UQ-Deisgn of a CF Composite Chassis Thesis

Benjamin Lii 4/26 Durham St. St. Lucia QLD, 4067 30th October 2009

Professor D.J. Mee Head School of Mechanical & Mining Engineering University of Queensland Queensland 4072

Dear Sir, I hereby submit my Thesis titled “Design and Manufacture of a Composite Monocoque Chassis” for consideration as partial fulfilment of the Bachelor of Engineering degree. All the work contained within this Thesis is my original work except where otherwise acknowledged. I understand that this thesis may be made publicly available and reproduced by the University of Queensland unless a limited term embargo on publication has been negotiated with a sponsor. Yours sincerely

Benjamin Lii Student ID: 41013580

Abstract This thesis report covers the design and manufacture of a composite monocoque chassis for a Formula SAE race car. A composite monocoque chassis is potentially much lighter with better mechanical performance than traditional designs. This has applications in racing where it provides increased performance, and also in more commercial automotive applications where fuel efficiency is becoming an ever greater concern. The existing space frame chassis was analysed in the first step of the design process and we determined that it weighs 23kg and a torsional stiffness of 6,820 Nm/deg/m. The maximum loads on the chassis were also investigated, and it was determined that the largest load on the chassis was 4,880 kN through the front arm of the lower wishbones under braking. This load case was used as the basis of insert design. The insert design was based primarily on physical testing. A rig was designed to hold a variety of test panels in place while the design loads were applied. The testing was conducted in 3 parts as results from the first tests enabled design improvements to be made, and these were subsequently tested for validity. From the testing it was determined the sandwich panels would not be able to support the design load regardless of insert configuration. Based on this conclusion, the design was altered to incorporate internal braces at the lower suspension supports to carry the large braking loads. The part geometry was realised through the use of Computer Aided Design (CAD) software. It was used to build a design which accommodates the components within the car, and the rules of the F-SAE class. The shock packaging was redesigned for the monocoque chassis as the existing configuration created too many compromises in the design. The final design analysis was performed using Finite Element Analysis (FEA) software Strand7 with this analysis validated with the mechanical testing of beam samples. In the validation process the datasheet elastic modulus for the composite material was revised from 65 GPa to 49.9 GPa. This loss of stiffness was due to the manufacturing process and the corresponding reduction in fibre volume fraction of the part, estimated to be having been reduced to 0.41 from the datasheet value of 0.55. i

The part was manufactured as a single component using a pre-preg composite system in a 2 part mould. The male moulds are expanded polystyrene foam with a fibreglass skin and a 6mm thick outer layer of tooling paste which is machined to an accurate finish via a Computer Numerical Controlled robot machining process. The mould is hand sanded and painted to a polished finish before being used to make a female mould. The final part is laid up within this female mould and cured in an oven at vacuum and temperature. Once cured and separated from the moulds, the edges of the part were trimmed to size before roll hoops and suspension support braces were fitted. The finished design weighs 18 kg, and has a torsional stiffness of 21,473 Nm/deg/m. The major structure is a single symmetrical sandwich structure with 5 plies of carbon fibre (MTM56/HTA5131) 2x2 twill weave face sheets with a 16 mm aluminium honeycomb core. Steel roll hoops are required by F-SAE rules and these are structurally integrated into the chassis and act as stiffeners as well as roll over protection. Aluminium inserts are used where major structural loads are carried. The composite monocoque chassis represents a significant step forward in the mechanical performance of the chassis compared to the existing triangulated steel tube frame. It will enable UQ Racings 2010 car to be lighter, stiffer, stronger and safer than the 2009 car. Combined these attributes will help lower the lap times of the car about a racetrack, the ultimate goal of any race car. More importantly a knowledge base now exists at UQ Racing for further composite structures to be rapidly designed and manufactured.

ii

Acknowledgements A. Prof. Martin Veidt: for supervising the project, and providing valuable guidance and insight throughout the project lifespan. Liam and staff at LSM Advanced Composites: for not only providing sponsorship but valuable advice, without which the manufacturing process might have been a much more expensive, mistake strewn and time consuming process. UQ Racing: supporting the project financially. Callum Jensen: for organising the part manufacture and assisting in testing. Graham Ruhle: for patience in teaching the finer points operating the Instron testing machine as well as valuable advice in designing the mechanical testing rigs and procedures. Mark James: for fielding last minute requests for components to be manufactured in the TSU workshop.

iii

Contents 1.

Introduction .......................................................................................................................................... 1 1.1.

Thesis Outline................................................................................................................................ 2

1.2.

The Aim ......................................................................................................................................... 3

1.3.

The Objectives............................................................................................................................... 3

1.4.

The Scope ...................................................................................................................................... 3

1.5.

Background ................................................................................................................................... 4

Composite Monocoque Chassis ............................................................................................................ 4 Composite Materials ............................................................................................................................. 6 Sandwich Structures ............................................................................................................................. 7 Formula SAE .......................................................................................................................................... 8 1.6. 2.

3.

4.

Anatomy of a Race Car .................................................................................................................. 8

Literature Review ................................................................................................................................ 10 2.1.

Composite Monocoques in the Automotive Industry ................................................................ 10

2.2.

Monocoque Design Process ........................................................................................................ 11

2.3.

Sandwich Structures and Honeycomb Cores .............................................................................. 13

2.4.

Corners and Joints....................................................................................................................... 15

2.5.

Load Introduction and Inserts ..................................................................................................... 16

2.6.

Finite Element Analysis of Composite Materials ........................................................................ 18

Design Approach ................................................................................................................................. 21 3.1.

Project Plan ................................................................................................................................. 22

3.2.

Schedule ...................................................................................................................................... 22

3.3.

Milestones................................................................................................................................... 24

3.4.

Budget ......................................................................................................................................... 24

3.5.

Resources .................................................................................................................................... 25

Design.................................................................................................................................................. 26 4.1.

Goal Definition and Load Analysis .............................................................................................. 26

4.2.

Sandwich Panel Selection ........................................................................................................... 28

4.3.

Insert Design and Testing ............................................................................................................ 30

Procedure ............................................................................................................................................ 31 Results ................................................................................................................................................. 34 Analysis and Design Modification ....................................................................................................... 37 iv

4.4.

Part Geometry............................................................................................................................. 39

4.5.

Beam Testing and FEA Validation ............................................................................................... 43

Test Procedure .................................................................................................................................... 43 Results ................................................................................................................................................. 45 Analysis ............................................................................................................................................... 46

5.

4.6.

Development of FEA Model ........................................................................................................ 48

4.7.

Evaluation of FEA Results ............................................................................................................ 53

4.8.

Sandwich Failure Modes ............................................................................................................. 57

Manufacturing .................................................................................................................................... 58 5.1.

Moulds ........................................................................................................................................ 58

5.2.

Lay Up.......................................................................................................................................... 61

6.

Testing ................................................................................................................................................. 63

7.

Overview of Final Design..................................................................................................................... 64 7.1.

Specifications and Features ........................................................................................................ 65

7.2.

Schematic Drawings .................................................................................................................... 66

8.

Design Evaluation................................................................................................................................ 68

9.

Conclusion ........................................................................................................................................... 69

References .................................................................................................................................................. 70 Appendicies ................................................................................................................................................. 72 Appendix A. Initial design schedule ............................................................................................................ 72 Appendix B. Torsional Stiffness Analysis of Chromoly Space Frame Chassis.............................................. 73 Appendix C. Material Datasheet MTM56/HTA5131 ................................................................................... 74 Appendix D. Material Datasheet EC Aluminium Honeycomb Core ............................................................ 75 Appendix E. Cockpit Opening Template ..................................................................................................... 76 Appendix F. Foot Well Template................................................................................................................. 77

Figures Figure 1 The rail chassis from a frame on rail design(Macogans Street Rods) ............................................. 5 Figure 2 The triangulated steel truss structure of a space frame chassis..................................................... 6 Figure 3 The location of composites on a material selection chart shows it is similar in stiffness to metals but much lighter (University of Cambridge, 2002) ....................................................................................... 7 Figure 4 The anatomy of a race car .............................................................................................................. 9 Figure 5 Finite element analysis of a F1 monocoque (Savage, Formula 1 Composites Engineering, 2009) .................................................................................................................................................................... 12 v

Figure 6 Strength, stiffness and durability testing on a composite F1 gearbox and rear suspension assembly (Savage, Formula 1 Composites Engineering, 2009) ................................................................... 13 Figure 7 Flex-Core description from the Hexcel product catalog(Hexcel, 2005) ........................................ 16 Figure 8 Overview of some of the methods for local load introduction into sandwich structures(Heimbs & Pein, 2009)............................................................................................................................................... 17 Figure 9 Numerical model of insert pull-out in a honeycomb sandwich panel (Heimbs & Pein, 2009) ..... 18 Figure 10 Revised schedule from end of semester 1 .................................................................................. 23 Figure 11 FEA model of the 2009 steel space frame chassis ...................................................................... 26 Figure 12 Stiffness, strength and weight vs core thickness (Savage, Formula 1 Composites Engineering, 2009) ........................................................................................................................................................... 28 Figure 13 Parallel axis theorem values(Pilling, 2005) ................................................................................. 29 Figure 14 An insert potted into the honeycomb core with syntactic epoxy .............................................. 32 Figure 15 Insert and bending test pieces .................................................................................................... 32 Figure 16 Insert test rig ............................................................................................................................... 33 Figure 17 Insert testing raw data ................................................................................................................ 34 Figure 18 Insert testing results shown after toe compensation has been applied .................................... 36 Figure 19 Shear failure in the honeycomb core is evident in these test pieces ......................................... 36 Figure 20 Finite element analysis results of monocoque with bracing installed........................................ 38 Figure 21 Early concept mock-up of design ................................................................................................ 40 Figure 22 Render of foot well area showing template location, cross brace and suspension configuration .................................................................................................................................................................... 40 Figure 23 CAD process from template to finished part .............................................................................. 42 Figure 24 Bending test rig configuration .................................................................................................... 43 Figure 25 FEA test beam setup ................................................................................................................... 44 Figure 26 test beam in 4 point bending rig ................................................................................................. 45 Figure 27 Raw data from beam 2 test 1...................................................................................................... 45 Figure 28 4 point bending results FEA and experimental ........................................................................... 46 Figure 29 Bulk modulus vs fibre volume fraction ....................................................................................... 48 Figure 30 Plates showing their local axis, it was important that these were all aligned in the same direction ...................................................................................................................................................... 49 Figure 31 CAD model and FEA model ......................................................................................................... 50 Figure 32 Laminate configuration ............................................................................................................... 51 Figure 33 Plate deflection and axial forces for braking load case, maximum deflection 0.5mm ............... 53 Figure 34 FEA results, braking load case showing safety factors................................................................ 55 Figure 35 Torsion stiffness model with plate elements hidden to reveal the rigid links and moment application .................................................................................................................................................. 57 Figure 36 The EPS core of male master moulds ......................................................................................... 59 Figure 37 The male mould after bog is applied and machined to final size ............................................... 59 Figure 38 The finished male mould for the shock cover, note the highly polished surface. ...................... 60 Figure 39 The chopped paste female mould formed over the male shock cover mould........................... 60 Figure 40 The shock cover laid up with the pre-preg material and core reinforcements .......................... 62 vi

Figure 41 The shock cover vacuum bagged and ready for cure ................................................................. 62 Figure 42 Chassis torsion test rig (Bateman, 2005) .................................................................................... 63 Figure 43 Final design ................................................................................................................................. 64

Tables Table 1 Budget estimate, GST non-inclusive............................................................................................... 25 Table 2 Design specifications ...................................................................................................................... 27 Table 3 Stiffness vs layup ............................................................................................................................ 29 Table 4 Materials, datasheets in Appendix C and D. .................................................................................. 29 Table 5 Insert test piece configuration ....................................................................................................... 31 Table 6 Load and freedom cases................................................................................................................. 52 Table 7 FEA model statistics ....................................................................................................................... 52 Table 8 Minimum safety factors in each load case ..................................................................................... 54 Table 9 Torsional stiffness analysis of monocoque .................................................................................... 56 Table 10 Comparison of composite monocoque and cromoly space frame .............................................. 65 Table 11 Mass breakdown of composite monocoque................................................................................ 65

vii

1. Introduction This thesis aims to tackle the lack of knowledge in composite component design by embarking on a project to design and manufacture a composite monocoque race car chassis. The project has been facilitated by the University of Queensland’s (UQ) participation in the Formula SAE (FSAE) student design competition through the university race team UQ Racing (UQR). The project has been assisted by the sponsorship of LSM Advanced Composites which has lessened the financial impact of the project and provided practical knowledge on the manufacture of composite components. While the composite monocoque chassis is widely thought to be the best chassis solution for a race car, the knowledge to design one is not. The study of the design of composite materials is not a standard part of the university syllabus and is focused on the understanding of its properties and behaviours but not on the design of complex components utilising it as a primary material. Within UQ Racing composite materials are more familiar as body work on the car has in the past been made from pre-preg composite materials. More significantly carbon fibre wheels were designed in 2004 and were run successfully on the car, however no evidence of the design or manufacture process remains apart from the moulds used in their production. The two main issues when it comes to fielding a composite monocoque chassis are cost and lack of design knowledge. The cost issue has been addressed through sponsorship by LSM and an improvement in the team’s financial situation. This thesis intends to address the second issue which is the general lack of knowledge in regards to composite components. The project in its simplest form is to design and manufacture a composite monocoque chassis for the 2010 UQ Racing F-SAE campaign. This will be as close as possible a direct replacement of the 2009 steel space frame chassis.

1

1.1.

Thesis Outline

This thesis report is arranged slightly differently to a traditional thesis report as it is design project. The report is organised into the main sections below: •

(Section 1) Introduction, aim, objectives, scope and background. This gives an overview of the project, why the project exists, what will be achieved and the reader a brief outline of the subject area.

•

(Section 2) Literature review. A summary of information gathered and important points which are relevant to the project. The review covers monocoques in Formula 1 and other automotive sources, the optimisation of sandwich structures and the design and testing of inserts in sandwich structures as well as other relevant topics.

•

(Section 3) Design Approach. Here we look at the project structure including the steps in the design as well as project details such as budget and scheduling. The manufacturing side of the project was not 100% complete due mainly to the global financial crisis and this and other factors are outlined here.

•

(Section 4) Design. The calculations, computer models and testing results are presented in this section.

•

(Section 5) Manufacturing. The process in which the monocoque is manufactured is outlined including notes on pitfalls and successes.

•

(Section 6) Testing. The testing regime which would validate the design and clear the monocoque for safe use is outlined.

•

(Section 7) Design Presentation. All of the critical specifications of the design presented in a condensed format.

•

(Section 8) Design Evaluation. A critical review of the design portion of the project looking at what was done well and what could have been done better.

•

(Section 9) Conclusion. 2

1.2.

The Aim

To design and manufacture a composite monocoque, for the UQ Racing F-SAE team to campaign in 2010. It should be interchangeable with the current chromoly space frame design utilising the current suspension geometry with minimal modification and on par with the current chassis in terms of weight, strength, stiffness, and driver safety. It should be compliant with F-SAE technical regulations.

1.3. •

The Objectives To research and develop and understand of the topic at hand, this includes composite materials in general and specifically the use of composites in motorsports.

•

To conduct initial calculations to determine the major design requirements of the chassis.

•

To use Computer Aided Design (CAD) software to visualise a model of the final component and also integrating major subcomponents.

•

To select an insert design based on mechanical testing.

•

To validate a Finite Element Analysis (FEA) software package with mechanical testing.

•

To use FEA software to analyse the overall structural performance of a monocoque design. The analysis should include stiffness, strength, failure modes. The FEA results should be validated with physical testing.

•

To produce detailed drawings and CAD files which communicate the design effectively to ensure the component is manufactured as designed.

•

To manufacture the designed component.

•

To validate the designed component in static and dynamic testing apparatus to correlate mechanical performance and ensure the safety of the final product.

1.4.

The Scope

Due to the possible breadth of the project the scope has been selectively narrowed down to a set of minimum requirements to make the project feasible within the 2 semester timeframe. The scope of the project is set out below:

3

•

Design of a carbon fibre and aluminium honeycomb cored monocoque chassis, the structure will extend from and include the rear roll hoop. It will be appropriately designed to support and protect the driver and support the front suspension, impact attenuator and other sub-components. However the design will not include the seat, or other subcomponents.

•

Design of main structural geometry for to meet stiffness goals and integration of attached components, using Computer Aided Design (CAD) and Finite Element Analysis (FEA) software packages. The analysis will be conducted in a static environment using the maximum dynamic loads.

•

Design of inserts to support the attachment of major components. This does not include the design of any external mounting hardware required.

•

Production of any required documentation or files necessary for the manufacture of the component/s.

•

Manufacture of the final design at LSM’s facilities.

•

Crash analysis requiring the use of non-linear explicit solvers.

•

The cost report and structural equivalency forms required by F-SAE are outside of the scope of this project. These would be produced by the team closer to the 2010 competition date.

1.5.

Background

This background will give a brief overview of the composite monocoque chassis, composite materials and sandwich structures that form the basis of this design. Finally a brief introduction to UQ Racing and F-SAE is given as this annual event provides a platform on which this design could be created. A thorough technical review of the relevant design issues is presenting in the literature review (Section 2, page 10). Composite Monocoque Chassis

The term monocoque comes from the Greek word for single (mono) and the French word for shell (coque)(Wikipedia, 2009). As the name suggests a monocoque design has a single shell that carries external loads. The monocoque is an extremely efficient structural design, and like many automotive technologies has its roots in the aeronautical field. In World War 2 aircraft 4

designers braced truss structures with structural skins to increase the stiffness. Designers soon found that there were advantages in making the skin thicker and minimising tubular truss structure. In this way the loads were carried through the skin of the aircraft instead of by an inner bracing structure. The use of sandwich construction furthered these designs as they made it possible to make efficient, lightweight and extremely stiff skins. Over the past 50 years, car designers have begun the migration from structurally less efficient traditional designs such as frame on rail and steel space frames to monocoque designs, albeit mainly from pressed steel and in some cases aluminium instead of expensive composites. One part of the automotive industry which has embraced the composite monocoque chassis more rapidly than others is motorsports, the most prominent being Formula 1. The composite monocoque chassis is currently also used in a limited amount of exotic road cars, and by some of the top F-SAE teams.

Figure 1 The rail chassis from a frame on rail design(Macogans Street Rods)

The adoption of the composite monocoque chassis has allowed designers to make F1 cars lighter, faster and much safer than in the past (Savage, Formula 1 Composites Engineering, 2009). However they are more expensive and hence are only present in parts of the industry where the extra costs can be justified. Light weight lowers the mass of the car and allows the vehicle to accelerate more rapidly in all directions, while improved stiffness makes suspension tuning more accurate and the dynamic performance of the vehicle more predictable.

5

Figure 2 The triangulated steel truss structure of a space frame chassis

Composite Materials

In this project we are using a subset of composite materials called Fibre Reinforced Composites (FRC), specifically carbon fibres in an epoxy matrix; this is generally referred to simply as carbon fibre. The fibres in this carbon fibre are what give it, its unique properties, that are its high stiffness and strength in the axial direction of the fibre in relation to the epoxy matrix making it an anisotropic material. Carbon fibre is usually available either as a unidirectional product where all the fibres are in the same direction or a woven product where the fibres are woven perpendicular to each other. Carbon fibre has a higher specific strength and stiffness compared to traditional materials such as steel and aluminium. This coupled with its excellent fatigue performance has seen become a significant material in the aerospace industry. However the high cost of the material means that in situations where the high performance is not necessary such as in much of the automotive industry, the material has found a lot less use.

6

Figure 3 The location of composites on a material selection chart shows it is similar in stiffness to metals but much lighter (University of Cambridge, 2002)

Sandwich Structures

Sandwich structures are able to significantly increase the specific stiffness of a part, with only a slight increase in weight. The sandwich structure is analogous to an I-beam, which separates the top and the bottom sections which carry the majority of the load, with a thin web section in the middle. This can result in a large increase in the second moment of area without increasing mass, greatly improving the structures resistance to bending. In a sandwich structure the top and the bottom are thin face sheets chosen for their high stiffness, and the core (middle section) is often a light weight material such as foam or a honeycomb. In this configuration similar to an I-beam, the structure carries loads bending loads much more efficiently. A properly designed sandwich structure can be many times stiffer then a solid design of similar weight and material. The downsides with this type of structure is it can be harder to determine the strength of the design as there exists more failure modes such as the wrinkling and delamination of face sheets as well as crushing of the core.

7

Formula SAE

Formula SAE (F-SAE) is a worldwide competition where student design teams compete in the development, manufacture, presentation and racing of a small formula style race car. The competition was originally run in 1980 in the USA, but has since spread worldwide. The competition consists of three key elements; engineering design, cost and static inspection, solo performance trials, and high performance endurance test(Society of Automotive Engineers Australasia, 2009). The competition gives students the opportunity to develop their skills by participating in a hands-on engineering project. The students manage the complete project from finding sponsors, scheduling and the design, to the manufacture and testing of components. UQ’s participation in this competition through their entrant UQ Racing has given this project an ideal environment in which to develop. The requirements of UQ Racing has given this project goals for which to design towards.

1.6.

Anatomy of a Race Car

The main structural component of a race car is known as the chassis; in this case it may also be referred to as the driver’s safety cell as it is where the driver sits and acts as a passive safety device. 1) Tyres, all of the loads on the car are reacted at the tyre ground interface. The goal of any race car is to maximise the performance of this interface allowing the car to accelerate, corner and brake more rapidly. 2) Wheels, these support the tyres and carry the load into the hub. The wheels spin along with the hub, and the assembly is supported by bearings with an upright. 3) Upright, this supports the bearings which hold the wheel and tyre package. The upright is connected to the chassis via the suspension arms. The uprights may also support other loads such as in this case the front brakes. Together with the wheels and tyres, these components form the unsprung mass of the car.

8

4) Suspension Arms, in this case they are in a double wishbone configuration. These are attached to the uprights with spherical bearings which allow total freedom of rotation. They are attached to the chassis at the suspension pickup points. 5) Coilovers/Shocks, are a unit combining a spring and dampener. This regulates the dynamic motion of the vehicles suspension, and tuning of the springs and dampener are an important part of setting up a race car. Together with the wishbones, uprights, tyres and wheels this forms the suspension system of the car. In this case the suspension is similar in the front and rear however this is not always the case. 6) Engine, the engine provides the motive force to move the car. In some designs the engine also forms part of the structure of the car often supporting the rear suspension. 7) Roll hoop, the roll hoop is a structural component whose primary role is to protect the driver in case of the car overturning. A roll hoop is also present at the front of the cars cockpit and together with the rear roll hoop this forms a safety envelope for the driver.

Figure 4 The anatomy of a race car

9

2. Literature Review A wide variety of sources were researched Both traditional articles such as books and journal articles and more informal information garnered from various sources primarily from the internet were used in this literature review. This includes forum postings which often were single sentences with no guarantee of relevance or accuracy. These were considered with great care and assumed to be false unless independently verified. Pictures of Formula 1 and other FSAE monocoques were also useful as a visual reference. The fairly elusive nature of the subject is the main reason why these sources were used as an additional source of information. A quick refresher on composite materials, and a review of classic lamination theory enforced the fact that when dealing with composite materials there is often 2-3 times more material data required to accurately define the material in comparison with traditional materials. It was also abundantly clear that physical testing would need to be an integral part of the design process. Lamination theory and basic knowledge of sandwich structures are not covered in this review, however we can recommend Composite Materials: Design and Applications (Gay & Hoa, 2007) for a basic review of composite materials. Honeycomb Technology (Bitzer, 1997) is a good resource dealing with a variety of core materials and also many practical considerations. Finally Finite Element Analysis of Composite Materials (Barbero, 2008) provides a good guide to using finite element tools to analyse composite structures, specifically with ANSYS.

2.1.

Composite Monocoques in the Automotive Industry

On the surface there was a wide variety of sources describing the superior performance of composite monocoque chassis’ in exotic applications. In non-motorsports applications, super cars such as the Porsche Carrera GT and the McLaren Mercedes-Benz SLR utilise an allcomposite primary structure(Marsh, 2006). These cars have inherited much from their company’s motorsports development programs and utilise a central composite monocoque safety capsule from which substructures are attached. The SLR also utilises composite crash attenuators which are conical CFRP members 620 mm long and weighing 3.4 kg which absorb crash energy through deformation. Similar to the SLR the Porsche Carrera GT has a composite monocoque chassis and this was chosen because the manufacturer wanted a design with the minimum weight. Other advantages were noted by the manufacturer in that the low thermal 10

coefficient of expansion of the composites allowed tighter tolerances, and the single integrated component reduced the parts count. However all these advantages come at a price, both the SLR and the Carrera GT cost well over $500,000. Moving into the motorsports sector, composite monocoques are used in classes such as the German DTM series, Le Mans Prototypes and most recognisably, in Formula 1. In Formula 1 composites and other advanced materials have produced cars that are lighter, faster and safer than ever before(Savage, Formula 1 Composites Engineering, 2009). These statements reinforce the possible advantages that a composite monocoque chassis would offer UQ Racing and the reasons for pursuing the design. The use of a composite monocoque design for a Formula SAE application was also investigated by a previous member of the team (Bateman, 2005). The thesis report concluded that a monocoque design would be a far superior choice; however the results of the calculations leave a lot of doubt as to their accuracy. The preliminary calculations in the thesis indicate that a monocoque chassis of similar specifications to the existing space frame would be 6 orders of magnitude stiffer, a claim that does not seem accurate.

2.2.

Monocoque Design Process

The design process for a structural composite component has many unique features due to the complex properties and behaviour of composites as well as the different manufacturing methods in comparison with traditional materials. The article Formula 1 Composites Engineering (Savage, Formula 1 Composites Engineering, 2009) has a good overview of the entire design process and the article is summarised here. In Formula 1 the process begins with concept studies which look at balancing aerodynamics, suspension dynamics as well as mechanical performance of the design, often some compromises have to be made. The majority of the concept work is investigated through CAD, and it is usually only once a design has been finalised that it is produced. Advanced immersive visualisation tools also exist now for designers to interact directly with the concept in a virtual space. 11

The performance of the final part is simulated using a FEA package. The complex mixture of forces which act on the chassis and the equally complex part geometry means that in F1 the design work is a combination of experience and FEA analysis. Hence data acquisition is very important in F1 as the basis for a good design, is a good problem definition. However it is important to note that at present no matter the sophistication of the FEA package, certain assumptions and estimations are made and mechanical testing to verify results are essential before committing to the final design.

Figure 5 Finite element analysis of a F1 monocoque (Savage, Formula 1 Composites Engineering, 2009)

Manufacturing also presents problems which are unique to composite materials, and these must be considered in the design process. The lay-up process used to produce composite parts is significantly different to metal manufacturing processes which designers may be more familiar with. Complex geometry in particular can cause problems as certain shapes may not allow plies to be formed around them, or cause wrinkles to form. These discontinuities in the fibres can affect the mechanical performance of the part. Also as the parts are formed within moulds there needs to be thought put into how the part will be separated from the mould after manufacture. Complex multi-part moulds can be used, but it may be more efficient to simplify 12

the design for ease of manufacture. Finally there should also be attention paid to the size of plies as it is possibly to exceed the available material size. From this we can understand that there is a much wider variety of issues facing the designer of a composite part. This means that design validation through physical testing is essential throughout the design process. This emphasis on physical testing was integrated into the design process used in this project as it suited the resources we had available. All major design decisions were based on the results of physical testing.

Figure 6 Strength, stiffness and durability testing on a composite F1 gearbox and rear suspension assembly (Savage, Formula 1 Composites Engineering, 2009)

2.3.

Sandwich Structures and Honeycomb Cores

The design of sandwich structures presents an additional challenge as there are certain assumptions made which will affect the accuracy of the final result. One of the most common assumptions made in the design of sandwich structures the in plane modulus of the core is assumed to be zero (Barbero, 2008). Another significant assumption that is often made is that the core shear modulus is often ignored. This can have a large affect on the total deflection calculated which is a sum of the deflections due to bending and shear. It is important to be aware of these assumptions as they will often highlight certain limitations in theoretical and FEA results.

13

However despite these additional design complexities sandwich cores present an opportunity to have an extremely efficient structure (Allen, 1969). This efficiency is so pronounced that the mechanical performance of an optimised sandwich structure can often be an order of magnitude greater than a poorly designed sandwich. Importantly this can also lead to designs which may be orders of magnitude cheaper (Walker & Smith, 2002). The use of honeycomb cores also present additional challenges as they are in fact a complex 3D structure in themselves with complex failure modes and behaviours. The hexagonal geometry of a honeycomb means that it has unique properties in different directions, making them highly anisotropic; perhaps more so then the fibre reinforced composites we use in the face sheets. Specifically with the expanded aluminium honeycombs we are using in this design, the transverse shear properties also vary in with the applied shear load (Bitzer, 1997). Because the lightweight aluminium honeycombs are formed by having alternate glue spots on thin foil which is then expanded, some walls are double thickness which alters not only the shear strength but stiffness. Analytical solutions (Qiao & Wang, 2005) exist however testing would still be necessary to validate results. For the purpose of this design a simplified approach is taken using the bulk properties from material data sheets validated with physical testing. The multitude of variables present in the design of a sandwich structure means that in many cases computational methods are used to optimise the structure for cost (Walker & Smith, 2002) and structural performance (He & Hu, 2008) and efficiency (Qiao & Wang, 2005). For the purposes of our design these analysis are useful, however due to the necessary performance requirements of our design, physical constraints and availability of certain products the analysis has a much narrower focus. In the design of a serial production component or possibly in applications where cost is a much smaller factor, more variables may be present as the economies of scale may allow customised core sizes. As a final consideration, the light weight and fragile nature of honeycomb cores means that they impose certain restrictions in manufacture of composite components. Because of the low strength of many of the core materials in use, the pressure at many pre-preg composites are cured in an autoclave can crush the core, and the components must be cured at lower pressures 14

(Bitzer, 1997). This would obviously have a negative effect on the final strength of the product, but the efficiency of a honeycomb structure typically far outweighs this. Never the less it is an important consideration.

2.4.

Corners and Joints

Corners and joints are a feature of all but the most simple of designs, and while a designer may strive to “achieve a smooth flowing shape avoiding sharp corners and joggles which act as stress concentrations”(Savage, Formula 1 Composites Engineering, 2009) this is not always possible and an understanding of how these features affect the design is important. In designing a corner in a sandwich structure it is important to be aware of how the materials will behave at the corner. The strength of the fibre reinforced composites is dominated by the tensile strength of the fibres and turning a corner negatively affects this strength(Gay & Hoa, 2007). However by introducing smooth corners we can reduce this affect. It is also important to avoid bridging the inside of the corner with the face sheets as this would compromise its performance (LSM, 2009). The behaviour of core materials around corners in particular honeycombs is more complex. Honeycombs tend to bend easier in certain directions then others, and when bent in certain directions can form a saddle like shape (Bitzer, 1997), this presents difficulties in forming the core to complex shapes. However there are products available on the market such as Flex-Core® which utilise a cell configuration that allows it to be formed into complex curves more easily. For the monocoque design we maintained large radius curves to dimensions recommended by LSM Advanced Composites to avoid any of these local effects.

15

Figure 7 Flex-Core description from the Hexcel product catalog(Hexcel, 2005)

Joints are also commonly found in many monocoques which may feature bulkheads that help feed suspension loads into the structure (Savage, Formula 1 Composites Engineering, 2009), as well as acting as stiffeners on larger structures. These bulkheads can be adhesively bonded, or attached using mechanical fasteners (Savage, Failure prevention in bonded joints on primary load bearing structures, 2007). For simplicity in this monocoque design we will avoid any joints, and attempted to make as much of the structure as possible one component.

2.5.

Load Introduction and Inserts

Load introduction is one of the most complex issues facing the designer of a sandwich structure. In search of efficiency, sandwich structures often feature thick fragile cores, with thin fragile face sheets that may form a component which is extraordinarily stiff and strong in certain orientations but quite fragile in many others. In these cases the introduction of loads may cause face sheets to fracture or buckle, and the core material to be crushed. Inserts of a solid material are often integrated into the component and this supports the structure locally and spreads the loads into the structure. Some of the possible insert configurations are shown in the figure below.

16

Figure 8 Overview of some of the methods for local load introduction into sandwich structures(Heimbs & Pein, 2009)

We can see that there are a wide variety of methods for introducing loads into the a honeycomb structure and some of the issues a designer has to consider beyond just the strength of the insert include additional weight, installation methods and stress concentrations around the inserts. A study conducted on inserts similar to configuration ‘f’ in the figure above found that the strength of the insert was strongly influenced by core thickness, core density, face sheet thickness, as well as face sheet laminate stacking sequence (Song, Choi, Kweon, Choi, & Kim, 2008). Various studies show numerical models which attempt to predict the performance of inserts. In one a numerical model is developed using FE-code Samcef, and this was able to predict the strength of a potted insert with good accuracy, however the process is quite complex and the authors suggest that it be used in non-conservative designs or when testing becomes too expensive. There are also limitations in the model specifically in dealing with the non-linear behaviour of the honeycomb core (Bunyawanichakul, Castanie, & Barrau, 2008). Another study used LS-DYNA and was able to accurately predict the onset and manner of failure of a potted insert (Heimbs & Pein, 2009).

17

Figure 9 Numerical model of insert pull-out in a honeycomb sandwich panel (Heimbs & Pein, 2009)

These methods were deemed to be too complex for the project, so to simplify the obviously complex design procedure, the process undertaken in this design was simply to identify the strongest insert configuration based on available literature, produce some samples and use physical testing to choose the final design. Based on the review of various articles we were able to deduce that inserts which engaged both face sheets were stronger than those that engaged one face sheet such as in configuration ‘f’ (Thomsen, 1998). Options such as ‘g’ and ‘h’ were avoided due to the added complexities in manufacturing. Option ‘d’ was discarded as the strength of epoxy would means that a screw threaded into it would result in a fairly low pull out loads. This left options ‘a’, ‘b’, ‘c’, and ‘e’ to be investigated in the design process. We will primarily be investigating the use of metal inserts however it is also possible to use more exotic materials such as monolithic blocks of 3D woven composite material as inserts (Naik, Rao, Agarwar, Raju, Pottigar, & Suresh, 2009)

2.6.

Finite Element Analysis of Composite Materials

Finite element analysis (FEA) is an important design tool for which was briefly mentioned in the section on the design process. FEA allows us to accurately predict the behaviour of complicated structures under the influence of complex loads and restraints. Like any tool, it is important to understand how it works; in particular, we need to understand the limitations of the FEA software due to assumptions made by the software to be able to accurately interpret the 18

results. As part of the project we assessed a variety of FEA software packages, as well as modelling methods to understand the capabilities and limitations of finite element analysis. The objective was to model the monocoque structure as a shell, and using the correct elements apply the appropriate laminate properties to this shell structure and then apply loads and restraints by importing the suspension geometry and major substructures into the program. The two programs that were investigated initially were Strand7, ANSYS and CosmosWorks. At the time CosmosWorks lacked the capability to model composite materials as this was only added in the 2010 version which was only recently released, CosmosWorks was therefore not considered as a solution. Strand7 was the program that the university taught FEA with, and as such was the first port of call. The main limitation with Strand7 was that it does not model the shear deformation in sandwich materials, therefore deformations may be calculated incorrectly and the failure criteria will not take into account shear failure of the core (Strand7, 2008). Its main advantages were that it has a fairly efficient user interface and a successful model was able to be built within a relatively short period of time. ANSYS was the other main software package considered. ANSYS has better elements included and these will model the shear effects within the core (Barbero, 2008) which gives it an advantage over Strand7. The disadvantages were that the user interface was not conductive of an efficient workflow. There was also a lack of documentation which meant that the process to import suspension geometry, creating restraints, and applying forces was not intuitive. A decision was made to use Strand7, as it would allow more time to be spent designing, and less time attempting to make the software do what was required of it. In the latter half of the project a Computer Aided Engineering (CAE) software package called HyperWorks became available. This package features a much improved user interface then both Strand7 and ANSYS and had a built in solver as well as the ability to export models to be solved in ANSYS (Altair Engineering, 2009). Additionally it included capabilities for the software to optimise variables such as core thickness and ply orientations (Gray & Aldous, 2009). This came 19

too late in the project, however it would be the recommended package for use in future as it allows the model to be rapidly created, features an accurate solver without Strand7’s element limitations and advanced optimisation possibilities.

20

3. Design Approach A design approach was devised which addresses the specific requirements in the design of a composite component. The main issues identified which differentiate the design of a composite component to traditional materials were: 1. The material properties are anisotropic meaning that the orientation of the material is important. This is not a significant factor in the design of most traditional materials. 2. The failure modes present in composite material are numerous and not as well understood. In a sandwich structure, even more failure modes are present, making analysis a complex process. 3. The material properties given in datasheets may not exactly represent the actual properties of the final product due to differences in manufacturing process. In this case the material properties for the composites used were given for an autoclave manufactured sample. In actual fact we are using only a vacuum cure process, and would expect a lower fibre volume fraction and hence slightly lower material properties. 4. The availability of design tools. With traditional materials it is quite easy to produce a cad model and use finite element analysis to easily predict the final performance of the part. These tools are also available for composite materials but their use mainly due to the issues noted in points 1 and 2, mean that producing an accurate result is a lot more difficult. Once these issues were identified a suitable design approach was determined. This approach has a emphasis on physical testing and was tailored to be the most efficient based on the available resources. The design approach is broken down into several stages, outlined below: 1. Identification of load cases and major requirements. 2. Selection of initial sandwich configuration based on stiffness vs weight vs cost. 3. Design of inserts via physical testing. 21

4. Qualification of FEA package using test beams in 4 point bending. 5. Creation of part geometry based on subcomponents and F-SAE rules. 6. Creation of FEA model. 7. Analysis of FEA model for deflections and identification of failure modes. 8. Finalisation of design, creation of drawings etc. 9. Manufacture of component. 10. Testing and commissioning of component.

3.1.

Project Plan

A project plan was developed which included a schedule, milestones, budget and resources. In this section we will have a brief review of these components as well as looking at issues which have caused the project to run behind schedule.

3.2.

Schedule

The project was originally scheduled to be completed over semester 1 and 2 of 2009. The initial project research and development occurred over the 1st semester, the final design was to be finalised by the winter break. The design was to be built over the winter break, and the second semester will allow for the design to undergo a qualification process and testing against design specifications. A Gantt chart which describes this original schedule and the major tasks is seen in Appendix A. Difficulties arose which meant that it was not possible to adhere to the original schedule. The FEA component of the design was delayed because of difficulties with the originally selected program ANSYS. Eventually a switch to Strand7 was made, and also investigations into using HyperWorks were initiated. The time needed to manufacture components were also underestimated, and this lead to a delay in the completion of test pieces. This prompted a reshuffling of the project schedule at the end of semester 1 and this revised schedule is present in the figure below. 22

Figure 10 Revised schedule from end of semester 1

There was ample time for the manufacture of the final part in the original schedule, despite the processes taking longer than expected. However an additional difficulty arose as the global financial crisis had an effect on the team and also the sponsorship. The effect on the team was that entry costs for the competition in 2009 rose from <$1000 to $5000 and this cut a large chunk out of the team’s budget. The main sponsor LSM was also affected and they were no longer able to sponsor the team the material cost, which would as shown in the budget, add another $3000+ to the project cost. The manufacture of the project therefore had to be revised as the original plan of using expensive tooling board as a mould material was no longer financially viable. The alternative was to use a tooling paste mould with an EPS core, this led to a blow-out in the time schedule as the process was untested and labour intensive. The process is described in more detail in the manufacturing section. At this stage the male moulds for the design are complete and it is expected that the manufacture will recommence shortly after the final examinations are 23

complete and the component finished within 2 weeks. This may change depending on the status of the 2009 car.

3.3.

Milestones

Based on the schedule a series of major milestones set out to measure the progress of the project. The main milestones and their completion dated are: Preliminary Design Completed:

9th April 2009 (mid-semester break, semester 1)

Interim Report Submitted:

5th June 2009 (end of semester 1)

Geometry Finalised:

27th July 2009 (winter break)

Design Analysis Complete:

29th September 2009 (mid semester break, semester 2)

Mould Built:

20th October 2009 (End Semester 2)

Chassis Completed:

Projected 1st December 2009

Chassis Static Tests Completed:

Projected 1st January 2010

Chassis Installed:

Projected 1st June 2010 (end of mid-semester break)*

Final Report Submitted:

30th October 2009 (end of Semester 2)

*The schedule of installing the chassis is highly dependent on whether or not the team decides to conduct testing using the existing space frame or the monocoque chassis.

3.4.

Budget

The project budget was allocated from UQ Racings budget. The project was originally allocated $2000, primarily to be spent on the purchase of materials for the tooling required to manufacture the monocoque. Additionally the team has a significant budget surplus this year and this would be accessed on a case by case basis. This budget is accessible through the UQ Racing via the University of Queensland finance department. However with the difficulties with the global financial crisis the project became more expensive as the team had to pay for materials. The expanded polystyrene (EPS) core of the male mould was donated by LSM.

24

Table 1 Budget estimate, GST non-inclusive

Item

Qty

Cost/Unit

Cost

MTM56

25.0 m2

$134/ linear metre

$3,250

ECM6.4

2.5 m2

$309 / 1250x2500 mm

$309

Tooling Bog

44 kg

$37 / 4kg tin

$370

Misc Tooling

$200 (MDF and Glue) Total to Date

$4,129

The values above are preliminary figures, and are give only an overview of the total cost. The eventual cost will most likely be slightly higher. There is also wastage however many of the smaller pieces of left-over carbon cloth were able to be used for small projects such as the production of carbon paddles for pneumatic shifting system. Other components such as the insert test rig and inserts were built from left over material in the workshop, and workshop time was sponsored through UQ Racing.

3.5.

Resources

The project has 3 main resources; 1. LSM advanced composites, which has sponsored the project with the use of autoclave facilities and donation of EPS foam for use in the male moulds. The use of this facility has been organised through Liam the founder of LSM, and access needs to be organised and cleared with him a few days before hand. 2. The TSU Mechanical Engineering Workshop, from which UQ Racing has sponsored workshop hours; these will mainly be used to produce testing apparatus, and any inserts and fixtures that the chassis may require. When components need to be manufactured, they are submitted with the appropriate technical drawings under UQ Racing’s job number.

3. The Instron Lab, was used for the mechanical testing of insert designs, and also for the bending tests whose results will be used to validate FEA results. Testing time needs to be booked through the laboratory manager Graham Ruhle. 25

4. Design The following section details the design steps.

4.1.

Goal Definition and Load Analysis

To begin with we need to analyse the existing chassis and determine the exact performance characteristics we require from the design. As stated in the introduction we wish to develop a monocoque chassis which will match the performance of the steel space-frame chassis. Based on past analysis (Bateman, 2005), FEA was determined to give a good match to the tested performance of a space frame component. Thus FEA methods were used to determine the stiffness of the space-frame, and statistics such as weight were also obtained from the computer model. The model is made from beam elements with the geometry manually transferred from the 3D CAD model of the space-frame and is shown in the figure below.

Figure 11 FEA model of the 2009 steel space frame chassis

26

Secondly we also obtain the loadings which will need to be supported by the chassis. To do this we first need to calculate the design loads on the car. The loads on the chassis are a result of inertial loads of components in the car due to acceleration, braking, cornering, gravity and combinations of the above being reacted through to the road via the tyres. The tyres on the FSAE car enable it to routinely generate loadings up to 1g, up to a maximum of 1.5g before the loss of traction limits the accelerations it is able to sustain. To measure the torsional stiffness of the space frame the car was fixed at the roll hoops and a moment parallel to the cars main axis was introduced through the suspension components. The deflection at the pickup points was measured together with the distance from the centreline of the car and we were able to calculate the rotation of each part, and hence the torsional rigidity. The reason for restraining the car at the rear roll hoop was to directly compare the stiffness of the composite tub against the space frame design independent of the rear of the car. This does mean that the rear of the car is restrained in a slightly unrealistic manner and it would be inaccurate to use this value to determine the stiffness of the whole vehicle, however it is valid for comparison purposes as the monocoque will be similarly restrained. The stiffness analysis of the steel space-frame chassis is shown in Appendix B. The two main design specifications, weight and torsional stiffness are shown in the table below: Table 2 Design specifications

Specification

Value

Weight

22.97 kg before welds.

Torsional Stiffness

6,820 N.m/deg/m

Equally important is the maximum loading which will need to be reacted by the chassis. To calculate the maximum loadings the FEA model was restrained by at the tyre contact patches. Translational masses were added to the model to simulate the actual mass of the vehicle, and linear accelerations were used to simulate the different loads which would be carried under combinations of acceleration, braking, bump and cornering as well as static loading. The maximum load was found to be during braking at the front pickup point on the lower 27

wishbones. The load calculated was 4,880 N. This value will be used to design the inserts which will be used to introduce the loads into the chassis.

4.2.

Sandwich Panel Selection

The skin/core thickness ratio has a large bearing on the stiffness to weight ratio and also the cost of a sandwich structure. Generally, the core material is significantly lighter and cheaper than the face sheets and the bending stiffness of the sandwich increases significantly by increasing the core thickness. In this case, the choice of core thickness was limited by packaging issues which meant that the only two choices available were 10 or 16mm thick aluminium core.

Figure 12 Stiffness, strength and weight vs core thickness (Savage, Formula 1 Composites Engineering, 2009)

Clearly increasing the core thickness dramatically improves the properties of a sandwich structure. 16 mm core material would be superior to the 10 mm in both stiffness and cost. However the larger loads and thinner face sheets of the 16 mm option means that it is more susceptible to failure modes such as face sheet buckling and yielding. The thickness of the face sheets were selected to be 5 plies. This value was suggested by LSM advanced composites, and was also an approximate value used in Formula 1(F1 Complete, 2008). The value was also confirmed as a suitable starting point by LSM. It was also determined that the layup suggested to approximately give the correct mass for the final component. From this starting point some calculations were performed using the parallel axis theorem the sensitivity of the material performance to changes in core thickness and ply layup.

 

         6 12 2 28

Figure 13 Parallel axis theorem values(Pilling, 2005)

In the table below we look at the results of this analysis on a range of sandwich structure configurations. Table 3 Stiffness vs layup

Layup

Core

Equivilent Flexural Rigidity, EIeq

[0,45,0,45]

10 mm

396

[0,45,0,45,0]

10 mm

534

[0,45,0,45,0,45]

10 mm

688

[0,45,0,45]

16 mm

903

[0,45,0,45,0]

16 mm

1186

[0,45,0,45,0,45]

16 mm

1493

The materials used are shown in the table below. These materials were selected because of LSMs familiarity with the particular products and their manufacturing characteristics. It was decided to use a woven fabric and no unidirectional materials to simplify the design and as woven materials were easier to handle. Table 4 Materials, datasheets in Appendix C and D.

Item

Material Used

Face Sheets

MTM56/HTA5131 199 gsm 3k 2x2 Twill, High strength pre-preg with MTM56 resin system and HTA5131 fabric.

10 mm Aluminium Core

ECM 6.4-82-10, 3003 series Aluminium honeycomb with 6.4 mm cell size, 82 kg/m3, 10 mm thick.

16 mm Aluminium Core

ECM 6.4-82-16, 3003 series Aluminium honeycomb with 6.4 mm cell size, 82 kg/m3, 16 mm thick.

29

The analysis in table 3 confirms our initial qualitative estimate that the 16mm core is far superior. However the 10 mm core was selected initially mainly due to availability and concerns about the formability of the 16 mm core. It is important to preserve the integrity of the honeycomb structure as we discover later as the first failure of the sandwich structure under our load cases usually occurs in the core due to buckling and shear failure of the honeycomb cell walls. Any damage to the core in the manufacturing process may advance the onset of this failure. Later in the project based on the radiuses that were achievable in the geometry of the part and advice from LSM, a switch was made to the 16 mm core as their experience was that it would be possible to form the core to the part geometry without damaging the core. The final analysis of the design was based on the 16 mm core.

4.3.

Insert Design and Testing

The introduction of out of out of plane loads into a sandwich structures is difficult and was the most challenging part of the design to resolve. The design of a sandwich structure is very much optimised towards carrying bending loads and large out of plane loads can cause issues as the core structure is very light weight, we risk crushing the core. Also while the face sheets are very strong for carrying in plane loads, normal loads expose the relatively low out of plane strength and stiffness of the face sheets. Therefore it becomes necessary to locally reinforce the structure at the point of load introduction via the use of inserts. The following figure shows a variety of different insert designs. For our design we have chosen initially to test a configuration similar to configuration ‘b’ in figure 8 (page. 17), this is because our loads are relatively large and this configuration would seem to be the strongest as it supports both face sheets, and uses a solid block to prevent crushing of the core. The question would be how large to make this insert. There were no analytical solutions available to accurately predict the strength of the chosen insert designs. Numerical methods may be possible, but are extremely complex and would need to be verified through physical testing. Therefore the most efficient method for the project to proceed based on the resources available, was simply to make an educated guess on the size of 30

inserts which would be required, build test pieces, and test them to determine which would be a suitable design. The chosen insert sizes were 30 mm, 50 mm and 70 mm in diameter, this and the panel configurations are shown in table 5. Procedure

The test pieces were prepared on a flat surface which acted as the mould and cured in the same process as would be used in the manufacture of the final part. The process is described in more detail in Section 5. The inserts were potted into the honeycomb core using a syntactic epoxy, which is an epoxy with Q-Cell to thicken it into paste. There is also a dark pigment used to colour the paste. A potted insert can be seen in figure 14. The test procedure was devised from the existing literature. A circular frame supported the test piece, and the force was applied through a round force applicator. The design was chosen based on a review of existing literature. In the journal article by Korean researchers (Song, Choi, Kweon, Choi, & Kim, 2008) similar panels were tested, they chose to use a 80 mm diameter supporting frame and 15 mm inserts. For our tests we modified this to a frame which was 140 mm in diameter with a 40 mm diameter puller. The tests were conducted using an Instron machine and the load vs deflection was logged. Because of the final size of our part, it was thought that this configuration would give a better correlation to the required performance of the inserts in the final design. Table 5 Insert test piece configuration

Panel Faces

Core

Insert

Comment Failure Load, N

i1

[0,45,0,45,0]

10mm

None

i2

[0,45,0,45,0]

10mm

30mm Al, full thickness

Type ‘b’

2,080

i3

[0,45,0,45,0]

10mm

50mm Al, full thickness

Type ‘b’

2,450

i4

[0,45,0,45,0]

10mm

70mm Al, full thickness

Type ‘b’

2,920

i5

[0,45,0,45,0,45,0]

10mm

50mm Al, full thickness

Type ‘b’

2,950

i6

[0,45,0,45,0]

16mm

30mm Al, full thickness

Type ‘b’

2,950

i7

[0,45,0,45,0]

10mm

50mm Epoxy

Type ‘a’

2,180

i8

[0,45,0,45,0]

10mm

70mm Epoxy

Type ‘a’

2,420

1,620

31

Figure 14 An insert potted into the honeycomb core with syntactic epoxy

Figure 15 Insert and bending test pieces

32

Figure 16 Insert test rig

33

The test was conducted at a fixed cross head speed of 0.5mm/minute. While the maximum design load was 4.880 kN it was decided to test the panels past this to 6 kN to better understand how the panels would fail. The load was brought to the final design load levels in 3 steps, a test to 2 kN, a test to 4 kN and a test to 6 kN.

Insert Testing - i4 - Load vs Extension 6000

5000

Load, N

4000

3000

i4 - 2kN return i4 - 4kN return

2000

i4 - 6kN return

1000

0 0

0.2

0.4

0.6

0.8

1

1.2

1.4

Extension, mm

Figure 17 Insert testing raw data

Results

The calculated failure loads of the panels can be seen in table 5. However significant processing of the raw data was necessary to reach this number. The figure above is an example of the raw data from the tests. We can see that there is an area of low gradient at the beginning of the test, this is referred to as the ‘toe’. The toe is a result of the slack being taken up in the testing rig, and in this case it is also a result of the test panels not being perfectly flat due to flexibility in the moulds. Toe compensation is applied to the results and this takes the linear part of the test data and re-zero’s the chart (ASTM International, 2006). This can be seen in figure 18.

34

In further examination of the data we can see that there is a region of linear deformation, after this stage the results start to curve, and we can see in figure 17 this occurs at around the 3,000 N mark for the test panel i4. At this point we begin to find that the panel is plastically deforming through shear buckling within the core. This deformation can be measured in the different between the raw data for the 4 kN test and the 6 kN test. The permanent deformation in the case of panel i4 is approximately 0.7 mm at 4 kN. However despite the failure of the panel, we can see that it continues to carry more load through to 6 kN without critical failure. The 6 kN tests were repeated for several panels and the panels were able to repeatedly support the 6 kN load. However at these large loads the deformation of the panels was quite large, over 3 mm in panel’s i1 and i2. For the monocoque, any permanent deformation of the shell would not be acceptable in the design as it compromises the structure of the monocoque despite not being a catastrophic failure. Therefore the point at which the panels are deemed to fail is when the deformation behaviour becomes non linear. The results of this are shown in table 5. The panels were tested in 3 phases, in the first phase of testing panels i1 to i4 were tested. From the results we were prompted to increase the stiffness of the panel to delay the onset of shear deformation, this was attempted in two ways by increasing the core thickness (i6), and by increasing the thickness of the face sheets (i5). The final two panel’s i7 and i8 used an epoxy and q-cell mix to fill the honeycomb cells, this configuration was tested due to savings in weight, and in manufacturing time. We can see from the results in table 5, and figure 18, that none of the panel configurations were able to reach the design load of 4,880 N without failing. The increase in panel stiffness in the second round of testing was able to improve the performance of the panels, but not sufficiently, and not to the degree that was expected. The reason for this is that the increased stiffness of the panels was calculated based on the bending rigidity of the panels where as in this case the deformation and failure were dominated by the shear characteristics of the core. Finally the third round of testing was conducted to determine the viability of using a syntactic epoxy filled insert, these performed poorly compared to aluminium, but were the easiest to manufacture. 35

Insert Testing - Load vs Extension w/ Toe Compensation 4000 3500 3000 i1 - 4kN return

Load, N

2500

i2 - 4kN return i3 - 4kN return

2000

i4 - 4kN return i5 - 4kN return

1500

i6 - 4kN return 1000

i7 - 4kN return i8 - 4kN return

500 0 0

0.2

0.4

0.6

0.8

1

Extension, mm

Figure 18 Insert testing results shown after toe compensation has been applied

Figure 19 Shear failure in the honeycomb core is evident in these test pieces

36

1.2

Analysis and Design Modification

The deformations in the panel are due to the shear failure of the core material as can be seen in figure 19. From the results we can see that the main limitation of the design is the shear strength of the core. In determining that the limiting factor we then designed to reduce the shear loads on the core. This was done in 2 ways, however other possibilities exist: •

Increase the stiffness of the panel, and hence delay the onset of the shear deformation, and hence shear failure. This was accomplished by either increase the core thickness, or by increasing the thickness of the face sheets. The method was somewhat effective as we can see in the test results.

•

Decrease the concentration of the shear load around the load point. This was done by increasing the insert size. This was the most effective method, however it is also adds a significant amount of weight. To reduce these disadvantages, we tried panels with the cells filled with epoxy + q-cell filler; however the modulus of the material was too low to be effective.

•

Another possibility which was not tested is to increase the shear strength of the core. This would have been accomplished by using a honeycomb with a thicker foil gauge. The limiting factor here is added complexity in design and manufacture as the new ‘reinforced’ honeycomb would need to be spliced into the ‘normal’ honeycomb. There were also issues in obtaining samples in the small amounts required.

However despite the attempts we can see that none of the panels were able to reach the design load point without failing. To solve this problem we turned to internal bracing. Initially it was thought that a composite stringer inside the chassis would be the solution, but this would be difficult to manufacture accurately and the loading on the stringers would not be optimal. Secondly the stringers would be susceptible to damage from the driver’s feet as he enters or leaves the vehicle. A horizontal steel tube was the final design for the bracing. This would fit between the lower pickup points and would be easy to install after the manufacture. It supports large loads 37

between the lower wishbones; this reduces the loads on the core and prevents shear failure of the core. The steel supporting member also has an advantage on a composite brace as it is not as susceptible to damage. It should be noted that the loads in the upper wishbones were much lower, and a 30 mm Aluminium insert in a 16 mm core (equivalent to sample i6) will be able to carry the 1785 N maximum load with a safety factor of 1.65. The maximum braking load case of 1.5 g is unlikely to be achieved and it is concluded that the design would be adequate. In the event of the forces being larger, the panel will not fail catastrophically until past 6 kN, protecting the drivers legs with a safety factor of 3.35. The load reduction which the brace would supply was calculated using FEA analysis. Shown below are the results of the FEA analysis of the braking load case with the bracing in place. By summing the forces, we can see that the load being carried by the shell is now 2400 N giving a safety factor of 1.23. However the FEA analysis does not accurately treat the shear deformation of a sandwich structure, and in fact the actual forces carried by the brace would be higher than calculated by the FEA as it does not account for shear deformation. The brace sufficiently stiff to greatly restrain the deformation, and as such, with a much lower deflection the shear failure of the core would not occur and the safety factor in the design would be much larger then calculated. A more accurate calculation could be produced by using HyperWorks and this is noted in the design evaluation.

Figure 20 Finite element analysis results of monocoque with bracing installed

38

4.4.

Part Geometry

The actually geometry of the chassis is influenced by a multitude of factors, the major ones are the F-SAE rules (in particular the template regulations, see Appendix E and F), suspension geometry, manufacturability and of course mechanical performance. We also cannot forget aesthetics which is possibly the most difficult thing to get ‘right’. An overview of the geometry creation process can be seen in figure 24. The process was iterative and many revisions were made before the final design was settled on. We began our design with a template; essentially the CAD of the 2009 car was stripped back leaving just the suspension geometry, cockpit templates and roll hoops. The cockpit templates are in place to ensure that there is adequate space around the driver’s legs, and also a sufficiently large cockpit opening for easy ingress and egress. The roll hoops importantly must cover a sufficient envelope so that drivers are protected in the event of a roll over(SAE International, 2008), the design is based on the existing geometry. A concept design based on past body work was used to influence the look of the car. With the template in place we could see that the area that would drive the design would be the driver’s foot well where the front suspension was attached to the car. This area was constrained by the location of the suspension components including the suspension wishbones and the steering rack and also the foot well template. We also attempted to make the areas where the suspension attached ‘flat’ to simplify the design. We can see that once we have decided on the design surrounding the foot well, we were able to take this and expand it to fit the rest of the car. Lines were chosen to give the car a clean appearance. We also attempted to make the car as compact as possible. Making the car compact meant that fewer materials would be used and a smaller cross section for the same sandwich structure there would mean a stiffer structure. The majority of the curved surfaces on the car are also single rather than compound curves, this makes it less likely that the accuracy of the geometry would be reduced in the hand finishing processes as more of the curve would be in contact with sand paper. 39

Figure 21 Early concept mock-up of design

Figure 22 Render of foot well area showing template location, cross brace and suspension configuration

40

An extension of the foot well design challenge was the packaging of the coil over units. In the 2009 car the coil over units are mounted in parallel with the vehicles direction of travel. It was found that it was not possible to package the shocks in this manner on the monocoque design without compromising both the structure of the tub, intruding into the foot well template area and maintaining the clean aesthetics of the car. It was decided that the shocks could be mounted perpendicular to the direction of travel and the front suspension package was completely redesign including rockers, shock mounts and front anti-roll bars. In figure 23 we can see that this method of packaging the shocks has reduced the cross sectional area of the monocoque, and also maintained smooth transitions between the surfaces. Once the major geometry was sorted, the roll hoops and wishbone bracings were created. This was a relatively straight forward procedure. The shock cover and nose cone were also created from the main shape as seen in part 5 of figure 24. As these were cut from the original geometry they should fit the final part with good accuracy. These are not structural components.

41

Figure 23 CAD process from template to finished part

42

4.5.

Beam Testing and FEA Validation

Before we could begin modelling our design in the finite element package we needed to first validate the performance of the FEA against a real world benchmark. The main reason for this is that different manufacturing techniques will lead to different fibre volume fractions in the composite material which can affect the stiffness of the final product. The final fin fibre volume fraction is difficult to predict and the material data from the data sheets were normalised to 55% fibre volume when cured in an autoclave process. We predicted that our fibre volume fraction would be less then this as we were using only a vacuum oven cure. To do this we first looked at testing standards, and aimed to develop an experiment around them. The two test standards that we will focus on are ASTM D 7250 (ASTM International, 2006) and EN 6061 (Aerospace and Defence Industries Assocoiation of Europe, 1995). 1995) These two standards detail the testing of flexural stiffness and strength respectively. The decision was made to test with the ASTM D 7250 standards. The decision for this was based on the fact that as part of another student thesis, strain gauges were in place on the beams and it was preferred that the beams were not destructively tested as in EN 6061. Test Procedure

ASTM D 7250 0 specifies a beam in 4 point bending. Based on the guidelines set out in the standard and the availability of a suitably sized test rig, it was decided that the test pieces would be 650 mm x 75 mm and as they were manufactured at the same time as the insert inse test pieces the layup consists of 5 plies of MTM56/HTA5131 in a [0, 45, 0, 45, 0] stacking sequence with a 10 mm ECM 6.4-80-10 10 3003 series aluminium honeycomb core. The key to this test was that we manufactured the test pieces using the same processes aass what the final part would be manufactured in. The manufacturing process is described in more detail in Section 5. 400mm

610mm Figure 24 Bending test rig configuration

43

Before beginning the mechanical testing the beams were modelled in Strand7 to give us a preliminary idea of the performance to expect. The beam was modelled to the dimensions mentioned earlier and restrained as per the test rig. The values for the plies and the core were drawn straight from the material data sheets as seen in the appendices. The value that was of the most interest was the deflection at the middle of the beam. The FEA model is shown in figure 26. The setup of composite analysis within Strand7 is discussed in more detail in Section 4.6 when the FEA analysis of the chassis is presented. The FEA results are presented along with experimental results in 27.

Figure 25 FEA test beam setup

The results from the FEA model showed that approximately 10mm of deflection could be expected at 1,200 N of load. As 10 mm of deflection was the maximum travel of our Linear Variable Differential Transformer (LVDT) it was decided that applying the load in 100N intervals would give a good range of samples. The FEA model also showed that we could expect the deflection to vary linearly with the load. The Instron test machine used in the test was slightly older than the one used in the insert test rigs and as such we were not able to log the loads. Instead the load was applied in 100 N steps with a pause between each increase in load. In this manner the logged values from the LVDT could be interpreted to see the deflection at each load level. The beams were each tested twice. 44

Figure 26 test beam in 4 point bending rig

Results

Beam 2 Test 1 - Raw Data

Deflection, mm

-2

1 20 39 58 77 96 115 134 153 172 191 210 229 248 267 286 305 324 343 362 381 400 419 438 457 476 495 514 533 552

0

-4 -6 -8 -10

Sample Number beam 2 test 1

Figure 27 Raw data from beam 2 test 1

45

4-point Bending Test Results 800 700

Load, N

600 500

Beam 1 Test 1

400

Beam 1 Test 2 Beam 2 Test 1

300

Beam 2 Test 2 Prelim FEA

200

Calibrated FEA 100 0 0.000

2.000

4.000

6.000

8.000

10.000

Deflection, mm

Figure 28 4 point bending results FEA and experimental

As seen in the results there was some variability in the data, one beam was slightly stiffer than the other. However the results from the test on each individual beams were very similar, this suggests that the reason was due differences within the beam rather than errors in the experimental procedure. However there was no measurable difference in the dimensions of the two beams, and they were cut to the same size and manufactured as one piece before being cut into separate parts. Ideally if resources allowed more samples would be tested. Analysis

The results show a linear response as predicted by the analytical calculations and preliminary FEA. From this we went on to calibrate the FEA model to fit the data. There are many variables in the FEA model which affect the mechanical performance of the beam in bending and we need to select and appropriate one to modify for an accurate model. These variables include; Laminate/Ply Thickness, Ply tensile Modulus, Core Shear Modulus and Core Thickness.

46

The mechanical properties which could be measured, that is ply/laminate thickness and core thicknesses were measured. This leaves modifications to be made to the modulus of the plies. The ply modulus is subject to change because with the manufacturing process of a honeycomb, the pressures applied to the part is much lower, and consolidation of the fibres in the part is lower, reducing the fibre volume fraction, and the modulus of the part. Also of note is that the FEA model does not take into account shear deformations, however in the case of the long beam, the deflection is dominated by bending, and the affect of shear deformation not accounted for. The next task now is to modify the material properties of our FEA to match the results with the experimental. It was necessary to reduce the modulus of the composite in the numerical model to 49.9 GPa, this is down from the datasheet value of 65 GPa. We can see that this is a large drop, and we will further analyse this to make sure our calculations are reasonable. To analyse this calibration further we obtained the individual material properties for the MTM56 epoxy and the HTA 5131 fibres and calculated the fibre volume fraction required to give us the modulus we determined experimentally. The equation for calculating the elastic modulus based on fibre volume fraction and the modulus of the individual components is:        1     Where,

Eb = Bulk modulus of the material, (GPa) f = fibre volume fraction Ef = 237 GPa (elastic modulus of the HTA 5131 fibres) Em = 3.6 GPa (elastic modulus of the MTM56 epoxy matrix)

Additionally in this case we are using a woven fabric, so the modulus would actually be half of that calculated in with this equation. The reason being half of the fibres are running in the perpendicular direction and do not contribute to the stiffness. The results are shown in the following figure.

47

Bulk Modulus vs Fibre Volume Fraction 160.0

Elastic Modulus, GPa

140.0 120.0 100.0 80.0 66.0

60.0 49.9

40.0 20.0 0.0 0

0.2

0.4

0.6

0.8

1

1.2

Fibre Volume Fraction Unidirectional

2x2 Woven

Figure 29 Bulk modulus vs fibre volume fraction

From the figure above at a fibre volume fraction of 0.55 the modulus is 66 GPa, very close to the 65 GPa from the data sheets. At the calibrated modulus of 49.9 MPa, the fibre volume fraction is 0.41. On inspection of the data sheet for the MTM56 pre-pregs the fibre volume fraction of a press cured unidirectional pre-preg only achieved a fibre volume fraction of 0.53. This leads to the conclusion that the 0.55 fibre volume fraction to which the values are normalised in the supplied data sheet are rather optimistic. The combination of the vacuum oven cure, woven material and the honeycomb means that it is difficult to obtain a high fibre volume fraction. The results of the validation process were deemed to be sound and the development of the full car FEA model could begin.

4.6.

Development of FEA Model

The finite element model of the chassis is built up of quad8 plate elements with composite properties assigned within Strand7. The specific design of roll hoops, suspension components, internal bracing and the rear of the car were not within the scope of the design, these were created as beam elements as they offered an accurate stiffness while reducing the overall complexity of the model. 48

The transition of the part model to the FEA software package occurred smoothly. From the part geometry a single part was created at the mid-surface of the sandwich structure. The IGES file format was used to transfer the geometry between the CAD and FEA packages. The suspension components and rear of the car were created manually within the FEA package based on dimensions extracted from the CAD model. The material properties and beam geometries were carried over from the space frame model. Importantly the specific modelling issues that are faced when analysing composite materials and sandwich structures are addressed. The composite sandwich structure meant that additional variables were defined such as: •

Additional material properties, elastic modulus in the 11 and 22 direction. As well as multiple failure criteria including stress and strain limits in both directions for tension, compression and shear.

•

The local axis directions of the plate elements used need to be aligned correctly so that the fibre orientations are accurately defined on the model.

The meshing process was perhaps the most time consuming because of these extra requirements, in particular ensuring the all the elements were properly orientated and aligned. The automatic meshing within Strand7 improved the speed of the process however minor glitches were found and these needed to be inspected. After the initial meshing it was found that many elements were inside-out, and the command to align normals was used to correct this. The drape feature was then used to align the local x-axis of the elements to the cars direction of travel. This would ensure that fibre orientations would be applied correctly.

Figure 30 Plates showing their local axis, it was important that these were all aligned in the same direction

49

Figure 31 CAD model and FEA model

50

As part of the modelling process it was found that the quad4 elements used in the first iteration of the model were not suitable because the complex geometry of the part meant that some of the elements had were skewed and had aspect ratios greater than 1 making the results from them unreliable (Strand7, 2008). The solution for this was to start from scratch and mesh the model with quad8 elements. These elements are tolerant of bad aspect ratios. Once the mesh was in place the laminate configuration was defined. Material properties from the previous analysis were carried over. In particular this included the properties for the woven fabric that were validated in the previous section. The laminate configuration was also carried over from the previous section. The quasi isotropic [0, 45, 0, 45, 0] stacking sequence was chosen as detailed load information was not available so optimisation of the design could lead to weaknesses in the design if any unaccounted loads are present.

Figure 32 Laminate configuration

Connections between the chassis and the other components were then defined. Minor components such as the suspension mounts were assumed to be rigid and rigid links were used to connect the monocoque to the surrounding structure including the roll hoops and lower wishbone support braces. The roll hoops were each attached to the chassis at 6 equally spread out locations. End releases were applied to the ends of the wishbones and pushrods to allow the appropriate range of movement. The coil-over units were assumed to be solid. 51

Restraints were created in three different freedom cases for acceleration, braking and torsion test. The model is restrained at the tyre contact patches, only one corner has its motion full restrained, this is the front right or rear depending on load case. The remaining corners are all restrained from vertical movement but allow lateral or longitudinal movement depending on their location, this is to allow the car forces to be reacted primarily at the rear of the car in acceleration, or the front in braking. Finally the loads were applied. These are carried over from the load analysis of the space frame chassis, which were based on the predicted performance envelope of the car. The loads were applied as linear accelerations and the magnitudes are shown in the table below. Masses were in place at nodes to simulate the mass of major components in the vehicle so that the accelerations would produce the correct loadings on the monocoque. Table 6 Load and freedom cases

Load Case

Freedom Case

Vertical Longitudinal Lateral Acceleration, ms-2 Acceleration, ms-2 Acceleration, ms-2

Static

Acceleration

-9.81

0

0

Acceleration

Acceleration

-9.81

14.72

0

Acceleration + Cornering

Acceleration

-9.81

14.72

14.72

Cornering

Acceleration

-9.81

0

14.72

Cornering

Braking

-9.81

0

14.72

Braking + Cornering

Braking

-9.81

-14.72

14.72

Braking

Braking

-9.81

-14.72

0

Bump

Acceleration

-29.43

0

0

Table 7 FEA model statistics

Nodes

71,922

Links

52

Beams

151

Faces

99

Plates

23,815 52

4.7.

Evaluation of FEA Results

Once the FEA model was successfully run we then proceeded to post processing where the results were evaluated. In this section we look at how results were interpreted to determine safety factors and also to gain an understanding of how the design is behaving under various load conditions. The tools available for this analysis include not only stress, strain, deflection and composite safety factors, but also curvature analysis which also helps interpret the way the chassis acts under load.

Figure 33 Plate deflection and axial forces for braking load case, maximum deflection 0.5mm

Laminate failure analysis was done using the built in maximum stress, maximum strain and modified Tsai-Wu criteria built into Strand7. As a conservative measure the safety factor which will be designed to is the smallest of the three failure criteria’s. Ideally destructive testing of samples would be performed and this would enable us to more accurately predict failure, however in absence of this a conservative approach will be used to guarantee the performance. The following table outlines the minimum safety factors in the design for each load case.

53

In the braking load case which has the largest loads, the maximum deflection of the pickup points is 0.5 mm. However this is partly due to the compliance in the suspension system, and the actual deflection as calculated as half the change in distance between two opposing suspension pickup points is 0.2 mm. The insert mechanical test results (figure 18) also show a deflection of than 0.2 mm of deflection for the load case (1,785 N). Table 8 Minimum safety factors in each load case

Load Case

Maximum Stress

Maximum Strain

Modified Tsai-Wu

Minimum S.F.

Static

43.01

35.48

42.84

35.48

Acceleration

186.01

153.72

186.05

153.72

Acceleration + Cornering

27.43

20.88

26.60

20.88

Cornering

17.53

14.19

16.90

14.19

Braking + Cornering

21.39

17.59

21.09

17.59

Braking

36.17

24.42

31.52

24.42

Bump

14.33

11.82

14.28

11.82

The safety factors show that the design is quite conservative when it comes to strength. This is to be expected as the design at this stage is stiffness driven. However the results must also be treated with caution, as when the locations of these minimum safety factors are seen to be around the inserts. The values in the immediate vicinity of the inserts would need to be viewed with caution as we know that the FEA model is not very accurate in the areas dominated by shear deformation. This has been accounted for in the mechanical testing of the inserts. Despite this we can see that overall the ply stresses over much of the monocoque are quite low leading to extremely large safety factors. This would suggest that there is a lot of optimisation possible in these areas to reduce weight, material usage and save money. A look at the strain levels shows much the same thing and these areas could be locations where it may be possible 54

to reduce the ply layup without compromising structural stiffness. The figure below shows the safety factors calculated using the Modified Tsai Wu method, and the non-colour areas are where the safety factor exceeds 1000. We can also see that the safety factors are the lowest surrounding where loads are introduced. One figure that stands out in particular is the large safety factors in the acceleration case. This is attributed to most of the load being carried in the rear of chassis. Finally of note is that the minimum safety factor calculated for all cases is via the maximum strain criterion.

Figure 34 FEA results, braking load case showing safety factors

The analysis of the loads also showed that the front bulkhead was not subject to significant loads. Based on this, it was redesigned as a monolithic structure with no core reinforcements. This proved to have no significant impact on the performance of the final design, and would make the manufacturing process simpler. The torsional rigidity of the chassis was calculated to check that the design goals were met. The model was restrained at the rear roll hoop and a rigid link placed between the front suspension 55

points. The rigid link has a pivot in the centre, simulating the torsion test rig. The torsion test rig setup is shown in the following figure along with the results. Table 9 Torsional stiffness analysis of monocoque Pickup Location

Node

Deflection, m

Distance from Centreline, m

Deflection in Degrees

Distance from roll hoops, m

Torsional Stiffness, N.m/deg/m

L:U:F

40

1.94E-04

2.41E-01

4.61E-02

1.055

22,875

L:U:R

37

1.35E-04

2.41E-01

3.20E-02

0.755

23,560

L:L:F

35

2.38E-04

2.64E-01

5.18E-02

1.055

20,386

L:L:R

33

1.86E-04

2.64E-01

4.03E-02

0.755

18,723

R:U:F

39

1.94E-04

2.41E-01

4.62E-02

1.055

22,859

R:U:R

38

1.31E-04

2.41E-01

3.11E-02

0.755

24,260

R:L:F

36

2.38E-04

2.64E-01

5.17E-02

1.055

20,397

R:L:R

34

1.86E-04

2.64E-01

4.03E-02

0.755

18,725

Location Notation:

Minimum

18,723

(Left/Right) : (Upper/Lower) : (Front/Rear)

Average

21,473

The torsional stiffness is measured as the average of the rotational deflection of the suspension mounting points per meter of the chassis. The applied moment was 1000 Nm and this was chosen to transmit similar loads into the chassis as the ‘dynamic’ load cases from table 8. It was found that the torsional stiffness of the monocoque was 21,473 N.m/deg/m, this is approximately 4 times stiffer then the steel space frame chassis. It should be noted that the space frame is analysed without stressed skins as there was no way to validate the effect of the stressed skins. Correctly applied they can dramatically increase the stiffness of the space frame, however if over stressed or loose as has been observed in previous cars then the effectiveness would be greatly reduced.

56

Figure 35 Torsion stiffness model with plate elements hidden to reveal the rigid links and moment application

4.8.

Sandwich Failure Modes

There are 3 main modes of failure for a sandwich structure. Yielding or wrinkling of face sheets, and core failure. Delamination of the face sheets from the core is not considered in this analysis as it was deemed not to be an issue based on the performance of the beam test pieces. Yielding of the face sheets occurs when the stress in the face sheets exceeds the strength of the face sheets; this is addressed previous section with the max stress safety factor analysis. The core failure is due to the shear strength of the core being exceeded and we address this in the insert design section. The wrinkling of face sheets occurs when the compression stresses in the beam reach a level of instability (Allen, 1969). In the beam testing two sandwich beams were tested nondestructively. Using FEA simulation the stress in the compressive face of the beams was determined at the maximum deflection. This stress level was 74.9 MPa and this was treated as a stress level below that which would cause instability in the face sheet.. From examination of the FEA results the highest compression stress in all load cases was 52.82 MPa in the bump load case. Therefore the minimum safety factor in face wrinkling is 1.42. This is a conservative estimate. 57

5. Manufacturing The test pieces and final components were manufactured under the expertise of the staff at LSM advanced composites. Callum Jensen organised the logistics of the manufacture of the final component. The manufacturing process was lengthy and still in process as there were some changes to the manufacturing process chosen due to unforseen financial difficulties. In this section we will not cover the basics in composites manufacturing but rather the unique challenges in manufacturing the monocoque.

5.1.

Moulds

The majority of composite components are manufactured in a mould. The mould is of critical importance and a good mould will guarantee an accurate reproduction of the final product with a good surface finish. The mould process in this case was a 2 step process, a male master of the mould was produced first, and this was identical in shape to the final product. From the master a female mould was taken, and within this female mould is where the final part is constructed. For simple parts such as the insert test panels and beams the moulds were merely a sheet of aluminium, or a plate of glass. However for the monocoque with its complex shape the production of the mould was a challenging task. The mould was produced in 2 parts split vertically down the primary axis of the car so that the component could be produced as one part. The first step in the production was to machine a block of EPS foam 6 mm under the size of the finished part. CAM software was used to generate tooling paths for a robot CNC process from the original part geometry. On those EPS core 2 layers of fibre glass were applied in a wet layup. This would give the mould strength. Once the fibre glass was cured a layer of car bog was applied over the top and the mould was built up over size ready to be machined to its final size. Car bog was chosen as it was less than half the price of specialist tooling paste and was tough enough for the application as it was only a single use item. Once the car bog was cured, the CNC robot was used to machine the moulds to the finished size. This stage of the process is necessary to ensure the accuracy of the final component. Minor inaccuracies can be compounded through the manufacturing process and maintain accuracy is paramount throughout the whole process. 58

Figure 36 The EPS core of male master moulds

Figure 37 The male mould after bog is applied and machined to final size

The bog is difficult to apply without trapping air, and the next process was to fill any holes with bog, and hand sand the mould to a smooth finish. Once a satisfactory finish has been achieved, the mould is then sprayed with a high quality automotive paint, and then polished extensively to a smooth finish. The paint provides a high quality surface from which the female mould can be produced. 59

Figure 38 The finished male mould for the shock cover, note the highly polished surface.

The male mould is then polished with mould release products in preparation for the female mould. The first step in producing the female mould is to spray a layer of gel coat onto the male mould. This is a critical step as this will form the mould surface of the female mould. A chopped fibreglass paste is then layered over the gel coat, this forms the main structure of the female mould and it is important that the paste is thick enough to form a stiff structure. This female mould is cured at above room temperature in an oven.

Figure 39 The chopped paste female mould formed over the male shock cover mould

60

5.2.

Lay Up

Once the moulds are complete the next step is to begin laying up the pre-preg material within the mould. However the first step is to ensure that the mould is well polished with a mould release agent so that the part can be separated once cured. Once the mould is prepared, the layers of carbon fibre cloth are laid into the mould. The sheets of carbon fibre would ideally be cut to size before hand with a CNC cutter, but as the ability to create templates was not available in the software package used a more traditional method was used. Paper was used to get a rough idea of the size and shape of each individual layer, and this was cut out of the carbon fibre sheet. Special care should be taken here to ensure that the fibre orientation is correct. Once in the mould the pre-preg sheets were worked carefully onto the mould surface taking care not to bridge corners and leave air gaps within the structure. The excess cloth was trimmed with a sharp knife. The layers are placed as required as well as any core reinforcements. At this stage with heavier layups, for example greater then 3-4 plies, the part is then bagged placed under vacuum for a period to consolidate the layers in a process known as debulking. The bagging process is also an important process and it consists of several layers. The first layer over the pre-preg material is the peel ply. This ensures that the part does not stick to the breather; the breather is the next layer and ensures that the vacuum pressure is spread evenly over the part. Finally all of this is placed into a vacuum bag which has valves sealed into the sides to allow the air to be removed from the bag. It is important that excess peel ply, breather cloth and vacuum bag are available so that the materials do not bridge any corners. This would result in reduced pressure in the area reducing the quality of the finished part. In the following figures the shock cover is shown laid up within the female mould. Once laid up and bagged the part is then placed through the curing cycle. Once through the cure cycle the part is removed from the mould and trimmed to size. The finished carbon fibre is easily sanded and any excess material is quickly removed. Finishes such as paint or clear coats can then be applied to the surface for protection. Other secondary structures can be attached at this stage either with adhesives or through mechanical fasteners. 61

Figure 40 The shock cover laid up with the pre-preg material and core reinforcements

Figure 41 The shock cover vacuum bagged and ready for cure

Special thanks must go to LSM Advanced Composites at this stage as they were instrumental in making the manufacturing process occur. From mould building and composite layup techniques to the specification of cure cycles, the staff at LSM Advanced Composites were available with knowledge and advice to ensure that the manufacture of the parts went smoothly.

62

6. Testing Once complete the monocoque would undergo static mechanical testing to validate its performance against design specifications, this would help ensure that it is safe to use in competition. The primary mechanism for testing would be the torsion test rig. The torsion test rig is shown in the figure below; it consists of a fixed framework and moveable frame mounted on a large bearing. This allows a torsion load to be applied to the chassis using weights and the deflection to be measured. The deflection can be measured with dial gauges and other measurement devices.

Figure 42 Chassis torsion test rig (Bateman, 2005)

If the test is successful in validating the design then dynamic track testing would commence. If the test was not able to validate the design then further testing would need to be conducted to investigate why the part deviates from the design specifications and whether or not the chassis is safe to use in competition.

63

7. Overview of Final Design The main component from this design is the monocoque chassis. The structural roll hoops are bolted to the chassis in 6 locations each with pairs of M8 bolts. Also included in this design are the nose cone and shock/coilover cover, these are non-structural components.

Figure 43 Final design

In the figure above the monocoque is made transparent and we can see the lower suspension braces. The mounting tabs for the rear roll hoops are shown, however the tabs for the front roll hoop are not shown. 64

7.1.

Specifications and Features

The major specifications of the composite monocoque and the chromoly space frame are compared in the table below. Table 10 Comparison of composite monocoque and cromoly space frame

Composite Monocoque

Chromoly Space Frame

Length, mm

1,510.0

1,479.5

Height, mm

1,145.4

1,145.4

Width, mm

711.5

825.4

Weight, kg

18.35

22.97

Stiffness, N.m/deg/m

21,473

6,821

The composite monocoque and the space frame share most leading dimensions however the monocoque is lighter and stiffer. A mass breakdown of the monocoque is shown in the following table. Table 11 Mass breakdown of composite monocoque

Carbon Fibre

4.80 kg

Aluminium Honeycomb

3.09 kg

Aluminium Inserts

2.02 kg

Roll Hoops

5.46 kg

Lower Suspension Braces

2.04 kg

Syntactic Epoxy

0.96 kg

Total

18.35 kg

A unique feature of the monocoque design is the lack of a rear bulkhead, from a design view point the roll hoop is over designed in the competition rules so instead of having it as ‘dead’ weight it acts as a brace for the rear of the structure. The front roll hoop acts in a similar way to brace the middle of the structure. This is only a feature in the context of the F-SAE rule set, due to the strict rules covering the specifications of the roll hoops.

65

7.2.

Schematic Drawings

66

67

8. Design Evaluation Throughout the design process there were many times where possibly improvements were identified but where not implemented mainly to reduce the complexity of the project or due to lack of resources. These occurrences are identified in this section and provide a suggested way forward for future evolutions of the design. The single piece of information which would have provided the most benefit to the design process is load data. This would be data from accelerometers, linear pots on the shocks and other data acquisition devices placed on an existing car as it is put through dynamic testing. This data would be used to give an accurate picture of the loads acting on the chassis, and allow the design to be more accurate and efficient. From the design view point, more test samples in the mechanical tests performed would give better results, and allow for more refinement of the designs. The insert designs could be much further optimised. A monolithic composite block would have similar modulus to the aluminium while being up to 50% lighter. There is also the possibility of removing material from the aluminium insert to produce a lighter hollow insert component. More test pieces would also have been useful in the beam test as there was a slight discrepancy between the performances of the two beams which could not be explained. The treatment of the shear performance of the core in FEA analysis is a major point for improvement. The shear failure of the core is most likely where the first failure would occur in the design, and it is also failure mode that is least able to be predicted using the FEA and analytical tools available. Whether through the adoption of different software, more detailed modelling techniques, or more extensive testing a better understanding of the performance of the honeycomb core would greatly improve the design. Finally optimisation of the design through ply drops, different core materials in different areas, the use of unidirectional fibres would all greatly improve the design. This would be possible through the use of better software.

68

9. Conclusion The goal of the project to design a composite monocoque chassis was achieved. However due to the impact of the global financial crisis the manufacture was delayed and was not complete before the hand in of this thesis. The manufacturing process is well on its way, and due to be completed within the next month. In the absence of being able to physically test the component, a preliminary evaluation of the design process shows that the important design decisions were all based on mechanical test data and there is confidence that the finished monocoque will perform as designed. A combination of mechanical testing and numerical analysis was used to develop the design. Mechanical testing to design small areas of high stress concentration where it was felt that the FEA software would not be able to accurately predict results without correlating with physical results. An FEA package calibrated by a 4 point beam bending tests was used to analyse the design as a whole. A mechanical trial and error method of designing such a large part would not be a viable option. It was felt that with this combination of the advantages of both approaches were able to compensate for their respective shortcomings. In particular the trial and error aspect of the mechanical testing was useful in developing an understanding of composite materials. Having an actual component in your hand that has been deformed is much more useful than any piece of software. And the available literatures all state that physical testing is a very important part of the design of composite materials. Through the project a better understanding of the behaviour of composite materials was developed and with the new software such as HyperWorks with advanced optimisation tools becoming available an understanding of the basics becomes more important, as it allows the designer to understand and efficiently use more powerful engineering tools.

69

References Aerospace and Defence Industries Assocoiation of Europe. (1995, November 1). AECMA PRED6061 Aerospace Series Fibre Reinfoced Plastics Test Method Determination of Sandwich Flexural Strength 4 Point Bending. Allen, H. G. (1969). Analysis and Design of Structural Sandwich Panels. London: Pergamon Press. Altair Engineering. (2009). About HyperWorks. Retrieved September 13, 2009, from HyperWorks - A Platform for Innovation: http://www.altairhyperworks.com/ ASTM International. (2006). ASTM D7250 / D7250M - 06 Standard Practice for Determining Sandwich Beam Flexural and Shear Stiffness. ASTM International. Barbero, E. J. (2008). Finite Element Anaylsis of Composite Materials. Boca Raton: CRC Press. Bateman, L. M. (2005). Investigation into a Monocoque Chassis for Formula SAE. St. Lucia: The University of Queensland. Bitzer, T. (1997). Honeycomb Technology: Materials, design, manufacturing, applications and testing. London: Chapman & Hall. Bunyawanichakul, P., Castanie, B., & Barrau, J.-J. (2008). Non-linear finite element analysis of inserts in composite sandwich structures. Composites: Part B , 1077-1092. F1 Complete. (2008). Hard Core - The Formula 1 Monocoque. Retrieved March 13, 2009, from F1 Complete: http://www.f1complete.com/content/view/1843/389/ Gay, D., & Hoa, S. V. (2007). Composite Materials: Design and Applications (2nd Edition ed.). Boca Raton: CRC Press. Gray, T., & Aldous, M. (2009, October 16). Hyperworks Presentation Session. (B. Lii, Interviewer) He, M., & Hu, W. (2008). A study on the composite honeycomb sandwich panel structure. Materials and Design 29 , 709-713. Heimbs, S., & Pein, M. (2009). Failure behaviour of honeycomb sandwich corner joints and inserts. Composite Structures 89 , 575-588. Hexcel. (2005, December). HexWeb Honeycombs Selector Guide. Retrieved June 26, 2009, from Hexcel.com: http://www.hexcel.com/NR/rdonlyres/7F70671B-ED6E-4562-9659ABA426A7453F/0/RevisedHexWebSelectorGuide.pdf LSM. (2009, March 25). Manufacturing Discussion. (B. Lii, Interviewer) Macogans Street Rods. (n.d.). chassis_32. Retrieved October 15, 2009, from Macogans Street Rods: http://www.macogans.com/32_chassis.jpg 70

Marsh, G. (2006). Composites conquer with carbon supercars. Reinforced Plastics , 20-24. Naik, N. K., Rao, G. N., Agarwar, U., Raju, K. A., Pottigar, S. A., & Suresh, V. (2009). Sandwich Structures with Composite Inserts: Experimental Studies. Polymer Composites , 639-648. Pilling, J. (2005). Design and Properties of Sandwich Core Structures. Retrieved April 12, 2009, from Composites Material Design: http://www.mse.mtu.edu/~drjohn/my4150/sandwich/sp1.html Qiao, P., & Wang, J. (2005). Transverse Shear Stiffness of Composite Honeycomb Cores and Efficiency of Material. Mechanics of Advanced Materials and Structures , 159-172. SAE International. (2008). 2009 SAE Rules. Retrieved May 28, 2009, from Formula SAE-A: http://www.saea.com.au/wp-content/uploads/2009/02/2009-fsae-rules1.pdf Savage, G. (2007). Failure prevention in bonded joints on primary load bearing structures. Engineering Failure Analysis 14 , 321-348. Savage, G. (2009). Formula 1 Composites Engineering. Engineering Failure Analysis , 2-28. Society of Automotive Engineers Australasia. (2009). Formula SAE-A. Retrieved May 28, 2009, from http://www.saea.com.au/formula-sae-a Song, K.-I., Choi, J.-Y., Kweon, J.-H., Choi, J.-H., & Kim, K.-S. (2008). An experimental study of the insert joint strength of composite sandwich structures. Composite Structures 86 , 107-113. Strand7. (2008). Strand7 Help Files. Retrieved June 11, 2009, from Strand7 Finite Element Analysis FEA Software: http://www.strand7.com/ Thomsen, O. T. (1998). Sandwich plates with 'through-the-thickness' and 'fully potted' inserts: evaluation of differences in structural performance. Composite Structures , Vol. 40, 159-174. University of Cambridge. (2002, February 25). Introduction to Material Selection Charts. Retrieved October 25, 2009, from Material Selection and Processing: http://wwwmaterials.eng.cam.ac.uk/mpsite/physics/introduction/default.html Walker, M., & Smith, R. (2002). A computational methodology to select the best material combinations and optimally design composite sandwich panels for minimum cost. Computers and Structures , 14571460. Wikipedia. (2009). Monocoque. Retrieved October 24, 2009, from Wikipedia: http://en.wikipedia.org/wiki/Monocoque

71

Appendicies Appendix A. Initial design schedule

72

Appendix B. Torsional Stiffness Analysis of Chromoly Space Frame Chassis Pickup

Node

Deflection, m

Location

Distance from

Deflection in

Distance from

Torsional Stiffness,

Centreline, m

Degrees

roll hoops, m

N.m/deg/m

L:U:F

40

6.04E-04

2.41E-01

1.44E-01

1.055

7349

L:U:R

37

4.45E-04

2.41E-01

1.06E-01

0.755

7139

L:L:F

35

7.00E-04

2.64E-01

1.52E-01

1.055

6938

L:L:R

33

5.84E-04

2.64E-01

1.27E-01

0.755

5957

R:U:F

39

6.10E-04

2.41E-01

1.45E-01

1.055

7278

R:U:R

38

4.44E-04

2.41E-01

1.05E-01

0.755

7161

R:L:F

36

7.10E-04

2.64E-01

1.54E-01

1.055

6839

R:L:R

34

5.89E-04

2.64E-01

1.28E-01

0.755

5907

Location Notation:

Minimum

5907

(Left/Right) : (Upper/Lower) : (Front/Rear)

Average

6821

73

Appendix C. Material Datasheet MTM56/HTA5131

74

Appendix D. Material Datasheet EC Aluminium Honeycomb Core

75

Appendix E. Cockpit Opening Template

76

Appendix F. Foot Well Template

77