis the blade was treated like a rectangular box which had a thickness of 3 mm. After the
analysis the stress distribution of both the analytical and the FE solution was plotted (Figure
52 and Figure 53). It could be noticed that from the analytical solution that the stress
distribution was following a pattern where the maximum stresses were in the top edge and the
in the FE solution maximum stress were only in one edge. This could be explained by the
simplification performed and also due to the curvature on the airfoil and the twisting of the
blade. From the results obtained in the stress distribution, it could be concluded the analytical
solution results is quite close to the FE solution and it could also be noticed that the maximum
62
stress result from the approximation of the FE solutions is quite close to the analytical
solution.
Figure 52. Approximation of FE solution of the bending stress (Pa)
Figure 53. Bending stress analytical solution (Pa)
63
12. CONCLUSIONS
As a general conclusion, from the above results, it could be noticed from the preliminary
design that an increase in the rotational speed together with the pressure load will cause an
increase in the maximum stress, stress in the blade fillet and stress in the neck of the blade.
Bearing in mind that a safety factor of 2 was the pre-requisite, and taking into account that the
safety factor is given as the ratio between the ultimate tensile strength over the design load or
the maximum stress, the following conclusions can be stated:
With respect to the parametric study, it can be suggested that the required operating speed
should be between 1500 rpm to 2000 rpm. The size of the diameter of the neck of the blade
should be 45 mm, while the size of the fillet of the blade should be between 30 mm and 35
mm. Also, according to the parametric study, including the reduction of the material with
respect to the old design, the new design is acceptable according to the mechanical
requirements explained along this dissertation.
All these results can be trusted according to the values obtained throughout the convergence
analysis, where the prerequisite of percentage of change less than 2% was fulfilled.
Regarding to the modal analysis, as mention in chapter 9, all the natural frequency maybe of
interested. However when taking into account the suggested rotational speeds, it could be said
that the first two natural frequency might be of interested and there may be problem at start up
since the recommended rotational speeds are between the first and second natural frequencies.
With respect to the fatigue analysis, evaluating the results obtained in the interested area, it
can be suggested that the design meets the requirements of use if it operates at 1771 and 1594
rpm since the number of cycle falls under the High Cycle Fatigue and both rotational speed
fulfilled the requirement of the Safety Factor. It can be also stated that the constant amplitude
graph obtained during the analysis implies full reverse condition, that is, R=-1.
64
13. FUTURE WORKS
As suggestions, further works would need to be done in order to get more accurate results.
Computational fluid dynamics (CFD) analysis should be performed in order to get a better
performance of the fan. A CFD analysis will also help in order to obtain the vibration loads
acting on the fan blades. The aerodynamics load like the lift force or lift coefficient should
also be provided in order to obtain results close to reality.
The use of composite material may also be suggested when designing the axial fan, since the
literature also shows that some axial fan are being designed using composite materials.
Obviously the pros and cons would need to be investigated in order to determine which
material would be best suited for the current type of application.
If the prototype is built, it can also be recommended to put strain gauges on the fan blades.
The results can be used in order to compare experimental results to the FE solutions.
It can also be suggested to the manufacturing section, the design should be as smooth as
possible in order to avoid stress concentration.
65
14. REFERENCES
• Abdullah, O. I. & Schlattmann, J., 2012. Stress Analysis of Axial Flow Fan. Adv.
Theor. Appl. Mech., Volumen 5, pp. Nº6, 263-275.
• Agricola, G., 1912. De Re Metallica. London: s.n.
• Amoo, L., 2013. On the design and structural analysis of jet engine fan blade
structures. Progress in Aerospace Sciences, November, Volumen 60, pp. 1-11.
• Anders, J. M., Haarmeyer, J. & Heukenkamp, H., 2009. A Parametric Blade Design
System (Part I + II), s.l.: BMWRolls-Royce GmbH, Dahlewitz, Germany.
• Anderson, T. L., 2005. Fracture mechanics fundamentals and applications. 5 ed.
s.l.:Taylor & Francis.
• Anon, 2000. The Basics of Axial Flow Fans. Texas: MCDermott Company.
• Anon, 2001. Introduction to Finite Element Analysis or Finite Element Method. [En
línea]
Available at: http://www.engr.uvic.ca/~mech410/lectures/FEA_Theory.pdf
[Último acceso: 20 March 2014].
• Anon, 2014. Fundamentals of Vibration. [Online]
Available at: http://www.newport.com/Fundamentals-of-
Vibration/140234/1033/content.aspx
[Accessed 11 February 2014].
• ANSYS, I., 2004. ANSYS Inc. PDF Documentation for Release 14.5, s.l.: ANSYS,
Inc..
• Arewar, A. & Bhope, D. V., 2013. Stress analysis of axial flow fan impeller.
International Journal of Engineering Research and Applications, Volumen 3, Issue 1,
pp. 2086-2090.
• Barlow, K. & Chandra, R., 2005. Fatigue crack propagation simulation in an aircraft
engine fan blade attachment. Naval Air Systems Command, Propulsion and Power
Engineering.
• Budynas, R. G. & Nisbett, K., 2006. Shigley’s Mechanical Engineering Design. 8 ed.
s.l.:Mc Graw Hill.
• Caicedo, S., 2007. Diseño y fabricación de un ventilador axial para un túnel de
viento, Bogota D.C.: Universidad de los Andes.
66
• Chambon, L., 2001. A unified fracture mechanics approach to fretting fatigue crack
growth., Massachusetts: Massachusetts Institute of Technology.
• Chattopdhyay, S., 2008. Design fatigue curves based on small crack growth and crack
closure. Journal of applied science and engineering technology.
• Ciavarella, M. & Demelio, G., 2001. A review of analytical aspects of fretting fatigue,
with extension to damage parameters, and application to dovetail joints. International
Journal of, Volumen 38, pp. 1791-1811.
• Cook, R., Malkus, D. S., Plesha, M. E. & Witt, R. J., 2002. Concepts and Applications
of Finite Element Analysis. 4 ed. s.l.:Rev. ed. Wiley.
• Cory, W., 2005. Fans and Ventilation: A practical guide. 2 ed. s.l.:Elsevier.
• Dixit, U., 2009. Finite Element Method: An introduction. En: C. L. India, ed. Finite
Element Methods For Engineers. Guwahati: Department of Mechanical Engineering,
Indian Institute of Technology.
• Fatemi, A., 2013. Fatigue tests and Stress- Life (S-N) approach, Toledo: University of
Toledo.
• Findlay, S. J. & Harrison, N. D., 2002. Why aircraft fail?. Materials today, pp. 18-25.
• Garcia, I., 2012. Aerodynamic analysis of axial fan unsteady simulations, s.l.:
University of de Basque Country.
• Golden, P. & Naboulsi, S., 2012. Hybrid contact stress analysis of a turbine engine
blade to disk attachment. International Journal of Fatigue, February, Volumen 42, pp.
296-303.
• Gopinath, K. & Mayuram, M. M., 2014. Fatigue consideration in design. Indian
Institute of Technology Madras, pp. 1-33.
• Ohman, H. & Singhal, R. K., 1993. Modal Analysis of an axial flow fan blade. 11th
International Modal Analysis Conference, pp. 1294-1298.
• Poursaeidi, E. & Salavation, M., 2007. Failure analysis of generator rotor fan blades.
Engineering Failure Analysis, pp. 851-860.
• Save, C., 2013. Type of contact in ANSYS. [Online]
Available at:
https://sites.google.com/site/mechanicalengineeringforlife/FEA/typesofcontactsinansy
s
[Accessed 20 March 2014].
67
• Schildhauer, M. & Minges, M. V., 2010. Development and structural of axial flow
fan. Mech CaL, pp. 1-3.
• Singh, O., Khilwani, R., Sreenivasulu, T. & Kannan, M., 2011. Parametric study of
centrifugal fan performance: experiments and numerical simulations. International
Journal of Advances in Engineering & Technology, 1(2), pp. 33-50.
• Stigh, U., 2012. Cylindrical bodies, s.l.: University of Skövde.
• Sullivan, A., 2006. Aerodynamic forces acting on an aerofoil. Ohio: College of
Wooster, Department of physics.
• Totten, G. & MacKenzie, S., 2003. Handbook of Aluminum: Physical Metallurgy and
Processes. 1 ed. New York: Marcel Dekker, Inc..
• Tsai, G., 2004. Rotating vibration behavior of the turbine blades with different groups
of blades. Journal of Sound and Vibration, Volumen 271, pp. 547-575.
• Wei, D., Yuan, S. H. & Wang, Y. R., 2012. Failure analysis of dovetail assemblies
under fretting load. Engineering Failure Analysis, pp. 381-396.
• Zare, A. et al., 2013. Finite Element Analysis of Axial Fan Blade with Different Chord
Lengths. Indian Journal of Science and Technology, May, Volume 6 (5), pp. 4403-
4405.
68

Analysis of Fan Blade

Uploaded by

Copyright:

Available Formats

Analysis of Fan Blade

Uploaded by

Document Information

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

Analysis of Fan Blade

Uploaded by

Copyright:

Available Formats

What is the purpose of the thesis?

What is the purpose of the thesis?

What methods are used in the analysis?

What methods are used in the analysis?

ANALYSIS OF FAN BLADE

Master Degree Project in Applied Mechanics

Supervisor: Dr. Karl Mauritsson

4. GOAL AND PURPOSE ...................................................................................................... 6

5. LITERATURE STUDY OF THE LOADS ........................................................................ 6

5.1. AERODYNAMICAL LOADS .......................................................................................... 6

5.1.1. LIFT ................................................................................................................................ 7

5.1.2. DRAG ............................................................................................................................. 7

5.1.3. SKIN FRICTION FORCE ............................................................................................... 8

5.2. MECHANICAL LOADS .................................................................................................. 9

5.3. STRESS CONCENTRATION ....................................................................................... 10

6.1. PARAMETRIC STUDY ................................................................................................. 10

6.2. THE FINITE ELEMENT METHOD ........................................................................... 11

6.3. MODAL ANALYSIS ...................................................................................................... 13

6.4. CONVERGENCE ANALYSIS ...................................................................................... 13

6.5. FATIGUE ANALYSIS ................................................................................................... 15

8.2. BOUNDARY CONDITIONS ......................................................................................... 21

8.3. MESH ............................................................................................................................. 22

8.4. CONTACT COUPLING BETWEEN PARTS .............................................................. 23

9.1 SOFTWARE ANALYSIS ................................................................................................. 25

9.1.1. ENGINEERING CONSIDERATIONS .......................................................................... 25

9.1.2. PARAMETRIC STUDY ................................................................................................ 26

9.1.3. MODAL ANALYSIS ...................................................................................................... 30

9.1.4. FATIGUE ANALYSIS ................................................................................................... 31

9.1 ANALYTICAL ANALYSIS .............................................................................................. 25

9.2.1. CONTACT PRESSURE ................................................................................................ 32

9.2.2. BENDING STRESS IN THE BLADE ........................................................................... 37

9.2.3. CENTRIFUGAL STRESS ............................................................................................. 38

9.2.4. STRESS CONCENTRATION AROUND THE WHOLE ............................................... 40

10. RESULTS .......................................................................................................................... 41

10.1. FILLET SIZE OF THE BLADE .................................................................................. 41

10.2. DIAMETER OF THE NECK OF THE BLADE .......................................................... 48

10.3. CONVERGENCE .......................................................................................................... 51

10.5. FATIGUE ANALYSIS ................................................................................................... 55

11. ANALYSIS ........................................................................................................................ 58

12. CONCLUSIONS ............................................................................................................... 64

13. FUTURE WORKS ............................................................................................................ 65

14. REFERENCES ................................................................................................................. 66

Figure 1. Sketch of Impeller PFJ1 and components .................................................................. 2

Cf = Skin friction coefficient

FD = Force due to drag

Fs = Force due to skin friction

KI = Stress intensity factor

TE= Trailing edge

σcen = Centrifugal stress

[K] = Stiffness matrix

[M] = Mass matrix

C, M = Fatigue Parameters (materials constants)

= Crack growth rate

∆K = range of the stress intensity factor

FEA-Finite Element Analysis

FEM- Finite Element Method

UTS- Ultimate tensile strength

SF- Safety Factor

HCF- High Cycle Fatigue

LCF- Low Cycle Fatigue