Sie sind auf Seite 1von 57

2014/15

Finite Element Analysis Report

By: Steven Goddard


Student Number: 10038749
Course: Mechanical Engineering (PT)
Student Number: 10038749 Finite Element Analysis Report Page 2 of 57

Executive Summary
This assignment details my investigations into the design and analysis of a composite pressure
vessel. An initial theoretical investigation is undertaken to determine an initial ply layout that
obtains a minimum safety factor of 1.2 based on common composite failure criteria. This initial ply
layout is used in Abaqus 6.14 and an FEA model is developed in order to be used a benchmark for
comparisons against various further investigations.

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 3 of 57

Contents
Introduction ...................................................................................................................................................................... 6
1. Background Reading and Research........................................................................................................................... 7
1.1. Composite Pressure Vessels (CPV) .................................................................................................................... 7
1.2. Filament Winding .............................................................................................................................................. 7
1.3. Optimum Winding Angle................................................................................................................................... 9
2. Theory of Thin Wall Pressure Vessels ................................................................................................................. 11
3. Calculating the Ply Layout ................................................................................................................................... 12
4. Abaqus® ................................................................................................................................................................... 13
4.1. Preparing the Abaqus® Model ......................................................................................................................... 13
4.2. Isotropic Model ............................................................................................................................................... 17
4.3. Composite Model Preparation........................................................................................................................ 18
4.4. Mesh Study ..................................................................................................................................................... 20
4.4.1. Global Mesh Study ...................................................................................................................................... 20
4.4.2. Local Mesh Study ........................................................................................................................................ 24
4.5. Design Analysis & Iterations............................................................................................................................ 27
4.5.1. Method of Data Analysis ............................................................................................................................. 27
4.5.2. Initial Run .................................................................................................................................................... 28
4.5.3. Iteration 1.................................................................................................................................................... 31
4.5.4. Iteration 2.................................................................................................................................................... 33
4.5.5. Iteration 3.................................................................................................................................................... 35
4.5.6. Iteration 4.................................................................................................................................................... 37
4.6. Investigation into Different Positions for the Concrete Support .................................................................... 39
4.7. Investigation into the Effects of Connecting Inlet & Outlet Tubes ................................................................. 41
4.8. Investigation into the Effects of Weight of the Vessel.................................................................................... 43
4.9. Investigation into the Effects of Environmental Temperature Change .......................................................... 47
5. Recommendations for Further Work ...................................................................................................................... 49
6. Conclusion ............................................................................................................................................................... 49
References ...................................................................................................................................................................... 50
Bibliography .................................................................................................................................................................... 51
Appendix 1 ...................................................................................................................................................................... 52

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 4 of 57

Figures
Figure 1 – Carbon Fibre CPV (McLaughlan et al, 2011)..................................................................................................... 7
Figure 2 - Drum Type Resin Bath (Miaris, 2010) ............................................................................................................... 8
Figure 3 - Fabrication of Filament Wound CPV with Hoop and Helical Layers (McLaughlan et al, 2011) ........................ 9
Figure 4 - Body Diagram of Axial and Hoop Forces and Internal Pressure (Sulaiman et al, 2013) ................................. 10
Figure 5 - Ply Stack Layout [+/-55]s with 1.1mm Layer Thickness .................................................................................. 12
Figure 6 - Sketcher .......................................................................................................................................................... 13
Figure 7 - Create/Edit Material ....................................................................................................................................... 14
Figure 8 - Create Composite Layup ................................................................................................................................. 14
Figure 9 - Step Creation .................................................................................................................................................. 15
Figure 10 - Field Output Request .................................................................................................................................... 15
Figure 11 - Boundary Conditions..................................................................................................................................... 16
Figure 12 - Create Loads ................................................................................................................................................. 16
Figure 13- Isotropic FEA Results...................................................................................................................................... 17
Figure 14 - Isotropic Pressure Vessel Mesh Study .......................................................................................................... 17
Figure 15 - Add a Ply Layup ............................................................................................................................................. 18
Figure 16 - Create Lamina Type Material ........................................................................................................................ 19
Figure 17 - Forming the Outlet Hole ............................................................................................................................... 19
Figure 18 - Global Mesh Study – Checking Areas (Red) .................................................................................................. 20
Figure 19 - Global Mesh Study Iterations ....................................................................................................................... 21
Figure 20 - Mesh Study - Global Convergence ................................................................................................................ 22
Figure 21 - Global Mesh Study - Tri Mesh Iterations ...................................................................................................... 23
Figure 22 - Tri Mesh Study - Global Convergence ........................................................................................................... 23
Figure 23 - Bias Seed Distribution ................................................................................................................................... 24
Figure 24 - Local Mesh Study - Inlet Checking Points ..................................................................................................... 24
Figure 25 - Local Mesh Study - Inlet Iterations ............................................................................................................... 24
Figure 26 - Inlet Hole - Local Convergence ..................................................................................................................... 25
Figure 27 - Outlet Hole Mesh Refinement ...................................................................................................................... 26
Figure 28 - Concrete Support Mesh Refinement ............................................................................................................ 26
Figure 29 - Results > Field Output ................................................................................................................................... 27
Figure 30 - Excel Import Wizard ...................................................................................................................................... 27
Figure 31 - Initial Run - Deformed S11 Stress Plot .......................................................................................................... 28
Figure 32 – Initial Run - S11 Outlet View (Ply 1) ............................................................................................................. 29
Figure 33 – Initial Run - S11 Inlet View (Ply 1) ................................................................................................................ 29
Figure 34 – Initial Run - S22 Outlet View (Ply 1) ............................................................................................................. 29
Figure 35 – Initial Run - S12 Whole View (Ply 1) ............................................................................................................. 30
Figure 36 - High Stress Areas (Typical All Plies) .............................................................................................................. 30
Figure 37 - Iteration 1 - Strengthened Regions ............................................................................................................... 31
Figure 38 - Iteration 1 Ply Stack Plot ............................................................................................................................... 31
Figure 39 - Iteration 1 - S11 Whole View (Ply 1) ............................................................................................................. 32
Figure 40 - Iteration 1 – Main Body Investigation (Ply 1) ............................................................................................... 33
Figure 41 - Iteration 2 - S11 Whole View (Ply 1) ............................................................................................................. 33
Figure 42 - Iteration 3 - Increased Ply Layup................................................................................................................... 35
Figure 43 - Iteration 3- S11 Whole Model (Ply 1) ........................................................................................................... 35

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 5 of 57

Figure 44- Iteration 3 – Element with Maximum S22 – Edge of Increased Ply Thickness .............................................. 36
Figure 45 - FOS Trend ...................................................................................................................................................... 37
Figure 46 - Iteration 4 - Strengthening Area of Maximum S22 ....................................................................................... 37
Figure 47 - Iteration 4 - S11 Whole View (Ply 1) ............................................................................................................. 38
Figure 48 - Modifying the Geometry of the Concrete Supports ..................................................................................... 39
Figure 49 – Distance of Concrete Supports - Safety Factors ........................................................................................... 40
Figure 50 - Comparison of Stress Contours when Moving Concrete Supports .............................................................. 40
Figure 51 - Example of Solid Pipe Fitting (Tryer Process Equipment) ............................................................................. 41
Figure 52- Inlet and Outlet Tubes - S11 (Ply 1) ............................................................................................................... 42
Figure 53 - Inlet and Outlet Tubes - S22 (Ply 1) .............................................................................................................. 42
Figure 54 - Inlet and Outlet Tubes - S12 (Ply 1) .............................................................................................................. 43
Figure 55 - 1/4 Model of the CPV.................................................................................................................................... 43
Figure 56 – Loading and Boundary Conditions ............................................................................................................... 44
Figure 57 - Weight Investigation - Safety Factors ........................................................................................................... 45
Figure 58 - Deformation Due to Gravity – S11 (Ply 1) ..................................................................................................... 45
Figure 59 - Comparison of S11 Stress on the Top and Bottom Outlet............................................................................ 46
Figure 60 - Adding Thermal Coefficients of Expansion to Material Properties .............................................................. 47
Figure 61 - Setting up a Pre-Defined Field ...................................................................................................................... 47
Figure 62 – FEA Effects of Environmental Temperature Change .................................................................................... 48
Figure 63 - Theoretical Effects of Environmental Temperature Change ........................................................................ 48

Tables
Table 1- Data Location Point and Ply Correlation ........................................................................................................... 28
Table 2 - Initial Run - Maximum Stresses ........................................................................................................................ 30
Table 3 - Initial Run - Safety Factors................................................................................................................................ 30
Table 4 - Iteration 1 - Maximum Stresses ....................................................................................................................... 32
Table 5 - Iteration 1 - Safety Factors ............................................................................................................................... 32
Table 6 - Iteration 1 - Main Body Safety Factors............................................................................................................. 32
Table 7 - Iteration 2 - Improved Main Body Safety Factors ............................................................................................ 33
Table 8 - Iteration 2 – Maximum Stresses ...................................................................................................................... 34
Table 9 - Iteration 2 - Safety Factors ............................................................................................................................... 34
Table 10 - Iteration 2 - Inlet Hole Check ......................................................................................................................... 34
Table 11 - Iteration 2 - Inlet Hole Safety Factors ............................................................................................................ 34
Table 12 - Iteration 3 - Maximum Stresses ..................................................................................................................... 35
Table 13 - Iteration 3 - Safety Factors ............................................................................................................................. 36
Table 14 - Analysis of Highest Stressed Elements ........................................................................................................... 36
Table 15 - Iteration 4 - Maximum Stresses ..................................................................................................................... 38
Table 16- Iteration 4 - Safety Factors .............................................................................................................................. 38
Table 17 - Inlet and Outlet Tubes - Maximum Stresses .................................................................................................. 41
Table 18 - Inlet and Outlet Tubes - Safety Factors .......................................................................................................... 41
Table 19 - Weight Investigation - Maximum Stresses..................................................................................................... 44

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 6 of 57

Introduction

This report describes the research, theoretical and computational methods, results and evaluation of work set out in
the FEA Composite Materials coursework brief.

This report is written by Steven Goddard (Student Number 10038749) as part of the Mechanical Engineering (Part
Time) course at UWE, Bristol during the 2014/2015 student year.

Fabricating components using composite materials gives the designer the ability to tailor the strength and stiffness
to match the applied loads. When high strength is matched with relatively low weight and corrosion resistance, this
makes composites an attractive material choice for numerous applications.

This piece of coursework details the process of designing a filament wound composite pressure vessel.

The report starts with details of research into the filament winding method and the advantages and limits of this
technique, from the research an initial ply angle is determined. Calculations using Pressure Vessel theory are
employed to determine applied axial and hoop stresses, based on given dimensions.

The design process starts with an Excel spreadsheet to calculate an appropriate ply layout using laminate theory. An
initial factor of safety is determined using the Maximum Stress, Tsai-Hill and Tsai-Wu failure criteria.
A model is developed using Abaqus v6.14, the mesh is refined and the design is re-iterated to determine the
optimum ply layout in order to achieve a factor of safety of 1.2.

Further work investigates the effect of; different distances to the concrete supports that hold the vessel, the weight
of the vessel and a consideration of environmental temperature change.

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 7 of 57

1. Background Reading and Research


1.1. Composite Pressure Vessels (CPV)

Figure 1 – Carbon Fibre CPV (McLaughlan et al, 2011)

Composite pressure vessels are widely used in many applications. With superior characteristics such as significant
weight reduction, high strength/stiffness and corrosion resistance, they are an excellent replacement for classic
metallic vessels. However, these benefits can be offset by higher cost of manufacture and certification as composite
pressure vessels require unique manufacturing, design and testing.

This technology was originally developed for military applications and then by NASA as part of its Firefighter’s
Breathing System Program in 1970 (McLaughlan et al, 2011). In 1975, the US Department of Transport certified CPVs
and commercial production started. Since then it has been adapted for numerous civilian purposes and the
commercial market. Some applications include SCUBA tanks for divers, high-pressure containers for fluids used in
propulsion & life support and paintball gas cylinders.

Another interesting point given by Önder (2007) is the emergence of a requirement for CPVs in the automotive
industry through the conversion of some cars to Compressed Natural Gas. A similar scenario is happening with the
implementation of hydrogen fuel cells and hydrogen gas storage such as the type IV tank used in the Toyota FCHV,
Mercedes-Benz F-Cell and the HydroGen4. (U.S DOE Hydrogen Program Review, 2002).

1.2. Filament Winding


A common method of manufacturing CPVs is Filament Winding. As the composite material is not considered
pressure tight, the vessel starts with a thin ductile liner (usually non-structural). This can be made from rubber,
plastic, or metal and acts as a fluid retention layer. The liner and end caps are girth welded; each weld is then
radiograph and penetrant inspected for acceptance. The complete liner is then leak tested and an adhesive is
applied immediately prior to the filament winding process. (Tam et al, 2002)

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 8 of 57

Figure 2 - Drum Type Resin Bath (Miaris, 2010)

In the filament winding process, continuous fibres (in the form of tow or rovings) are impregnated with resin using a
bath system such as the one shown in Figure 2. Alternative methods are the dip type and enclosed systems. Enclosed
systems are currently less widespread in their use but they are being developed to improve the resins viscosity and
pot life. The large surface area of the bath type impregnation methods expose the resin to air, degrading the resin
and causing the release of monomers such as styrene that also raises concerns over workplace hygiene. (Miaris,
2010).

The liner is set up on a mandrel and the composite material is wrapped around as the mandrel rotates. The fibre is
fed through a guide and a basic machine can move the guide from left to right. The speed of the guide synchronised
with the rotation of the mandrel controls winding angles and the placement of fibres.

The process is not limited to axis-symmetrical structures. More complex parts can be wound providing the machine
has the appropriate degrees of freedom. Modern advances in filament winding particularly in the 1970’s with the
use of servos and in the 80’s and 90’s with computer technology has enabled the use of high speed computers to
give smoother motion and greater fibre placement accuracy on multi-degree of freedom machines.

There are 3 main types of filament winding patterns are:

• Hoop Winding – This is a high angle winding approaching 90 degrees. Each rotation of the mandrel
progresses the fibre down by one bandwidth.
• Helical Winding – This method can produce winding angles down to 10 degrees.
• Polar Winding – Fibres pass tangentially from pole to pole as the mandrel is rotated along the longitudinal
axis. This method is used to wrap fibres on domed end pressure vessels.

After winding is finished, the mandrel is cured either naturally or in an autoclave. The mandrel is continuously
rotated during curing to maintain the uniform quality of the composite material and resin.

Advantages

• Easy to reproduce parts • Fibre and resin is used in its lowest cost form.
• Continuous fibre over the whole part • Process automation can result in cost savings.
• Structural properties of laminates are very good due • Fast
to the ability to lay fibres in complex patterns to
match applied loads.

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 9 of 57

Disadvantages

• Limited to convex shaped components • Wound external surface may not be satisfactory
• Mandrels can be complex and expensive. for some applications

Figure 3 - Fabrication of Filament Wound CPV with Hoop and Helical Layers (McLaughlan et al, 2011)

1.3. Optimum Winding Angle


Numerous sources of literature reviewed have concluded that a ply angle of 54.7° is the optimum winding angle for
filament wound CPVs. This is value is obtained by an analytical technique called Netting Analysis; it is one of the
most popular methods of investigating multi-layered composite materials.

The main assumption in netting analysis is that all loads are carried by the fibres, neglecting the stiffness of the
matrix and internal pressure subjected to the vessel produces a hoop-to-axial-stress ratio of 2:1. (Evans, 2002).

sin , cos (1)

tan tan √2 54.7° (2)

The optimum winding angle can be estimated using equations (1) and (2) where = ultimate tensile strength. The
body of the cylinder consists of axial and hoop stresses wrapped with fibres at the angle α as shown in Figure 4.

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 10 of 57

Figure 4 - Body Diagram of Axial and Hoop Forces and Internal Pressure (Sulaiman et al, 2013)

Shultz and Smith (2004), identify that there is a lack of studies that investigate the accuracy of netting analysis. They
perform experimental investigations using a newly developed biaxial test fixture and conclude that a 50° is the
optimum ply angle for carbon/epoxy pressure vessels. Their results also show that the Maximum Stress failure
criteria more accurately describe fibre dominated strength than the interaction criteria (Tsai-Hill and Tsai-Wu).

Onder (2007) conducts an extensive investigation into the failure of CPV’s and optimal angle-ply orientations. He
explains and develops theoretical methods based on the Lekhnitskii theory and studies finite element analysis and
experimental approaches to verify optimum winding angles. The investigations each use 4 layers at various
symmetric and anti-symmetric angles and makes particular use of the Tsai-Wu failure criteria.
Onder concludes that based on theoretical studies, the optimum winding angle for CPV’s with internal pressure
loading is obtained as 61° with slight differences between symmetric and anti-symmetric winding, however the
experimental results show no difference between symmetric and anti-symmetric and give an optimum winding angle
of 55°.

Sulaiman, Borazjani and Tang (2013) also study the winding angle of CPV’s under pure internal pressure using finite
element analysis. Plies were orientated from 0°-90° in intervals of 15° each model consisted of 6 plies with a
constant thickness of 0.762mm per layer. Data was analysed using an exact elastic solution along with Tsai-Wu, Tsai-
Hill and Maximum Stress failure criteria. They also concluded that a 55° winding angle was the optimum solution and
confirmed this by comparing to experimental data.

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 11 of 57

2. Theory of Thin Wall Pressure Vessels


The theory of thin walled pressure vessels can be applied to the CPV in order to determine the hoop and axial
stresses.

When a closed end cylinder is subjected to an internal pressure the vessel will expand, causing a tensile stress
tangential to the circumference, this is known as the hoop stress. Longitudinal fibres in the vessel will also stretch
creating another tensile stress; this is known as the axial stress.

Shear stress is not experienced due to the stresses being applied to the surface of the vessel. If the stress element
were taken on the inner surface, it would be treated as a three dimensional element.

The following simple equations enable the hoop and tangential stress to be calculated:
$%
! "#!! (3)
&
"
'()'* ! "#!! (4)
2
Where:
p= internal pressure
r = radius of cylinder
t = thickness of cylinder

In order to calculate the forces per unit length, equation (3) and (4) are multiplied by t.

The variables stated in the assignment brief are as follows:

Pressure = 60 Bar

Cylinder Radius = 400mm

This gives:
$%
Hoop force = " 2400000 N/m and Axial force = 1200000 N/m.

The hoop and axial stresses acting on the wall are membrane stresses; no shear stresses are present on these
elements because of the loading and symmetry on the vessel.

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 12 of 57

3. Calculating the Ply Layout


In order to determine the best ply layout I have developed a spreadsheet capable of analysing up to 20 plies. The
spreadsheet can account for material changes, temperature and moisture inputs along with differences in angle and
thickness.

For full details of the calculations involved in determining the safety factors, please refer to the spreadsheet and
course notes (Xia and Amali, 2014).

As the specific material was not identified in the assignment brief, I based my calculations on a layer thickness of
1.0mm.

Based on my research In Section 1.3 I decided to trial a 4 ply symmetrical layout of [55/-55]s, this produced the
following factors of safety:

Maximum Stress = 1.28 Tsai-Hill = 1.10 Tsai-Wu = 1.47

With the Tsai-Hill safety factor falling below 1.2 this was not acceptable. I used the Solver tool in excel to determine
the change of angle required in order to increase the FOS, this was not possible. Therefore, I have raised the ply
thickness to 1.1mm which gave sufficient safety factors of:

Maximum Stress = 1.41 Tsai-Hill = 1.21 Tsai-Wu = 1.62

Using the solver again, the optimum ply angles for my laminate would be [55.392°,-55.111°]s, this makes the Tsai-Hill
safety factor equal to 1.2 to a precision of 6 decimal places.

Being mindful of angle accuracy limitations in some filament winding processes I have decided to remain with a +/-
55° symmetrical ply layup.

Figure 5 - Ply Stack Layout [+/-55]s with 1.1mm Layer Thickness

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 13 of 57

4. Abaqus®

4.1. Preparing the Abaqus® Model


The following section describes the method used to produce the initial isotropic vessel, for the composite pressure
vessel the same method applies with some additions detailed in Section 4.3.

To begin, a new model database was created; this model database is used throughout the whole project. A new
model instance is then created. Further models will be copies of this initial model with incremental changes.

Figure 6 - Sketcher

Once the model was created, within the part module a shell feature was created the shape was based on a sketch on
the XY plane (Figure 6).
The geometry for the CPV was roughed out using the line and circle tools. Dimensions from a fixed (0,0) point were
applied along with geometric constraints to constrain the model to the specified measurements. Any other lines
were trimmed or deleted to suit.

Once the sketch was complete the shell was revolved 90 degrees about the y axis to form an 1/8 of the CPV. The
reason for this is that in subsequent steps symmetry can be applied to the edges of the model, reducing the model
like this greatly improves computing efficiency by reducing the amount of elements and nodes.

Next a material was defined (Figure 7), for this first model “Steel” was created. Density, Young’s modulus and
Poisons ratio were used to define the material properties. In Abaqus, there are no units, therefore all values that are
input need to be consistent. Because millimetres were used in defining the geometry or the CPV, all other values
were scaled to this form. (Kg/mm3, N/mm2, Poisons Ratio is unit-less).

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 14 of 57

Figure 7 - Create/Edit Material

In a standard model the material would now be applied to a section of the model, but because the model will
eventually consist of a series of composite plies, a composite layup is created (Figure 8).
For this initial model the composite layup will consist of one ply, made of the newly defined material “Steel” at a
thickness of 4mm to represent the values on the spreadsheet. The rotation angle doesn’t matter because steel is an
isotropic material. In Section 4.3 the full ply layup will be defined along with rotation angles.

Figure 8 - Create Composite Layup

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 15 of 57

The next step is to create an instance of the part as an assembly. The part can either be dependent or independent
based on how it is going to be used. I have chosen to make this a dependant part and to edit subsequent instances in
the part module.

After this, Steps need to be set up. These indicate a major change in the model. Every model starts with an initial
step to represent initial conditions. For this project one additional step is needed which will define the applied 60 bar
pressure, this will be a “Static, General” type step (Figure 9).

Figure 9 - Step Creation

Once the step is produced a “Field Output Request” needs to be defined in order to output the relevant results.
As a preparation step for the CPV, “Composite Layup” is selected as the domain for the Field Output. In post
processing this will allow individual plies to be selected for contour plotting of variables (Figure 10).

Figure 10 - Field Output Request

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 16 of 57

Figure 11 - Boundary Conditions

The next stage is to create boundary conditions only the symmetrical boundaries are set by selecting the appropriate
edge and selecting a symmetry based on the axis system. Figure 11 illustrates symmetry being applied in the x-
direction.

Figure 12 - Create Loads

Once the boundary conditions have been set, one of the most important steps in the process is to apply the loading
on the component. As pressure inside the cylinder exerts itself in all directions uniformly this is relatively easy to set
up. The create load tool is selected and a pressure type load is chosen to be applied at the “Pressurise” step with a
uniform distribution. The magnitude of the load is again unit-less and must be consistent with the other values used.

Finally, the model is ready for meshing, for this initial run the distance between mesh seeds was set at approximately
100mm. This is a sufficient model in order to use as a baseline for a mesh study.

A job is now created in order to run the simulation. The job is then submitted.

Once complete, results can be viewed and plotted onto the model. Individual nodes can be probed and data can be
exported for analysis in Excel or Matlab etc.

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 17 of 57

4.2. Isotropic Model


As explained in Section 4.1, the first model created was an enclosed pressure vessel made out of steel. Due to steel
being an isotropic material the results should compare accurately to the calculated hoop and longitudinal stresses.
stress
This will provide a good indicator
cator of into the accuracy of the spreadsheet.

The stresses are calculated as per equations (3) and (4):

Axial/Longitudinal Stress =

Hoop/Circum
Hoop/Circumferential Stress =

An initial mesh with nodes spaced approximately 100mm apart gave the following results:

Figure 13- Isotropic FEA Results

The results obtained from this initial model were accurate when compared to the theoretical values. It is clear that
the hoop stress is 608.133 MPa (0.85% % error) and axial stress is 299.173 MPa (0.78%
% error).

Overall, the FEA results for hoop stress also conform to theory in that the stress at the end of the cap (280.535 MPa)
is approximately equal to the axial stress (299.173
( MPa); this correlates with theory for a spherical pressure vessel.
vessel

As a further investigation into the isotropic pressure vessel results a brief mesh study was conducted to determine
the effect of a finer mesh. All jobs ran with a quad, structured mesh varying from 8mm
8 to 150mm spacing between
seeds. Figure 14 shows the results were more consistent when using finer meshes but accuracy did not change
significantly between mesh sizes.

Figure 14 - Isotropic Pressure Vessel Mesh Study


Steven Goddard –Mechanical
Mechanical Engineering (Part Time)
Student Number: 10038749 Finite Element Analysis Report Page 18 of 57

4.3. Composite Model Preparation


In order to represent a composite pressure vessel the model required some modification from the Isotropic state.

Firstly a ply layup was created which defined each layers in the composite laminate (Figure 15). Thickness and
rotation angle were also specified here based on the initial layout calculated using the spreadsheet.

Figure 15 - Add a Ply Layup

As with Steel, the plies each required material properties. A new material was created based on the coursework
specification (Figure 16). The difference between this and Steel is that it is no longer an isotropic material. The elastic
type was set to “Lamina” which allowed the specification of transverse and longitudinal strengths. These
represented the difference directional characteristics of the composite material.

The assignment gives a material property for G12 but does not specify additional Shear Modulus components G13 and
G23. Kurtz (1990) suggests that an experimental approach is the best way to determine these values. Although not a
solid reference, forums such as Eng-Tips.com (Eng-Tips.com, 2011) and Abaqus User Forum (Abaqus User Forum,
2008) confirms that the experimental approach or a micromechanical analysis is currently the only way to determine
these properties.

All values for Shear Modulus are assumed equal to 6000 N/mm2.

Next the geometry has to be modified to match the CPV shown in the specification. This vessel features 4 holes
(Inlets and Outlets) along with two concrete supports. The inlet positioned on the end caps of the vessel was formed
by creating a datum plan on the end of the vessel. Once the plane was set up a sketch could be created, the sketch
was then used with the cut>extrude tool to form the hole.

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 19 of 57

Figure 16 - Create Lamina Type Material

Figure 17 - Forming the Outlet Hole

A similar process was implemented to create the outlet hole (Figure 17), this was located on the side wall of the
vessel at the half way point. On the 1/8th model, this meant the hole could be simply positioned as a coincidence
with the origin.

Figure 17 also shows a square and additional circle surround the outlet hole. These were added due to errors when
meshing. In Section 4.4.2 these partitions proved useful in setting up a local mesh around the area.

Lastly an additional partition was created to represent the concrete support. Even though the thickness is not
numerically specified in the specification there is a clear thickness represented on the diagram. A 50mm thickness
has been assumed which is an approximate scaling from the diagram. A commercially available block of this width is
Aircrete which is an ethical and sustainable structural material used in the building industry (Tarmac Building
Products, 2014).
Without a thickness a constrained boundary edge used to represent a support is unrealistic and could give a high
stress concentration which could adversely affect the result of the investigation.

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 20 of 57

4.4. Mesh Study


A mesh study is a technique used to measure convergence of results by incrementally varying a mesh. Usually a finer
mesh will result in a more accurate result, but more nodes will increase computation time. A mesh study will identify
a point at which the results are good enough so that the user can maximise the model efficiency.

The definition of a “good mesh” was hard to pinpoint, rather it seems a mesh needs to be tailored to the specific
model. Baker (2002) suggests some good points on what typically makes a good mesh. The following points are
important aspects when creating a mesh:

• The density of the mesh should be high enough to capture all relevant features.
• A uniform mesh will not allow detailed computation at the important features where it is most needed.
• More cells/nodes will give higher accuracy but will increase CPU time.
• Avoid high aspect ratios.
• Errors occur when the mesh:
o Is too coarse
o Has high skewness.
o Has large jumps between adjacent cells.
o Inappropriate local meshing.

4.4.1. Global Mesh Study


To start with, the model was analysed as a whole. The idea of this was to refine the global mesh first and then move
on to specific area. A structured quad type mesh was used across the whole model.

Figure 18 illustrates the checking areas that were probed for values during each iteration in order to compare
convergence. These areas were chosen as they should stay quite similar throughout the mesh refinement. They are
also not located near any stress concentrations or dominant features of the model such as the holes and concrete
supports. By increasing the mesh density in a step-by-step manner, a suitable number of elements are obtained to
find a balance between efficiency and CPU time. The stabilization of three stress components S11, S22, and S12 is
considered as the convergence criteria.

Convergence has been analysed by comparing the change from initial value. A successful convergence is indicated
when the values do not differ more than is required.

Figure 18 - Global Mesh Study – Checking Areas (Red)

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 21 of 57

100mm Mesh 75mm Mesh

50mm Mesh 40mm Mesh

30mm Mesh
20mm Mesh

10mm Mesh 8mm Mesh

Figure 19 - Global Mesh Study Iterations

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 22 of 57

Figure 20 - Mesh Study - Global Convergence

Figure 20 shows that the majority of data points converge sufficiently at approximately
approximately 30mm -35mm mesh spacing.

Point 3 – S12 shows no clear convergence, this may be caused by the point being too close clos to the concrete support
and the start of the end cap. Point 3 - S11 also varied significantly, in order to obtain higher resolution on the rest of
the data Point 3 - S11 is omitted from Figure 20.

Point 2 – S11 deviates approximately 115MPa from the initial reading at 100mm spacing but shows clear signs of
convergence after 35mm spacing.

The software reached maximum node limit at 6mm global mesh spacing. In order to ensure an accurate simulation,
simulation a
20mm global node spacing for the mesh was selected,, this represents improved convergence whilst allowing further
nodes to be available for mesh refinement around the holes and support along with a sensible computation time.
time

Tri Mesh Investigation

As an additional investigation, a tri mesh between 25mm


25 and 10mm
mm global mesh distance was analysed.

Point 3 still shows inaccuracies when using a triangle mesh,


mesh although Point 3 – S12 has improved over the quad
mesh. It is likely that this area will require a local mesh study to account forr the close boundary conditions of the
concrete support.

The tri mesh did not appear to improve accuracy significantly, therefore a 20mm quad structured mesh was
considered adequate with further local meshing near the holes, concrete support and end cap.

Improvements

The global mesh study could have been improved by probing values at specific coordinates; the method of selecting
elements in an area gave rise to potential error and even bias selection.
select As the mesh spacing decreased the exact
area was not always available to probe.

Steven Goddard –Mechanical


Mechanical Engineering (Part Time)
Student Number: 10038749 Finite Element Analysis Report Page 23 of 57

25mm Mesh 20mm Mesh

15mm Mesh 10mm Mesh


Figure 21 - Global Mesh Study - Tri Mesh Iterations

Figure 22 - Tri Mesh Study - Global Convergence

Steven Goddard –Mechanical


Mechanical Engineering (Part Time)
Student Number: 10038749 Finite Element Analysis Report Page 24 of 57

4.4.2. Local Mesh Study


Local mesh refinement was required at the hole locations, around the concrete support and end cap. Local mesh
refinement ensures that there are increased nodes at the more critical areas of the model.

Starting with the inlet hole, a partition was created in the geometry to produce biased seeding local to the hole area
whilst maintaining global mesh size on the remainder of the end cap.

Figure 23 - Bias Seed Distribution Figure 24 - Local Mesh Study - Inlet Checking Points

Seed biasing is illustrated in Figure 23 with the seeds towards the bottom of the image at 20mm spacing. As the
seeds approach the inlet hole they converge to 5mm spacing giving a gradually finer mesh.

Figure 24 shows the check points used to check for convergence.

Ratios from 20-5 to 20-0.5 were experimented with in an attempt to improve the convergence. The convergence of
values was investigated in the same way as the global mesh study in order to select an appropriately accurate mesh
setting for this local area.

20-5 20-2 20-1 20-0.5

Figure 25 - Local Mesh Study - Inlet Iterations

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 25 of 57

After the 20-1 seed ratio the mesh started to increase in aspect
aspect ratio in the opposite way. Two points were
measured around the hole show in Figure 24 and the values appear to converge after 20-1 20 (Figure 26). Increasing the
mesh ratio further may lead to higher aspect ratios
rat and larger inaccuracies, therefore a 20-1
20 seed ratio for the inlet
has been chosen.

Figure 26 - Inlet Hole - Local Convergence

A similar process was performed on the outlet hole and concrete block boundary condition until a suitable mesh was
produced.

For the Outlet Hole I used the previously created partitions (the square around the hole and larger concentric
circular section) to
o bias the seed distribution
distribu accurately around the area. The
he cube area (shown as pink in Figure 27)
was set as an unstructured quad mesh. Although an unstructured mesh is not as efficient as a structured one due to
the difficulty in expressing the 2-dimensional
dimensional array in the computer memory, this meshing technique
tec allowed finer
meshing around the hole whilst maintaining a good aspect ratio and no skewed elements here and throughout the
main body of the model.

As a result of a similar convergence


ence study to the Inlet, the following seed distribution was set around
arou the outlet hole,
see Figure 27:

• A – 20mm Spaced Mesh


• B – Single Bias of 20-10
• C – 10mm Spaced Mesh
• D – Single Bias of 10-1
• E – 2.5mm Spaced Mesh

This method was chosen as it ensures a gradual reduction in element sizes rather than jumping from large to small
elements which would likely result in inaccuracies (Baker, 2002).

Steven Goddard –Mechanical


Mechanical Engineering (Part Time)
Student Number: 10038749 Finite Element Analysis Report Page 26 of 57

Figure 27 - Outlet Hole Mesh Refinement

For the concrete support and the edge dividing the end cap from the main body the mesh type was set to structured
quad and the seed distribution was biased 20mm to 10mm. The finer mesh occurs on the edge of the main body/cap
and on the two edges of the concrete support.

Figure 28 - Concrete Support Mesh Refinement

With an appropriate mesh defined, this model can be used confidently for analysis and a series of design iterations
to improve the safety factor of the vessel.

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 27 of 57

4.5. Design Analysis & Iterations

4.5.1. Method of Data Analysis


In order to investigate the model in detail, data was exported to Excel. Excel is an appropriate choice of software to
analysis small to medium sized data sets. With the elements/nodes used and the limited variables that were being
examined Excel represented the best choice for manipulating and investigating the data. If the model contained
significantly more elements or vast amounts of variables were being analysed software that can better handle these
sized datasets such as MatLab or Maple would be a better choice.

In order to export the data the “Report > Field Output” option was selected within Abaqus. From this window (Figure
29) the Stress Components S11, S22 and S12 were selected. On the setup tab an output directory for the data was
chosen.

Figure 29 - Results > Field Output

The data was output in a .rpt file, when opening in Excel the data was in a format that could be interpreted by the
Excel Import Wizard (Figure 30). By delimiting the data by spaces, this separated each set of values into columns.
This method did result in some conversion errors, but this was mainly within the text and descriptive information at
the beginning of the file and also the offset of some of the data, this was easily rectified manually.
As an improvement to this method and if this was done more commonly, it would have been suitable to write a
macro script to arrange the data more efficiently.

Figure 30 - Excel Import Wizard

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 28 of 57

Once the data was processed, the stresses were displayed for each of the location points, for the initial run with 4
plies these locations points were taken through the laminate in the following manner:

Location Point 1 Ply 1


Location Point 2 Ply 1
Location Point 3 Ply 2
Location Point 4 Ply 3
Location Point 5 Ply 4
Location Point 6 Ply 4
Table 1- Data Location Point and Ply Correlation

In larger ply layups more location points were created and the correlation between them and plies became more
complex.
Maximum values for S11, S22 and S12 were taken from the full set of data (which includes the data for every ply)
and were used in the calculations for the safety factors. Even though these peak values may come from different
areas in the of the model, this method was chosen as It gives a good indication of a worst case combination of
stresses.
Any value can be traced back to the exact element it was calculated at, this is particularly useful when identifying key
areas for improvement.

4.5.2. Initial Run


Using the ODB display options in Abaqus, the deformed S11 stress plot can be mirrored to give a representation of
the full vessel. Figure 31 shows deformations around both sides of the concrete supports with larger deformations
on the outside side and at the top and bottom (X-Axis) of the vessel. The vessel around the outlet hole also deforms
with larger magnitudes in the circumferential direction, this correlates well with the theory described in Section 2.
There is also no significant deformation visible at the inlet.

The data shows that the maximum stresses occur at the outlet in Ply 1 (Figure 32); this is due to the hole creating a
stress concentration and being located on the area where hoop stresses are highest. From further investigation the
peak stresses at the outlet, 1728 MPa (Figure 33) are also approximately half of the peak stress at the inlet (3435
MPa) which is expected based on theory.

Figure 31 - Initial Run - Deformed S11 Stress Plot

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 29 of 57

Figure 32 – Initial Run - S11 Outlet View (Ply 1)

Figure 33 – Initial Run - S11 Inlet View (Ply 1)

Figure 34 – Initial Run - S22 Outlet View (Ply 1)

Figure 34 shows peak S22 stresses occurring at the Outlet on Ply 1 in the axial direction with a small compressive
stress in the circumferential direction. Figure 35 shows the whole cylinder with clear peak stresses on edges of the
concrete support and at the outlet hole. It is conclusive that with the current ply layup the outlet hole area is the
source of the highest stress and a weak point in the design.
Steven Goddard –Mechanical Engineering (Part Time)
Student Number: 10038749 Finite Element Analysis Report Page 30 of 57

Figure 35 – Initial Run - S12 Whole View (Ply 1)

Overall the peak stresses were:

S11 3440 MPa


S22 343 MPa
S12 583.429 MPa
Table 2 - Initial Run - Maximum Stresses

This resulted in Safety Factors of:

Maximum Stress 0.359


Tsai-Hill 0.198
Tsai-Wu 0.196
Table 3 - Initial Run - Safety Factors

Safety factors were lower than calculated. This was expected due to the stress concentrations at the holes and
supports, which were not incorporated into the initial calculations. The composite theory which drives the initial
spreadsheet for this project is not able to predict stresses on such a complex composite part, this is main reason for
using an FEA computer based approach when adding features such as supports and holes.

In order to improve the factor of safety the exact elements in each layer where the high stresses occur were
identified (Figure 36), this gave an indication on the most appropriate places to strengthen the laminate.

Figure 36 - High Stress Areas (Typical All Plies)

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 31 of 57

4.5.3. Iteration 1
Based on the locations identified in Figure 36, the laminate was strengthened by adding 4 additional plies (Figure 38)
in the same orientation setup as before [+-/55]s. Using the region selection in the composite ply layup settings, four
sections representing the high stressed areas were chosen.

Figure 37 - Iteration 1 - Strengthened Regions

Figure 38 - Iteration 1 Ply Stack Plot

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 32 of 57

Figure 39 - Iteration 1 - S11 Whole View (Ply 1)

After running Iteration 1, the data was exported as per Section 4.5.1.

The data was more complex this time round as the results were output as multiple regions due to the ply layup
modifications in some areas. After some processing, the peak stresses were identified:

S11 1522.61 MPa


S22 140.7 MPa
S12 222.1 MPa
Table 4 - Iteration 1 - Maximum Stresses

Resulting in safety factors of:

Maximum Stress 0.788


Tsai-Hill 0.491
Tsai-Wu 0.480
Table 5 - Iteration 1 - Safety Factors

The effect of adding the new four layers has more than doubled the safety factor when compared to the initial run
but overall it is still short of the required 1.2 safety factor required in the specification.

As an additional investigation, several nodes on the main body of the vessel, shown as the dominant green area in
Figure 40 were probed. This location is known to be a main area of low stress, therefore calculating the factor of
safety using the stresses here will indicate if the base composite layup is strong enough. The safety factors are as
follows:

Maximum Stress 1.788


Tsai-Hill 1.02
Tsai-Wu 0.969
Table 6 - Iteration 1 - Main Body Safety Factors

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 33 of 57

Although the Maximum Stress criteria gives an acceptable safety factor, the Tsai-Hill and Tsai-Wu are more
conservative. As the minimum safety factor is being used this indicates that the base ply thickness needs to be
increased.

Investigating the highest stressed elements identified by the data indicates the outlet hole is still a problem.
Maximum values for S11, S22 and S12 are all located around the hole edge. For iteration 2, the thickness of the base
layup including the additional 4 plies will be thickened and the results of this will be evaluated. If this is not
sufficient, investigation into adding further plies in close proximity to the outlet hole will be considered.

Figure 40 - Iteration 1 – Main Body Investigation (Ply 1)

4.5.4. Iteration 2
Based on the results of Iteration 1, each ply thickness has been increased to 2.2mm.

Figure 41 - Iteration 2 - S11 Whole View (Ply 1)

The stresses across the main body have decreased around 30% resulting in safety factors of:

Maximum Stress 2.53


Tsai-Hill 1.43
Tsai-Wu 1.40
Table 7 - Iteration 2 - Improved Main Body Safety Factors

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 34 of 57

These safety factors meet the required specification of 1.2 which means the ply lay up for the main body is sufficient.
However, the higher stresses at the holes require further attention. Maximum stresses across the whole vessel are
given as:

S11 1083.33 MPa


S22 99.78 MPa
S12 136.57 MPa
Table 8 - Iteration 2 – Maximum Stresses

Resulting in safety factors of:

Maximum Stress 1.11


Tsai-Hill 0.74
Tsai-Wu 0.72
Table 9 - Iteration 2 - Safety Factors

The inlet hole was also checked due to high stresses noticed visually in Abaqus. By probing the area in Abaqus, the
maximum stresses around the inlet were calculated as:

S11 800.86 MPa


S22 14.86 MPa
S12 -14.42 MPa
Table 10 - Iteration 2 - Inlet Hole Check

Resulting in safety factors of:

Maximum Stress 1.50


Tsai-Hill 1.49
Tsai-Wu 1.48
Table 11 - Iteration 2 - Inlet Hole Safety Factors

These values are within specification therefore no further modification was needed on the ply layup around the inlet
at this stage.
It is also interesting to note that the three safety factor methods agree to a tolerance of ±.02, which is unlike any of
the previous results where Maximum Stress was far higher.

For the next iteration, the small circular section around the outlet hole will be strengthened by adding four more
additional plies.

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 35 of 57

4.5.5. Iteration 3
Four plies have been added to the circular area around the outlet hole at 2.2mm thickness, I have angled these plies
with a [+/-55]s layout (Figure 42). Local strengthening of this area should increase the safety factor whilst maximising
weight saving and potentially reducing material cost around the main body.

Figure 42 - Iteration 3 - Increased Ply Layup

Figure 43 - Iteration 3- S11 Whole Model (Ply 1)

A visual inspection through Abaqus suggests the output was improved (Figure 43), apart from the concrete support
the rest of the model appeared less heterogeneous than previous iterations. Exporting the data uncovered the
highest stresses to be:

S11 847.395 MPa


S22 88.979 MPa
S12 107.779 MPa
Table 12 - Iteration 3 - Maximum Stresses

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 36 of 57

Resulting in safety factors of:

Maximum Stress 1.42


Tsai-Hill 0.93
Tsai-Wu 0.89
Table 13 - Iteration 3 - Safety Factors

These safety factors still do not meet the specification, but on further investigation of the data the maximum stress
values come from varying locations across the model rather than being concentrated at the outlet hole as before.
This further confirms the initial observations made from the visual data. The values in Table 13 represent the safety
factors based on the worst-case combination of stresses.

Due to this each of the elements on the model that had the highest stresses were probed for the values of S11, S22
and S12 in each location. The stresses and resulting safety factors are as follows:

Element S11 S22 S21 Maximum Tsai-Hill Tsai-Wu


(MPa) (MPa) (MPa) Stress
Iteration 3 - S11 Max 847.395 20.401 -41.691 1.416 1.319 1.300
Iteration 3 - S22 Max 542.209 88.976 -72.011 2.213 1.292 1.202
Iteration 3 - S12 Max 99.280 64.019 107.779 3.124 1.268 1.424
Table 14 - Analysis of Highest Stressed Elements

The areas where maximum stresses occur are all within specification. At this point the laminate may be concluded as
a suitable layout that meet specification. However, there still may be areas where a combined set of stresses could
cause the factor of safety to drop below specified limits. Therefore, a further iteration will be carried out to ensure
the worst-case scenario gives a safety factor of at least 1.2.

The element of maximum S22 has also uncovered a possible side effect of sudden discontinuities between ply
thicknesses. This point, identified in Figure 44 is located on the edge between a 8.8mm and 17.6mm layup. This
effect is also causing higher stress on the axial edge that the circumferential one.
A recommendation for further work would be to compare the stepped change in thickness (modelled here) with a
more gradual increase in ply thickness.

Figure 44- Iteration 3 – Element with Maximum S22 – Edge of Increased Ply Thickness

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 37 of 57

4.5.6. Iteration 4
To make a more educated decision for the ply modification, the safety factors and ply thicknesses from the four
previous runs were plotted on a simple graph. A line of best fit was applied to Figure 45 and extrapolated this until it
reached the desired safety factor. According to the graph, a maximum ply thickness of 35.6mm should give the
correct safety factor.

Figure 45 - FOS Trend

Based on this information the ply thickness was increased by 3mm across all twelve plies,
plies giving a total thickness of
36mm at the outlet through to 12mm at the main body.

Based on the observations related to the element with maximum S22 S2 in Iteration 3 (Figure
Figure 44), this area was also
strengthened by adding an additional four plies to this section (Figure 46).

Figure 46 - Iteration 4 - Strengthening Area of Maximum S22

After the layup modifications were complete, the simulation was run.

Steven Goddard –Mechanical


Mechanical Engineering (Part Time)
Student Number: 10038749 Finite Element Analysis Report Page 38 of 57

Figure 47 - Iteration 4 - S11 Whole View (Ply 1)

Once processed, the data shows the three highest stresses as:

S11 609.39 MPa


S22 61.959 MPa
S12 81.772 MPa
Table 15 - Iteration 4 - Maximum Stresses

Resulting in safety factors of:

Maximum Stress 1.97


Tsai-Hill 1.27
Tsai-Wu 1.22
Table 16- Iteration 4 - Safety Factors

These safety factors show that at a worst-case scenario of combined stresses, the vessel will have a factor of safety
of at least 1.22, which is within specification. This also matches the predicted thickness from Figure 45.

This model will be used as a benchmark to compare to in the following investigations.

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 39 of 57

4.6. Investigation into Different Positions for the Concrete Support


In all previous simulations of the composite pressure vessel, the concrete supports have been set at 1.6m apart.
Using the ply layout from design iteration 4, Abaqus calculates a minimum factor of safety of 1.22, which is within
specification.

The next two investigations examine the effects of moving the concrete supports to distances of 1.5m and 1.4m
apart.

A review of relevant literature which examines the support of composite and traditional pressure vessels suggests
that two saddled, evenly spaced supports is the ideal setup for supporting a pressure vessel (Zick, 1951). This is
based on the load being equally shared over the two supports as opposed to a multi-support setup that would result
in a static indeterminacy and difficultly in predicting the load distribution in the event of a foundation settlement.
Zick’s work which is based on beam theory is also included in the British Standard BS 5500 (British Standards
Institution, 1997) and ASME Boiler and Pressure Vessel codes (ASME, 1989).
Zick also mentions that the two supports should be constrained in a way that one support is fixed to the vessel in the
axial direction whilst the other only supports circumferentially and allows axial thermal expansion to happen without
incurring additional axial stresses. The model analysed in this section represents an 1/8th of the pressure vessel and
is symmetrical in all directions meaning that both supports are rigidly fixed to the vessel. A change in temperature is
not yet considered but a recommendation for further work would be to model half of the vessel and investigate the
effect of temperature combined with an axially fixed and non-axially fixed support.
Finally, Zick comments on the use of rings around the pressure vessel, which is similar to the conditions in this
investigation. He concludes that the rings will act as stiffeners around the vessel. Therefore, it would be reasonable
to expect that moving the supports closer to the inlet hole will decrease stress in this area, improving the overall
safety factor.
A more recent paper (Khan, 2010) investigates the ideal construction of supports for pressure vessels. Khan states
that the most critical condition of loading is the self-weight of the vessel along with the weight of fluid inside.
Another further investigation would be to model the supports including the additional effect of gravity with the self-
weight of the vessel along with any fluid inside.

Figure 48 - Modifying the Geometry of the Concrete Supports

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 40 of 57

In order to complete the investigation, the geometry was changed incrementally to reflect the 1500mm and
1400mm support distances (Figure 48).

The investigations show that the safety factors increase by approximately 0.002 each time.

As an additional investigation supports at 800mm separation were simulated, this also had the effect of increasing
the safety factor further proving the trend. This agrees with the literature (Zick, 1951), in that the concrete support
ring acts as a stiffener to the stress concentration at the outlet hole.

1600mm 1500mm 1400mm 800mm


Maximum Stress 1.9692 2.0067 2.0134 2.0942
Tsai-Hill 1.2651 1.2553 1.2569 1.2572
Tsai-Wu 1.2214 1.2243 1.2260 1.2389
Minimum 1.2214 1.2243 1.2260 1.2389
Figure 49 – Distance of Concrete Supports - Safety Factors

The comparison of stress contours shown in Figure 50 (See Appendix 1 for enlarged versions) show that in S11,
moving the supports inward reduces the compressive stresses occurring on the outward side of the vessel. In the S22
direction, the magnitude of the high stress area shown in red is maintained no matter where the supports are
positioned but the effective area is decreased. The shear stress, S12 is highest around the support edges and
reduced by around 70Mpa from 1500mm to 800mm.

S11 S22 S12


1500mm
1400mm
800mm

Figure 50 - Comparison of Stress Contours when Moving Concrete Supports

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 41 of 57

4.7. Investigation into the Effects of Connecting Inlet & Outlet Tubes
In order to investigate the effects of tubes attached to the inlet and outlet, encastre boundary conditions have been
applied around the inlet and outlet holes in Abaqus. This assumes that the tubes are not flexible and are rigidly
fastened to the inlet and outlet similar to Figure 51 such that the pipes do not apply any additional load.

Figure 51 - Example of Solid Pipe Fitting (Tryer Process Equipment)

The results from the simulation give maximum stresses as:

S11 471.6 MPa


S22 68.4 MPa
S12 -70.8 MPa
Table 17 - Inlet and Outlet Tubes - Maximum Stresses

Resulting in safety factors of:

Original With Tubes


Maximum Stress 1.97 2.54
Tsai-Hill 1.27 1.46
Tsai-Wu 1.22 1.39
Table 18 - Inlet and Outlet Tubes - Safety Factors

The additions of the tubes clearly act as a supporting structure directly on the inlet and outlet holes reducing the
stress concentration from this area and increasing overall safety factors.

Maximum stresses in the S11 direction have reduced by almost 140 MPa, the stress concentrations now appear
around the concrete support and on the edges adjacent to the first strengthening stage (Figure 52).

In the S22 direction, maximum values have decreased but the outlet hole is still experiencing the highest stresses
along with portions of the main body that could be due to the reduced ply thickness in this area (Figure 53).
In the S12 direction, the maximum tensile stresses have more than halved, resulting in dominant negative stresses
acting on the main body (Figure 54).

The method used to simulate the inlet and outlet holes was quite basic. A more realistic method would be to model
the geometry of a connecting flange or interface. Another method of connecting pipes to the CPV would be to use a

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 42 of 57

threaded insert bonded into the composite layup. This would usually be of solid construction further strengthening
the hole locations and reducing stress in the area.

Figure 52- Inlet and Outlet Tubes - S11 (Ply 1)

Figure 53 - Inlet and Outlet Tubes - S22 (Ply 1)

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 43 of 57

Figure 54 - Inlet and Outlet Tubes - S12 (Ply 1)

4.8. Investigation into the Effects of Weight of the Vessel


In order to represent gravity adequately, the Abaqus model has been recreated as a 180 degree revolved shell (1/4
model). This avoids issues with applying gravity whilst the symmetrical boundary conditions are active. In order to do
this, a copy of the initial part was mirrored about the YZ axis.

This new part was then added to the assembly module. Once these two instances were in the assembly module they
were merged together into one connected part.

The new merged part was partitioned through the middle down the y-axis and re-meshed identically to the initial
1/8th model to maintain the accuracy determined in the mesh study (Figure 55).

Figure 55 - 1/4 Model of the CPV

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 44 of 57

Due to the change in the model the pressure loads and boundaries were reset. As the model was now a ¼ of the
cylinder the X-SYMM boundary condition was no longer needed.

In order to apply gravity to the part, density properties for the material were applied.

In order to ensure gravity is already in place before the internal pressure is applied, a new step called “Gravity” was
been created before “Pressurise”. A new “Gravity” type load was also created and applied to the Gravity step, these
are the yellow arrows pointing in the x-direction in Figure 56. The distribution was uniform and a value of 9810
(consistent with units used) was input in the component 1 direction (X-Axis).

Field and History Output options were also updated to account for the additional step.

The job created and run.

The assumptions made during this investigation were that the fluid inside the vessel had a negligible effect on the
weight. Only the self weight of the vessel was taken into account.

Figure 56 – Loading and Boundary Conditions

The data gives the three highest stresses as:

S11 602.7 MPa


S22 63.2 MPa
S12 -86.5 MPa
Table 19 - Weight Investigation - Maximum Stresses

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 45 of 57

Resulting in safety factors of:

Original With Gravity


Maximum Stress 1.97 1.99
Tsai-Hill 1.27 1.23
Tsai-Wu 1.22 1.19
Figure 57 - Weight Investigation - Safety Factors

Figure 58 shows the deformed vessel (exaggerated). It deforms around each side of the concrete support as
expected. Towards the end cap the maximum stress on the top of the vessel is actually a compressive force in the x-
direction with a lower tensile force on the bottom of the vessel. The vessel is also noticeably deformed in this
location more in the x-direction than the y-direction which confirms that gravity is having a bias effect on the bottom
of the vessel.

Figure 58 - Deformation Due to Gravity – S11 (Ply 1)

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 46 of 57

Top of the Vessel Bottom of the Vessel


Figure 59 - Comparison of S11 Stress on the Top and Bottom Outlet

Figure 59 shows the S11 stresses on both the top and bottom outlet holes. There is a clear difference with the
maximum stress occurring at the bottom of the vessel and minimal stress occurring on the top hole.

Despite the observational impacts of gravity mentioned above, the effect of the weight of the vessel has not had a
significant impact on the factor of safety. The Tsai-Hill and Tsai-Wu criteria both decrease whilst the maximum stress
criteria rises. This could indicate the weaknesses in the Maximum stress failure as this method assumes no
interaction between modes of failure. The Tsai-Hill and Tsai-Wu criteria however take into account the interaction
between failure modes which is why they are more conservative.

By querying the properties of the vessel, the mass was obtained (for the ¼ model) and the force acting on the
support was calculated to be approximately 300 N. Compared with the force exerted by the pressure, the weight of
the vessel represents a minor portion of the total load on the vessel. Therefore, it makes sense that the safety
factors are not changing by a large amount.

If however the vessel was made of thicker and denser material or if a dense fluid was contained in the vessel, gravity
would have a much larger effect. Increased sagging of the lower portion of the vessel resulting in bending moments
about the supports would also occur based on the distance between the concrete supports as the concrete supports
are rigidly fixed to the vessel this would most likely increase stress concentrations at the joining edge.

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 47 of 57

4.9. Investigation into the Effects of Environmental Temperature Change


To prepare the model for this investigation the material properties were updated to include the Thermal Coefficients
of Expansion given in the specification (Figure 60) keeping to the consistent form of units.

Figure 60 - Adding Thermal Coefficients of Expansion to Material Properties

A predefined field was also been set up to define a 20 degree change in the environmental temperature.

Figure 61 - Setting up a Pre-Defined Field

A job was then created and run.

As a further investigation into the effects of temperature change, a number of incremental runs were simulated
from 5 degrees up to a maximum temperature change of 40 degrees.

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 48 of 57

Figure 62 – FEA Effects of Environmental Temperature Change

Figure 62 shows the almost negligible change in Tsai-Hill


Tsai Hill and Tsai Wu methods up to 40 degrees, the Maximum
Stress criteria decreases more rapidly during this period. The theoretical results follow the same form with a lower
safety factor as the temperature goes up but the theory shows a more rapid decrease (Figure
(Figure 63).

Figure 63 - Theoretical Effects of Environmental


Environ Temperature Change

There are a number of assumptions that were made prior to this investigation and also some improvements for a
more accurate representation.

The simulation and theoretical results do not account for any fluid inside the vessel, depending
pending on the fluid
contained in the vessel a temperature increase will also have an effect on the internal pressure exerted assuming the
volume is fixed. In the simulation, the pressure remained constant throughout the temperature changes.
changes

The materials properties of the laminate may also be temperature dependant.


dependant. (Parnas, 2002) explains, “Longitudinal
strength and modulus remain constant but off-axis
off axis properties are significantly reduced when temperatures
approach the glass transition temperature of the polymer”.
p During this simulation, no reference temperature was
given and no temperature dependant material properties were defined,, it is assumed that the temperature changes
in this investigation were based around standard room temperature (23 degrees Celsius).
Celsius).
Steven Goddard –Mechanical
Mechanical Engineering (Part Time)
Student Number: 10038749 Finite Element Analysis Report Page 49 of 57

5. Recommendations for Further Work


Some further work has already been discussed in the investigations above these include:

• Investigation of moving supports with gravitational effects applied.


• Investigation of moving supports with hygrothermal effects applied whilst fixing one support axially.
• Modelling the geometry of flanges, threaded inserts and other commercially available tube/pipe
connections and investigating the effect on the vessels strength.
• Investigating the effect of having various different gases in the pressure vessel.
• Simulating temperature dependant data.
• Combining temperature, moisture and gravity effects on the model.

Additional investigations could also try to understand the affect of loading other than internal pressure such as
impact or the effects of delamination within the composite. Moisture effects to the composite material could
also be compared with the calculations set up within the initial spreadsheet.

Buckling of the CPV could also be taken into account as a potential failure mode along with the effect of adding
a metallic liner which is common to the conventional method of filament wound composite pressure vessel
manufacture.

6. Conclusion
In this work, an initial ply layup was created based on laminate theory. A design including inlets, outlets and
supports was then developed using FEA software until a design that reached a safety factor of 1.2 was achieved.
This design was then used as a benchmark to compare against a range of effects applied to the vessel.

From the majority of investigations it is clear that the Tsai-Hill and Tsai-Wu failure theories result in more
conservative results than the maximum stress failure criteria, the literature reviewed in Section 1 shows that
Tsai-Wu criteria is generally the more preferred method for composite analysis. The Tsai-Wu criteria uses
numerous parameters and is an interactive tensor polynomial method that accounts well for the differences in
direction dependant material properties. From the results of the investigations, the Tsai-Hill and Tsai-Wu results
are similar. Therefore, it is recommended a factor of safety can be confidently estimated from these methods as
opposed to the Maximum Stress criterion. The only limiting factor is in a compression failure mode where the
Tsai-Wu criteria is less effective.

It is apparent that the theory used to determine the initial ply layout is not suitable as a comparison for the more
complex geometry and loadings of the further investigations. The use of FEA software with an appropriate mesh
and a suitable combination of boundary conditions and loading is vital to analysing complex structures in more
detail.

As with all simulation and theoretical based work it is paramount that safety factors are reinforced by a suitable
degree of experimental testing. This is especially important when dealing with pressurised fluids.

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 50 of 57

References
Önder,A (2007) First Failure Pressure of Composite Pressure Vessels, MSc, Dokuz Eylül University.

Sirosh N (2002), U.S DOE Hydrogen Program Review [online], Irvine, California, USA. Available from:
http://www1.eere.energy.gov/hydrogenandfuelcells/pdfs/32405b27.pdf [Accessed 04 December 2014]

McLaughlin, P. et al (2011) Composite Overwrapped Pressure Vessels [online]. Report Number: NASA/SP–2011–573. Johnson Space Center,
Houston, Texas: NASA. Available from: http://ston.jsc.nasa.gov/collections/trs/_techrep/SP-2011-573.pdf [Accessed on 04 December 2014].

Tam, W. et al (2002) Design and Manufacture of a Composite Overwrapped Pressurant Tank Assembly [online]. Report Number: AIAA 2002-
4349. Pressure Systems Inc. Available from: http://www.psi-pci.com/Technical_Paper_Library/AIAA2002-4349%20Astrolink%20Pres.pdf
[Accessed on: 04 December 2014]

Miaris, A (2010) Continuous Impregnation of Carbon-Fibre Rovings, JEC Magazine [online]. Volume 56. [Access 04 December 2014]

Evans J T et al (2002) Composite Angle Ply Laminates and Netting Analysis, Proceedings of the Royal Society of London. Series A: Mathematical,
Physical and Engineering Sciences. Volume 458, Issue 2028. Pages 3079-88.

Shultz M. Et al (2004) SEM X International Congress & Exposition on Experimental & Applied Mechanics [online], Washington State University,
Society of Experimental Mechanics. Available from: http://sem-proceedings.com/04s/sem.org-SEM-X-Int-Cong-s009p01-Optimal-Fiber-
Orientation-Fiber-Reinforced-Pressure-Vessels.pdf [Accessed 04 December 2014]

Xia, Y. Amali, R. (2014) Composites Analysis and FEA Course Notes. Simulation & Modelling [online]. Avaialble from: MyUWE [Accessed
05/12/14]

Tarmac Building Products (2014) Durox Supabloc Aircrete Blocks. Avaialble from:
http://www.tarmacbuildingproducts.co.uk/products_and_services1/blocks_and_mortar/blocks/lightweight_aircrete_blocks/durox_supabloc.a
spx [Accessed 06 December 2014]

Bakker, A. (2002) Applied CFD – Lecture 7 – Meshing [Presentation]. Available from: http://www.bakker.org/dartmouth06/engs150/
[Accessed: 07 December 2014]

Kurtz, R.D and Sun C.T. (1990) Composite Shear Moduli and Strengths from Torsion of Thick Laminates. Composite Materials: Testing and
Design. Volume 9, ASTM STP 1059, American Society for Testing and Materials. Philadelphia, 1990, P 508-520

Eng-Tips.com [Online] (2011) Available from: http://www.eng-tips.com/viewthread.cfm?qid=295070 [Accessed 09 December 2014]

Abaqus User Forum [Online] Available from: http://abaqus-users.1086179.n5.nabble.com/question-about-lamina-material-properties-


td9547.html [Access 09 December 2014]

Zick, L.P. (1951), Stresses in Large Horizontal Cylindrical Pressure Vessels on Two Saddle Supports (Revised Edition) [Online]. The Welding
Journal Research Supplement. [Accessed 10 December 2014]

British Standard Instituion (1997) BS 5500 Unfired Fusion Welded Pressure Vessels. London: British Standards Institution

ASME (1989) Boiler and Pressure Vessel Code, Section VIII. New York: ASME

Khan S. Stress Distributions in a Horizontal Pressure Vessel and the Saddle Supports [online] International Journal of Pressure Vessels and
Piping. Volume 58, Issue 5, Pages 239-244. [Accessed 10 December 2014]

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 51 of 57

Bibliography
Butt. A. M. (2010), Comparative Study for the Design of Optimal Composite Pressure Vessels. Key Engineering Materials Vol 442. Pp 381-388.
Trans Tech Publications, Switzerland.

Federal Aviation Adiminstration (2013) Chapter 7. Aviation Maintenance Technician Handbook, Volume 1.

Echtermeyer A. (2013) Safety Factors and Test Methods for Composite Pressure Vessels. D6.3. Delivery Hy.

Nettles A.T. (1994) Basic Mechanics of Laminated Composite Plates. NASA-RP-1351. Marshall Space Flight Centere, Alabama: NASA

Wang J. Et al. (2011), Stability Analysis in Wound Composite Material Axial Impeller. Journal of Mechanical Engineering Science. ImechE,
London. Available: http://pic.sagepub.com/content/early/2011/09/24/0954406211420335. [Accessed 11 December 2014]

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 52 of 57

Appendix 1

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 53 of 57

Enlarged Contour Plots of Stress Distribution when Modifying the Distance between Supports

1500mm

S11

S22

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 54 of 57

S12

1400mm

S11
Steven Goddard –Mechanical Engineering (Part Time)
Student Number: 10038749 Finite Element Analysis Report Page 55 of 57

S22

S12

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 56 of 57

800mm

S11

S22

Steven Goddard –Mechanical Engineering (Part Time)


Student Number: 10038749 Finite Element Analysis Report Page 57 of 57

S12

Steven Goddard –Mechanical Engineering (Part Time)

Das könnte Ihnen auch gefallen