Final Report Summary - M-RECT (Multiscale reinforcement of semi-crystalline thermoplastic sheets and honeycombs)
Executive Summary:
Executive Summary
The broad aims of the M-RECT project were several fold:
• Functionalisation of carbon nanotubes (CNTs) to aid the dispersion of the CNTs in the polymer matrices, which were PPS and PEEK.
• Manufacturing of compounds where the CNTs were dispersed and disentangled within the polymer matrix.
• Incorporation of the CNT loaded matrices into long-fibre composite materials.
• The development of computer based models to allow for the simulation of the materials and evaluation of the components produced as part of the project. The intention was to provide designers with the tools to be able to accurately model components.
• Development of tools for the risk analysis of the project and the materials being developed during the project.
• Manufacture of validation components to test the technologies developed and comparison with the output from the computer models being developed. These included automotive components, satellite applications, inflatable structures and jet engine inlet fan blades.
• The project sought to develop surface coatings to improve the UV resistance of the polymer matrices and to control heating effects due to solar radiation.
The project involved the development of a novel technique for damage detection within composite structures.
The project was therefore, very wide ranging and was aimed at covering and developing the complete supply chain from CNT functionalisation through to finished component design, manufacture, testing and use. Where possible the project used industry-standard approaches to operations and also innovative alternative approaches.
Technologies for the successful functionalisation of CNTs were developed which were found to aid the dispersion of the CNTs and scalable chemical processes for manufacture of the functionalised CNTs were developed. The development of the chemical functionalisation has resulted in one patent application being filed. The CNTs were incorporated into the matrix polymers although the anticipated improvement in simple mechanical properties was not observed. This was thought to be related to the orientation of the CNTs as the dispersion appeared to be good.
Composite laminates were manufactured from the materials developed in the project both via standard approaches and novel impregnation and bagging technologies. The development of a novel impregnation technology has resulted in one patent application being filed. Problems were experienced in the manufacture of the PPS laminates due to the formation of cracks within the matrix of the laminates when CNTs were incorporated.
The mechanical and damage tolerance properties of these materials were investigated. The results indicated that whilst the tensile properties of the CNT-containing laminates exhibit inferior properties to those without CNTs, the fracture toughness of the PEEK/CNT laminates showed promising results.
The computer models developed showed good correspondence with the performance of the validation components and the development of one of the modelling approaches is continuing within a follow-on project.
A range of novel morphing honeycomb systems have been developed based on a SILICOMB geometry. Whilst the addition of CNTs to these honeycombs did not produce the anticipated improvement in mechanical properties, the recovery of the honeycombs following compression was superior. A further follow-on project has been initiated around the novel honeycomb structures developed as part of the project.
Durable UV and heat resistant coatings were successfully produced for both types of composite matrix.
The tools relating to the evaluation of risk were fully developed and have been deployed within the consortium.
A technique for the assessment of damage within a composite structure was developed and shown to be robust enough for moulding into a thermoplastic composite material, this being vital for use in the health monitoring in large composite structures.
Project Context and Objectives:
The M-RECT project aimed to create multiscale – reinforced semi-crystalline thermoplastics, which would outperform all reinforced polymers in terms of strength, stiffness and damping, by developing and upscaling state of the art production procedures, within cost-efficient manufacturing routes.
The use of composite materials in engineered components has increased significantly within recent years due to the need to the reduce weight and the environmental impact of artefacts, particularly in the case of transportation. The desire for more fuel efficient and environmentally friendly aircraft and the regulations concerning emissions from road vehicles is driving both the aerospace and automotive industries to adopt composites as materials of construction. Examples of this trend are:
• The BWM i3. In order to maximise the range of electric vehicles, the body weight must be reduced and so composites are being used.
• Boeing 787 where approximately 35T of composite are used per aircraft. The use of composites results in significant weight savings for the aircraft which in then yields better fuel efficiency this giving longer range and better environmental impact.
The majority of composite materials are currently based on thermosetting matrices but thermoplastic matrices offer many advantages. The potential advantages include recyclability, faster and more energy efficient production methods, improved fatigue resistance and better mechanical performance, in particular impact performance.
It was in this context that the project was developed in order to seek to:
• Improve the performance of the composite systems by the inclusion of carbon nanotubes (CNTs)
• Provide scalable manufacturing technologies for the raw materials and products made from these multi-scale reinforced composites
• Fully characterise the performance of the materials
• Develop modelling software to aid the design of components made from the multi-scale materials
• Develop UV and heat resistant coatings to extend the performance of the materials
• Develop health monitoring systems for use in multi-scale composite structures
• Develop novel honeycomb systems
It was anticipated that multiscale reinforcement of PEEK or PPS in aircraft manufacturing would greatly enhance composites durability for long-term utilisation at intermediate temperatures (100-150°C). On top of the strength and damping enhancement, the crack-bridging capability of the CNTs was expected to result in a reduction in micro-cracking under thermal-cycling ageing or isotropic ageing with oxidation.
The semi-crystalline thermoplastics selected for the project were Poly-Ether-Ether-Ketone (PEEK) and Poly-Phenylene-Sulphide (PPS). The matrix materials selected were very-high performance thermoplastic resins which are used extensively in the aerospace and allied industries due to their properties. The products derived from these materials are also high value.
However, whilst these matrix materials were specified, the proposed project sought, primarily, to develop generic technologies for the production, modelling and design of high-performance, light-weight reinforced thermoplastics. The multiscale reinforcement envisaged, would comprise of disentangled, dispersed, straightened and aligned multi-walled CNTs and also fully impregnated long carbon fibres (CFs).
It was envisaged that, depending on the extent of reinforcement and manufacturing process, four products would emerge with a PEEK or PPS matrix:
i. Polymer sheets reinforced with CNTs.
ii. Laminates reinforced with CNTs and long fibres (multiscale reinforcement).
iii. Honeycomb cores with bonded panels as skins (multiscale reinforcement).
iv. Laminated parts with complex shapes (multiscale reinforcement).
The effective utilization of CNTs in a polymer matrix depended strongly on the development of processing technologies that would ensure their disentanglement, dispersion, straightening, alignment and good interfacial bonding with the thermoplastic matrices used.
The risks associated with failure of any of the newly developed technologies would be mitigated by always developing processes in parallel with the use of the existing state-of-the-art technologies in order to ensure delivery of the objectives.
The main objectives of the proposed project are the following:
1. To develop CNT-reinforced sheets from semi-crystalline thermoplastic PEEK and PPS at low CNT concentrations in order to obtain a nanocomposite material showing improved performance over the matrix material. The primary objective of this work was to functionalise the CNTs by directly synthesizing polymer-specific molecular chains on their surface. This functionalisation was intended to promote de-aggregation of the CNT bundles with minimal destructive effects on their structure. Their functionalisation with polymer specific organic groups would promote bonding with the matrix. As part of the project, the chemical procedures being used would be modified to increase the rate of the functionalisation and also to simplify it.
2. To develop novel technologies for the disentanglement, dispersion and straightening of the CNTs.
3. To investigate the use of microwave heating to reduce heating times and improve the energy efficiency of the processing of the CNT containing materials. Microwave heating of the CNTs and eventually of the polymer should be a very efficient process since the material is heated rapidly at the molecular level and therefore volumetrically. It reduces material degradation (surface micro-cracks), while the targeted localized melting along the CNT-polymer interface should minimise heat losses and improve the control of the melting and bonding processes.
4. To use the envisaged CNT-PEEK and CNT-PPS nanocomposites as matrices for continuous carbon fibre reinforcement, achieving high levels of impregnation and fibre distribution.
5. To use multiscale-reinforced PEEK sheets to manufacture strong honeycomb cores and panels based on the Directionally Reinforced Integrated Single-yarn (DIRIS) architecture.
6. To develop novel morphing honeycomb structures using CNT reinforced films.
7. To improve the quality and further develop a novel dry powder impregnation technique in order to impregnate performs for laminates and honeycombs.
8. To integrate the multiscale reinforced polymers for the production of representative aerospace and automotive components by developing smart out-of-autoclave manufacturing processes.
9. To develop on-line monitoring tools for rapid measurement of selected electrical properties and almost real-time quality assessment of the outcome of the production line.
10. To verify and measure the mechanical properties of the multiscale reinforced materials and to assess the quality of the consecutive stages of manufacture through testing at nano-scale and macro-scale at a wide range of testing conditions covering the full spectrum of anticipated operating environments.
11. To determine the damage tolerance of the materials developed.
12. To develop suitable numerical tools for modelling and designing structures based on Finite Element Analysis techniques and validate the developed models and software by comparing them with reference to the mechanical test results.
13. Using the numerical tools, simulate and validate any damping mechanisms observed.
14. To demonstrate the feasibility of readily incorporating the new products in large European industrial sectors such as aerospace, automotive, building and space structures. This will include the design, manufacture and testing of real-scale prototypes, which will be characteristic of the specific applications.
15. To ensure through optimized design that the cost benefit achieved through use of the high-performance CNT and multiscale-reinforced thermoplastics developed is effectively transferred into the envisaged structures.
16. To incorporate low-cost smart health monitoring optical fibre sensors in this new-generation of structures for continuous assessment of progressing structural damage.
17. To greatly enhance the strength of the materials to chemical attack, weathering and UV rays though the use of improved novel fluorocarbon-based coatings. The fluoropolymer based coating will be bonded, through a suitable primer, onto the reinforced PEEK and PPS components, the coating protecting the structure against gamma radiation and heating up to 260°C.
18. To increase the cost effectiveness of the materials developed through automation and optimisation of the manufacturing procedures (particularly the CNT functionalisation process), to increase the strength-to-weight and stiffness-to-weight ratios in order to maximize fuel and energy savings, to minimize the use of active protection systems, to maximize safety and also to promote recyclability.
19. To improve techniques for the recovery and EOL disposal of waste materials to ensure that the cost of the fabrication process is minimised.
Project Results:
Introduction
This project dealt with:
The functionalisation of carbon nanotubes (CNTs), incorporating microwave heating
The incorporation of the CNTs in thermoplastic matrices and assessment of the quality of the resulting material
The manufacture of composite pre-pregs and laminates using these CNT filled matrices, including novel consolidation technologies
The evaluation of the properties of the laminates
The development of sensor technologies for monitoring the health of the multi-scale materials
Development of DIRIS honeycomb structures
Recycling of multiscale-reinforced thermoplastics
Numerical modelling of the multi-scale materials
Manufacture and testing of components
Life-cycle assessment of a multi-scale reinforced engine stiffener
Risk assessment
Assessment of Uncertainties
The development of surface coatings for the laminates to prevent attack of the laminates by UV and to control temperature build-up within the laminates due to solar radiation
This structure will be used to detail the development and results obtained within the project.
Functionalisation of CNTs
Two high performance thermoplastics were chosen for the development work within the project, these being poly(ether ether ketone) [PEEK] and poly(phenylene sulphide) [PPS].
Two types of CNTs were considered for use in the project, both coming from European manufacturers. These were:
Baytube C70P. These CNTs are supplied in a pelletised form. This has the advantage that they are better from a health and safety perspective as regards handling. However, the pelletised form makes them more difficult to disperse and good dispersion is important in respect of functionalizing the CNTs and also in the properties they confer on the polymer once the materials are combined.
Nanocyl NC7000. These CNTs are supplied as loose nanotubes and thus there are increased hazards in handling them but they are easier to disperse. For the reasons of ease of dispersion, the Nanocyl CNTs were chosen for this project.
The intention of the functionalisation process was to attach oligomers to the CNTs which would improve the compatibility and hence dispersability of the CNTs in the chosen polymers. In the case of PPS, the oligomers were short PPS chains and in the case of PEEK the oligomers were poly (ether ketone) [PEK] oligomers. PEK oligomers were chosen due to the simpler electrophilic chemistry associated with the polymerisation and its compatibility with PEEK.
The original intention had been to develop a functionalisation process based on the use of a baffled continuous-flow meso-reactor. However, the functionalisation chemistry initially chosen utilised polyphosphoric acid [PPA] and this had a high viscosity which was not compatible with the chosen process. As a consequence, two alternative approaches were adopted in order to ensure a solution to this problem, which was fundamental to the project, and these were:
Development of new design of reactor which could accommodate the high viscosity PPA reaction mixture. This route was used for the PEEK-like functionalisation of the CNTs.
Development of alternative chemical routes by Nanozar, these involving the use of low- viscosity solvents such as NMP. This was used for the PPS-like functionalisation of CNTs and also investigated as a route to the PEEK-like functionalisation.
Details of the various chemical routes investigated are given below.
PEEK-like Functionalisation
Synthetic route #1
This route involved functionalising the CNTs by following a Friedel–Crafts acylation, initially proposed for a PEEK matrix by Baek et al (Polymer 44, 2003, 4135-4147). In the reaction PPA and phosphorus pentoxide [P2O5] are used as drying reagents but also act as a solvent, Friedel-Crafts catalyst and dehydrating agent. The first stage of the process involves 4-phenoxybenzoic acid reacting and bonding onto the CNTs, this being followed by the in-situ polymerisation of 4-phenoxybenzoic acid to form the PEK oligomers. The published method was carried out at 130°C for 48 hours followed by water extraction over three days and a further three days extraction utilising methanol.
Synthetic route #2
This involves the oxidation of the CNTs and anchoring them to the PEEK backbone. A patent has been applied for relating to this route.
Synthetic route #3
This route involved the functionalisation of CNTs with PEEK monomers and the in-situ polymerisation of PEEK, the CNTs being introduced during the synthesis of the PEEK. This route involves the addition of phenolic groups onto the surface of the CNTs, the phenolic groups featuring hydroxyl groups which can take part in the PEEK polymerisation process. The initial stage of this process takes approximately twelve hours at 60°C with DMF as the solvent.
All three routes proved to yield functionalised CNTs.
Route #1 was used for the manufacture of the PEEK-like functionalised CNTs. This route initially involved several steps and several days for completion. An extensional flow reactor was developed, this reducing the timescale for the functionalisation process to thirteen hours. The reciprocating extensional flow reactor featured 10µm filters through which the reaction mixture flowed, this promoting disentanglement and dispersion of the nanotubes, this being evidenced by the changing rheology during processing. Micrographs from an SEM and images from an AFM indicated that the nanotubes were no longer clustered. The process developed:
Had the capability of splitting a volume of more than 1lt into millions of micro-cylinders, with lengths a few times greater than the length of the CNTs.
Produced elongational flow rather than shear flow using convergent – divergent flow geometries.
Had the capability to provide good control of the flow parameters.
Exhibited upstream and downstream flow to re-open blocked channels.
Was energy efficient.
Exhibited limited re-aggregation and temperature increase.
Resulted in limited attrition of the CNTs.
Was capable of being scaled–up.
Reduced the reaction time for functionalisation by a factor of ten from that originally used by Baek.
The productivity of the extensional flow mixer (from CNT disentanglement to homogenisation) was 1kg/24h (1kg of functionalized CNT/PEEK slurry).
The procedure was validated periodically by the use of FTIR examining the spectrum for the loss of the peak at 2922 cm-1, this indicating no free defect sites or open ends on CNTs, and a peak at 1650 cm-1 for the C=0 bond, this indicating the functionalisation. Analysis using an SEM indicated that entangled CNTs were found in <1% of all the images examined.
If the CNT disentanglement is 100% successful every defect site is available for reaction with the monomer. If the CNTs are fully functionalised, the C-H stretching band disappears. [Jasco, Ir application note 02-03 http://www.jascoint.co.jp/asia/literature/application_notes.html ]
Numerical simulation, using a Molecular Mechanics Method, and also observations from SEM images of functionalised CNTs indicated that the degree of functionalisation was small (< 10 chains per CNT) but that the PEK polymerisation around the functionalised CNT increases the CNT diameter by, on average, 3nm.
Upscaling the PEEK-like Functionalisation Process
Following the validation of the suitability of the extensional flow mixer in performing the tasks of CNT disentanglement, dispersion, distribution and functionalisation and their homogenisation with PEEK grains, the up-scaling to a pre-industrial model was carried out by improving the efficiency and increasing the capacity of the equipment. The improved mixer configuration included a simultaneous translation and rotation of an assembly of a great number of micro-channels, these being part of the piston this process resulting in:
Flow splitting, recombination and rearrangement
Flow twisting, folding and reorientation (shearing of the stream lines)
Transition from laminar to turbulent flow at the exit of micro-channels
It should be noted that the mixer operation was changed from disentanglement to functionalisation by simply changing the filter unit.
The new extensional cross-flow mixer, performed all the functions for a mixture over a 24 hour period.
Microwave Heating of Nanocomposites
Microwave heating of materials is preferred to conventional heating, as microwaves heat the material at the molecular level, i.e. volumetrically, and there are no temperature gradients when heating homogeneous materials. In contrast to that, only the material surface is heated in conventional heating, and heat has then to be transferred to the inside of the material by conduction, which is a very inefficient process for materials with low thermal conductivity.
A detailed investigation was carried out and it was concluded that the use of microwave heating in combination with the given scheme of mechanical and thermal processing of the nanocomposite had several severe disadvantages:
The need for simultaneous heating of the high pressure chamber.
High temperature gradients caused by inhomogeneity of the energy absorbed in the cylinder, both along its axis and along its diameter.
The high cost of microwave generators.
The sequence of the proposed mechanical and thermal processing which made the microwave method difficult to apply in a continuous flow production line.
The harmful effects of microwaves on the human organism which make it necessary to take additional measures for cutting off the electromagnetic energy from the exit slot from the equipment.
Thus, the microwave method proved to be incompatible with the proposed scheme of mechanical and thermal processing of the nanocomposite material, since the material is permanently in contact with the walls of the chamber. The material had to be heated by conventional means, which were much less expensive as well as being more reliable.
PPS-like Functionalisation
Two routes were studied in the case of functionalisation of CNTs with PPS-like compatible groups.
Synthetic route #1
This route involved a first stage of reacting CNTs with PPA and 4-chlorobenzoic acid in the presence of phosphorous pentoxide and, in a second step, reacting the product with 4-methylbenzenethiol in NMP/toluene in the presence of sodium carbonate. This mechanism introduced 4-(4- methylbenzenethio)benzoyl groups onto the CNTs, these being highly compatible with the PPS. The functionalised CNTs obtained in the first step were then reacted with chlorobenzenethiol or dichlorobenzenethiol to give anchored linear poly(phenylene sulphide) (LPPS) or hyperbranched poly(phenylene sulphide) (HPPS).
In the second step PPS compatible groups were attached to the grafted monomers. The first reaction uses methyl thiol as reactant and 2,4-dichlorothiol as reactants and the products obtained were characterized by TGA and FTIR.
The first intermediate product was synthesised in the laboratory with a good yield. The reaction was conducted at 130ºC over a period of 72h under mechanical stirring. The product obtained was soxlhet extracted with water and methanol for 3 days with each solvent to remove the non-grafted molecules. The product obtained was dried under vacuum and characterised by FTIR and TGA.
Synthetic route #2
The first route had problems associated with it due to the high viscosity of PPA and so up-scaling was difficult and thus an alternative was considered.
In a second step the grafted molecules react with sodium sulfide in the presence of dichlorobenzene resulting in PPS compatible groups grafted onto the surface of the tubes.
All the products were characterised by means of TGA and FTIR in order to determine if the reaction had occurred and to estimate the amount of grafted polymer. With both strategies the results obtained where optimised and characterised.
Upscaling the PPS-like CNT Functionalisation Process
An oscillatory flow reactor (OFR) was developed and validated within The Polymer Fluids Group at The University of Cambridge.
The initial functionalisation process, synthetic route #2, was a two-step reaction with long reaction times. Tests were carried out using a small lab scale stirred 250 ml reactor to establish if reaction times could be shortened and whether performing a single stage reaction was viable. It was found that it was possible to reduce the reaction time to 6 hours (4 hours for the first reaction and 2 hours the second reaction) compared with the 24 hours originally proposed. In addition it was discovered that it was possible to move from a two-step reaction to a single-step sequential reaction taking 6 hours using DMF as a solvent, thereby saving reaction, washing and separation times.
Key factors in the scale up process were the initial dispersion of the CNTs and the concentration of CNTs in the reactor. Without surfactants, CNT aggregation occurs even at low concentrations and so good dispersion is difficult to achieve. However, aggregation greatly enhances the separation time of CNTs from the solvent after reaction and also has benefits in handling the functionalised CNTs on completion of the reaction.
The initial CNT concentration used at lab scale was around 0.15 %, which was considered far too low for economic up-scaling. In order to increase the initial concentration and to optimise the up-scaling process the possibility of increasing the initial concentration to 1 % was assessed at lab scale. Despite CNT aggregates being present in the bulk of the reaction, there was clear evidence that functionalisation yields were in the same range as those obtained using lower concentrations.
Scale up of the functionalisation process was carried out using an OFR. By combining oscillation with periodically spaced baffles in a tube, vigorous eddy mixing can be achieved between the baffles and this leads to excellent mixing (Mackley MR and Ni X. Experimental fluid dispersion measurements in periodic baffled tube arrays. (1993) Chemical Engineering Science, Vol. 48, No 18, 3293-3305 and Ni X, Mackley MR, Harvey AP, Stonestreet P, Baird MHI and Rama Rao NV. Mixing through oscillations and pulsations. A guide to achieving process enhancements in the chemical and process industries. (2003) Trans IChemE, Vol. 81, Part A, March 2003). These baffles are usually low-constriction orifices plates where the fractional open cross-sectional area is based on a compromise between minimising frictional losses and maximising the mixing effect. The baffles are typically spaced at a uniform distance apart, in the range 1-2 times the tube diameter, with a spacing of 1.5 being the most common. A linear actuator causes the oscillation, moving the set of baffles up and down from the top of the reactor.
The OFR consisted of a jacketed reaction vessel of 104 mm internal diameter and 3 litres capacity although only 2.5 litres of reaction fluid were used. The vessel was operated vertically and connected to a linear actuator which caused the oscillation of the set of baffles, in this case in a reciprocating plate column. During oscillations, it was possible to specify both the frequency and the amplitude of the movement, giving a wide range of mixing conditions. The oscillation conditions were a frequency of 1 Hz and the amplitude centre to peak was 10 mm giving a moderate oscillation in the eddies formation region. The temperature was controlled by a Huber CC302 oil circulator which was able to reach the temperature needed for the reaction. Once the reaction had finished, the CNT aggregates were allowed to settle for a few minutes under gravity and then washed and filtered using Omnipore Membrane filters 1.0 µm coupled with a vacuum pump.
Following all the reactions TEM results showed a layer coating the CNTs. IR spectroscopy was also indicated that the CNT functionalisation was successful in both the one and two-step reactions.
TGA analysis was performed to quantify the amount of functional groups attached and to compare the yield of surface functionalisation between the one-step and two-step reaction process.The functional groups generated using the OFR with 1 % CNTs present showed high thermal stability and the yield was higher than in the initial lab scale experiments. The yield was slightly higher (35% compared to 30 %) if the reaction was carried out in two steps but the additional yield was considered not large enough to justify the longer processing time associated with the two-step process. In particular, the additional washing and filtration in the two-step reaction slowed production significantly and utilised a large volume of solvent.
Once the reaction protocol in the MRECT OFR was established and validated for a single step reaction with an initial concentration of 1 %, another trial was carried out with an initial concentration of 2 %. The TGA analysis showed encouraging results with a yield of grafted functional groups higher than 25 % and with high thermal stability. The output of functionalised nanotubes from the OFR was around 500g/month.
The OFR produced aggregated CNT dispersions at all concentrations tested. Initially there were concerns that the aggregates would inhibit functionalisation, however it was realised that this was not the case and the aggregation provided a strategic advantage in that on completion of a reaction fluid oscillation could be stopped and the CNT aggregates would settle in the bottom of the reactor within minutes. This feature meant that it was not necessary to use very long settling times or high pressure small pore size filters and had advantages in respect of handling the materials at the compounding stage.
Incorporation of CNTs into Thermoplastic Matrices
Incorporation of CNTs into PEEK
Problems were experienced incorporating the functionalised CNTs in to PEEK and two routes were adopted to overcome the problems:
Developing and scaling-up a melt mixer which operates on similar principles to the equipment used for the functionalisation process as discussed previously.
Industry standard process using a twin-screw extruder.
Novel Melt-Mixing Device
The construction of the initial equipment comprised of an outer static chamber, a symmetrical piston and a double rod system which rotated and translated simultaneously within the chamber.
Techniques were developed to determine the uniformity of the CNT dispersion by analysing the density of CNTs in systems of concentric rings of constant area on SEM images. The product from the melt mixer was subsequently processed by hot-compaction in order to produce a mechanically-stable tape which could be handled and fed into the next stage of the process. The compaction conditions were 250 MPa biaxial pressure at 370°C, this producing a micro-void free tape 600mm long.
The hot-compaction system increased the density of the material from 0.91g/cm3 to 1.38g/cm3 which is higher than normal crystalline PEEK. The tapes produced in the hot-compaction process were then fed into a calender system to further enhance the dispersion, distribution and orientation of the CNTs.
The tapes were then subject to sequential iso-stretching and annealing aimed at increasing the levels of orientation. The localised heating system on the equipment developed during the project ensured local uniform deformation of the tape at the minimum acceptable temperatures. This process also sought to reduce the natural waviness of the CNTs, aligning them in the direction of stretching.
Due to damage sustained in the mixing of the CNTs with PEEK, the melt-mixing device had to be rebuilt such that it was fed via a single screw extruder and featured a simple slit die at the exit to produce a film. The film was cooled rapidly with water sprays so that it was amorphous. This film was then subjected to stretching as before. Samples of the film evaluated indicated that there may be some residual undispersed and unbound carbon nanotubes and suggested that some further work may be needed to perfect the design. Outputs from the equipment were quite low when compared to the capacity of the alternative route, process of mixing and dispersing the CNTs in the machine taking several hours and so the process did not really constitute a viable commercial process with PEEK.
A CNT concentration of 0.5 wt% was adopted for all the materials investigated in the project on the basis of:
It was found that even with a concentration of 0,5 wt% CNT, the dispersion of the CNTs was not feasible following processing for several hours in the melt-mixing equipment.
Examination of the electrical conductivity of aligned PEEK/CNT tapes with CNT loadings from 0.15 to 1.5 wt% indicated that the percolation threshold occurs at about 0.2 wt% after processing on the melt-mixer.
The optimum MWCNT loading in PEEK composites for thermal stability is 0,5 wt%, according to P. Patel et al [P.Patel et al. Flammability properties of PEEK and CNT composites, Polymer Degradation and Stability (2012) 1-11].
Twin-Screw Extrusion
The second route to producing a 0.5% CNT loaded PEEK was by compounding of the functionalised CNTs with PEEK using a Coperion ZSK32MC twin-screw extruder. The compound was then ground down to produce a powder with a d50 of approximately 1000μm for subsequent use.
Initial trials to process the PEEK/CNT powders using a twin-screw extruder resulted in the PEEK degrading. Elemental analysis indicated that the material contained a high level of phosphorus which can be an issue when processing PEEK in the molten state. The phosphorus was assumed to originate from the polyphosphoric acid used in the functionalisation process. Compounding PEEK with unfunctionalised CNTs did not result in any similar degradation-related problems and the elemental analysis results indicated that the phosphorus levels were very low in comparison. Analysis of the material using DSC, TGA and thermal desorption with GCMS showed that the polymer was degraded.
A second sample of PEEK/CNT material was supplied for compounding, elemental analysis indicating that the residual phosphorus levels were significantly lower than in the first sample, 210ppm as opposed to 800ppm. This material was compounded successfully.
It was unclear as to whether some aggregation and attrition of CNTs would occur via this route. However, subsequent results indicated that this was not the case.
Incorporation of CNTs into PPS
Initial work involved processing and testing PPS neat resin as a reference, PPS with 2% CNTs and
3% CNTs and PPS with pre-dispersed CNT masterbatch as a reference for CNT dispersion. The CNTs used in this work were BaytubeC70P and the compounding was carried out using a Coperion ZSK32MC machine with the low-shear screw profile.
The properties of the compounds were assessed by:
Surface Resistivity: correlated to the dispersion of CNTs creating a network inside the polymer matrix
Tensile elongation at break and Impact Resistance: suitable to reveal the presence of agglomerate of CNTs in the compound
As a consequence of the high surface resistivity found with the Baytubes, a higher-shear screw profile (VHS) was examined. The work indicated that better dispersion was obtained if the CNTs were fed into the extruder part way along the barrel. It was also found that feeding the CNTs part way along the barrel whilst using the HT screw profile gave better, more consistent results. This approach produced the optimum shear and residence time for the materials. The optimum screw profile was found to be a compounding length of 20/1 L/D with the CNTs being added half way along the extruder barrel. The PPS is added at the rear of the machine so that it is molten before the addition of the CNTs. This is to prevent the CNTs sintering together as this makes dispersion more difficult. The CNTs are added part way along the machine in order to limit the amount of work carried out on them and so limit any possible attrition effects.
The optimised process resulted in moulded specimens with a resistivity of 104 ohm at a CNT loading of 3.25% using Baytubes C70P.
Baytubes are supplied in a pelletised form and so tend to present less of a potential health hazard than Nanocyl NC7000 CNTs which are very dusty. Techniques had to be developed as part of this work to ensure good levels of safety.
The work resulted in the manufacture of 3.0% loaded Nanocyl NC 7000/PPS compounds with a surface resistivity of 104 ohms. Further optimization of the processing parameters was carried out, the result being CNT/PPS compounds with a surface resistivity of 104 ohms at a loading of 2.5% of Nanocyl NC7000.
Attempts were made to produce masterbatches with CNT loadings of 17.5% but it was found that the CNTs tended to be packed together and the dispersion was poor. This approach was abandoned.
The initial screw profile used with functionalised CNTs, the High Shear profile, was found to be the best profile for the unfunctionalised CNTs. The results obtained with this configuration indicated that the product was sub-optimal and so the material was re-processed on the extruder to determine if this would generate an improvement. Reprocessing did produce improved results and so this indicated that process modifications were required to improve the quality of the compounds produced.
The revised screw configuration was the same basic screw configuration but the CNTs were added earlier in the process. Some mechanical properties showed an improvement over the compound produced with the initial machine configuration. The effect of the second loop through the machine was less pronounced in this case.
A third configuration of the machine was evaluated where the compounding length for the CNTs was maximised.
Following decisions within the consortium, the CNT loading was reduced to 0.5%. Several further iterations were carried out using:
• Functionalised CNTs
• Powder and pellet PPS
• Different molecular weights of PPS
The results indicated that:
The functionalised CNTs were less susceptible to sintering than the non-functionalised CNTs.
The dispersion of the functionalised CNTs was best with the maximum compounding length.
The process of dispersing the functionalised CNTs was different to that found with the unfunctionalised CNTs.
It was noticed that the theoretical reinforcing action of CNTs was not observed in the injection moulded samples produced.
Compounding work had been carried out on a small scale at another partner and this indicated an improvement in mechanical properties and this did not correlate with the results obtained on commercial scale equipment.
It was noted that there were some inconsistencies in the various data sets for the polymer and the compounds:
The base PPS resin is Fortron 0205 but the data produced on the lab scale was different from those reported in the technical data sheet of the manufacturer.
The mechanical data for PPS with unfunctionalised CNT compounds generated on the lab scale appeared to be much worse than those generated on the commercial scale at the same CNT concentration.
In contrast, the data for PPS with functionalised-CNT compounds generated on the lab scale appeared to be much better than those obtained on the commercial scale at the same f-CNTs concentration.
The reason for the differences was unclear and so extensive evaluation of the dispersion of the CNTs in the PPS was undertaken using SEM. The SEM work indicated that the CNTs were well dispersed in all the materials as well as in composite laminates produce using the materials.
There were a number of differences in the sample preparation methods used, these being:
Lab scale processing involved premixing the materials before processing them, in a molten state, on a twin screw extruder. The materials were milled as part of the premixing process.
The lab scale compounding was carried out using a Haake Minilab which is a small laboratory twin-screw extruder.
The lab scale materials were converted into test samples by compression moulding whilst the commercial scale materials were converted using injection moulded.
In injection moulding it is well know that with short-fibre reinforced polymers the cross-section of the component has a skin/core structure. In the skins, the fibres are predominantly oriented parallel to the flow whilst in the core they are transverse to the flow. The processing conditions and rheology of the polymer have a significant impact on the thicknesses of the skin and core layers. Thus it is possible that the core layer of the injection moulded samples was significantly large and so most of the CNTs were oriented perpendicular to the test direction and thus had little effect on the properties.
The flow occurring during compression moulding is limited and as a consequence, the CNTs were more likely to adopt a random orientation in the plane of the moulding. A random orientation would result in a significant number of CNTs being aligned in the direction of test and thus one would expect the improvement in performance that was observed.
It is believed that these flow-related effects were the primary reason for the differences in the test results.
For the impregnation of the carbon fabrics to form a pre-preg, the material had to be ground into a powder. The target particle sizes for this work was a d50 of 1000μm for Ten Cate and a d50 of between 50 and 100μm for Fibroline.
A d50 of 1000µm powder was relatively easy to achieve by normal polymer processing techniques. Particle sizes below 100µm were, however, found to be more difficult to achieve and a number of methods were investigated.
Initial work was carried out using a pin mill and this produced powder which had a particle size of under 1000µm but the particle sizes required by Fibroline could not be achieved.
Following this an underwater pelletiser system was investigated using pure PPS. This produced very regular sized spherical particles around 400µm in diameter. Whilst the process was capable of easy industrialisation, the particle size was still too large for Fibroline.
The next process investigated was particle size reduction by impact at very high speed. A laboratory device was used where the powder was fed tangentially at subsonic speeds (approximately 50 m/s) into the flat cylindrical milling chamber through a Venturi system using pressurized air or nitrogen. Once inside the milling chamber the particles were then accelerated by a series of jets around the perimeter to supersonic speeds (300 m/s), in a spiral movement. The micronising effect occurs when the slower incoming particles and the faster particles, in the spiral path, collide. While centrifugal force retains the larger particles at the periphery of the milling chamber, the smaller particles exit with the exhaust gas from the centre of the chamber.
Tests on small sized samples showed that 100µm could be achieved but unfortunately in the parallel work on impregnation, it had been found that powder with a particle size of 100µm was unsuitable for their process, the process requiring a particle size of around 20µm.
A Ball Mill machine was developed and some preliminary tests on small samples of PPS with a 15% loading of CNTs and PPS with a 17,5% loading of CNTs were by IMMG, the materials being processed for 4, for 8 and 12 hours. The outcome of these tests was not satisfactory as the particles became flat, a few millimeters in size, with a thickness of around 0.3 mm.
A ring mill was constructed, this possessing superior grinding performance when compared to ball mills as the energy density is around a thousand times greater. The equipment featured a chamber which contained a revolving shaft that had sets of rolls attached to it. As the shaft rotated, the rolls were forced radially out against the wall of the chamber. This type of mill was reported to have worked well with minerals but unfortunately was unsuccessful in meeting the particle size requirements specified.
External trials did indicate that grinding down to a d50 of 20μm was possible, one design of grinder successfully manufacturing a powder with a d50 of 23µm, but the production rate was so low that it would not be a commercially viable process.
Subsequent work on the novel impregnation technology resulted in powders of 50μm being suitable for the process. However, the problem of grinding the CNT reinforced materials down to a d50 of 50μm was not resolved.
Quality Measurement
Techniques need to be available for the quality assessment of the compounds manufactured and also, ideally, for the on-line assessment of the quality of product.
Differential Scanning Calorimetry (DSC)
Data obtained using DSC indicates that the levels of crystallinity achieved with a CNT loaded PEEK is less than that obtained with pure PEEK. It is believed that this arises due to the restricted chain mobility arising from the interaction of the functionalised CNTs and the PEEK molecules. This indicates that the functionalisation has been successful and that there is significant interaction between the CNTs and the PEEK.
DSC was used to investigate the glass transition region. Experiments were performed at a heating rate of 5°C/min. In all experiments, a weak glass transition is obtained, which is expected for a semicrystalline material. Figure 46 shows heating thermograms obtained with PEK/CNT nanocomposites at various compositions as indicated on the Figure. The slight Tg reduction in the nanocomposites, obtained after evaluation following standard procedures, is attributed to the increment of polymer mobility [Ash, B.J. Siegel, R.W. Schadler, L.S. 2004. Glass-transition temperature behaviour of alumina/PMMA nanocomposites. J. Polym. Sci. B 34, 4371-4383], indicating that the interface between matrix and nanotubes must be poor and mechanical properties are not improved [Zax, D.B.,Yang D.-K. Santos, R.A. Hegemann, H., Giannelis, E.P. Manias, E., 2000. Dynamical heterogeneity in nanoconfined poly(styrene) chains. J. Chem. Phys. 112, 2945-2951].
AC and DC conductivity measurements
Conducting networks formed within an insulating matrix can be utilised as highly sensitive sensors for detecting the onset, nature and evolution of damage in advanced polymer-based composites. The samples firstly had gold stripes sputter coated onto them, copper electrodes were attached to the stripes using a conducting silver-loaded epoxy, the samples then being mounted on the tensile testing machine.
DC techniques are mainly sensitive to fibre failures whereas AC measurements provide information on the development of both intra-ply and inter-ply matrix cracks, the latter being associated with delamination.
The results showed a normalised resistance change, this indicating that there is a significant increase in (ΔR/R0) before the plastic deformation of the sample. It is also noted that (∆R/Ro) changes faster than strain in the elastic region.
Various configurations of electrode have been evaluated, the work showing that 4-wire sensing was superior to 2-wire sensing and it was noted that the electrical resistance linearly increases with strain until fracture and the resistance changes immediately from minimum strain. Changes in resistance were also observed when cyclic loading was applied. The results showed a reversible change of resistance on every cycle, the cycles being repeatable. A small increase in resistance was observed at zero stress after many cycles.
Dynamic Mechanical Analysis (DMA)
Dynamic mechanical analysis at 1Hz has shown significant differences in the behaviour of PEEK and CNT loaded PEEK materials around the glass transition temperature.
DMA measurements for the evaluation of glass transition and the assessment of improvement of mechanical properties were performed using the TA Instruments DMA Q800 in the tension mode with a pre-strain equal to 0.7%, a strain amplitude of 0.5 %, a temperature range 20-230°C, at five frequencies of 1, 5, 10, 20 and 40 Hz, and a heating rate of 3°C/min. The storage and loss modulus curves versus temperature were evaluated, as well as the corresponding master curves, applying the time-temperature superposition (TTS) principle.
Scanning Electron Microscopy (SEM)
Various samples have been examined using a Nova NanoSEM 230, Fei-Innova Nanoscope, and micrographs are indicate the presence of the nanotubes on the surface of the polymer powder particles and the nature of the compacted samples prepared using the hot-compaction system discussed previously. All samples were coated with a thin film of gold (~7 nm).
Attempts were made to observe the CNTs below the surface of the tape. Chemical etching with phosphoric acid [Olley, R. H., Bassett, D. C., Blundell, D. J., & Division, P., 1986. Permanganic etching of PEEK. Polymer 27, 344–348] did not reveal the CNTs in the matrix. Further attempts were made using laser ablation at 213 nm, using a Nd:YAG laser. The results were poor and this method would need optimisation in order to improve the results. The use of a laser at 355 nm was attempted [Romoli, L., Fischer, F., Kling, R., 2012. A study on UV laser drilling of PEEK reinforced with carbon fibers. Optics and Lasers in Engineering 50, 449–457] but the results were worse due to thermal diffusion in the polymer. There is also a question as to whether laser ablation results in redistribution and/or damage of CNTs [Qian, D., Dickey, E.C. Andrews, R., Rantell, T., 2000. Load transfer and deformation mechanisms in carbon nanotube-polystyrene composites. Appl. Phys. Lett. 76, 2868-2870].
Assessment of the attrition of the CNTs
SEM and AFM observation at the different stages of processing and ac electrical conductivity measurements on the CNT reinforced PEEK sheets were used to assess possible attrition of CNTs caused by processing.
SEM images were used to provide evidence regarding possible attrition of the CNTs at the different stages of processing.
PEK/CNT samples of varying CNT content prior to calendering were selected for ac conductivity measurements. Two electrodes configuration measurements were performed using an Alpha Analyzer (Novocontrol). Measurements were performed using gold sputtered circular electrodes on opposite sides of the sample in order to assure good electrical contact between the sample and the electrodes of the apparatus. The sputtering process was performed using an EMS 550 Sputter Coater operating with sputtering current of 25mA and sputtering time of 4min. From the ac conductivity plots values of dc conductivity were determined, following standard procedure [Logakis, E., Pandis, Ch., Pissis, P., Piontek, E., Poetschke, P. 2011. Highly conducting poly(methyl methacrylate)/carbon nanotubes nanocomposites: Investigation of thermal, dynamic-mechanical, electrical and dielectric properties. Comp. Sci. Technol. 71, 854-862]. These values were used to construct the percolation plot and the following equation from percolation theory was fitted to the data and percolation threshold was determined to be φc = 0.13 % (corresponding to pc=0.17 wt%).:
σdc ~ (φ-φc)t
where σdc is the dc conductivity, φ is the volume fraction of the filler, and t is a critical exponent related to the dimensionality of the investigated system. The extremely low percolation threshold provides strong evidence in support for a fine dispersion of CNTs in the PEK matrix.
A model from the literature, the Li model [Li, J., Ma, C., Chow, W. S., To, K., Tang, B. Z., Kim, J. K.2007. Correlations between percolation threshold, dispersion state, and aspect ratio of carbon nanotubes. Adv. Funct. Mater. 17, 3207-3215] was used to calculate the aspect ratio of CNTs in the nanocomposite from the measured percolation threshold and from the mean length of the CNTs which was taken as1.2 µm this being close to the value of 1.5 µm given by the manufacturer. Thus, the results of analysis suggest that the initial length of CNTs is preserved in the nanocomposites.
Determination of PEEK crystallinity
The main experimental technique employed for the assessment of crystallinity in this project was DSC as discussed earlier. Additional information on crystallinity has been provided using X-Ray Diffraction (XRD).
Three different samples were measured, one pure PEEK as reference (the sample was commercial product from Victrex), one PEEK+0.5wt%CNTs and one PEEK+0.5wt%CNTs which had been strained and annealed. The first sample was in the form of sheet and the other two were in the form of tapes of 0.5-1.0 cm width. An X'Pert Pro X-Ray Diffraction System (PANalytical) was used for the measurements which were in the range of 2θ=10°-35°.
The results show that the position of peaks in the pure PEEK sample agrees with the literature [Goyal, R. K., Negi, Y. S., & Tiwari, A. N. 2005. Preparation of high performance composites based on aluminium nitride/poly(ether–ether–ketone) and their properties. Eur. Polym. J. 41, 2034–2044]. Some of those peaks also appear in the PEEK/CNT nanocomposites. A new peak at 27.3° suggests that the incorporation of the CNTs and processing create new interfacial phases and affect the crystallisation [Wang, L., Weng, L., Song, S., Zhang, Z., Tian, S., Ma, R., 2011. Characterization of polyetheretherketone–hydroxyapatite nanocomposite materials. Mater.Sci. Engin. A 528, 3689–3696]. The degree of crystallinity can also be determined from XRD measurements, but the results are less reliable when compared to DSC, and the main power of XRD is in the study of crystalline structures.
Alignment of CNT in the Nanocomposites
One of the goals of the project was to design, develop and employ special preparation/processing methods resulting in an alignment of CNTs in the thermoplastic matrix. Thus, there was a need from the point of view of characterisation to design and develop methodologies for assessing the alignment of CNTs in the thermoplastic matrix. The methodology used in this work was based on conductivity measurements before and after annealing of samples. Other techniques were used with the electrical measurements in an attempt to confirm the observations made.
The methodology of using electrical conductivity measurements before and after annealing for assessing alignment of CNTs in the nanocomposites is based on the observation that conductivity increases after annealing [Cipriano, B.H. Kota, A.K. Gershon, A.L. Laskowski, C.J. Kashiwagi, T., Bruck, H.A. Raghavan, S.R. 2008. Conductivity enhancement of carbon nanotube nanofiber-based polymer nanocomposites by melt annealing. Polymer 49, 4846-4851], this probably being due to redistribution of CNTs.
Values of conductivity are significantly higher with no calendering. Moreover, the increase of conductivity after annealing is significantly more pronounced in the case of calendering. Three main reasons could explain the remarkable decrease of conductivity after calendering:
With increasing alignment electrical conductivity in the nanocomposites is lost, in other words the percolation threshold increases.
The calendering process may lead to breakage of the CNTs, which leads to a decrease in their aspect ratio. The result of the above is again to increase the percolation threshold.
Mechanical mixture of PEEK and CNT powders may create a structure with PEEK particles covered by the powdered nanotubes. Thus, the conductive phase of the CNTs is located on the surface of the polymer particles. Hot pressing deforms polymer particles and results in formation of a compacted continuous polymer phase, where conductive paths of CNTs are located on the boundaries between particles. After calendering, these paths are destroyed and conductivity is lost.
Thus, the results of ac measurements provide additional evidence that CNTs are aligned in the polymer matrix after calendering and that alignment is partially destroyed by annealing, which leads to an increase of conductivity.
Raman spectroscopy is a well-established technique to study and quantify alignment of CNTs. Repeated attempts to make use of this technique in the case of the PEEK/CNT nanocomposites manufactured during this project were unsuccessful due to the high fluorescence of the PEEK used. Change of the wavelength of excitation could not solve this problem.
In order to establish this technique for the needs of the project, PPS/CNT nanocomposites prepared within the project were investigated. Based on the method of preparation, no alignment of CNTs was expected in these nanocomposites. No differences at the intensity of the G mode at about 1590 cm-1 have been recorded, which provides evidence that the CNTs are not aligned [Haggenmueller, R., Gommans, H.H. Rinzler, A.G. Fischer, J.E. Winey, K.I. 2000. Aligned single-wall carbon nanotubes in composites by melt processing methods. Chem. Phys. Lett. 330, 219-225] as was expected from the conditions of preparation/processing.
Manufacture of Composite Pre-pregs and Laminates
State-of-the-Art Process
The normal approach to the manufacture of fabric pre-pregs is via a powder coating process. This process consists of:
Finely ground polymer powder is first sprinkled onto a fabric.
The powdered fabric then runs through an infrared heated oven, which melts the polymer powder.
Finally, a set of calendering rolls are used to fuse and quench the polymer onto the fabric.
Double sided powder coating is achieved by reversing the single sided coated fabric and repeating the process.
Carbon woven fabric reinforced CNT/PPS
Fortron 0205 PPS reinforced with CNTs was evaluated, together with virgin Fortron 0205 powder which was used as the reference material. Four types of carbon woven fabric reinforced PPS prepreg were initially produced:
Carbon fabric with Fortron PPS 0214 powder. The PPS 0214 is the standard polymer used by Ten Cate Advanced Composites [TCAC] for prepregs and laminates that are applied in aerospace structures. An extensive database has been built up at TCAC, so the PPS 0214 was used as a reference against which the processing of alternative PPS grades could be compared.
Carbon fabric with Fortron PPS 0205 powder, which was used as a reference to compare the CNT/PPS 0205 with.
Carbon fabric with CNT/PPS 0205 on both sides.
Carbon fabric with CNT/PPS 0205 on one side and with PPS 0205 on the other side. It was anticipated that the CNTs would considerably increase the viscosity of the PPS, which might result in poor impregnation of the carbon fabric during consolidation. By adding virgin PPS the chance of achieving acceptable impregnation would be increased. If the CNT/PPS does not migrate into the fabric, then it might improve the interlaminar properties of the woven fabric composite. The CNT/PPS acted as an interply reinforcement in this case.
Laminates were produced using the different prepregs. The quality of the laminates was assessed on the basis of the following aspects:
Visual appearance; the surface of the laminates that were based on the CNT/PPS showed whitish cracks. The original CNT/PPS was black. A pronounced relief of residual stresses was observed in the form of noise during cooling.
Flow of the matrix; the laminates were consolidated between heated platens which allow free flow of the melted resin during the consolidation. The amount of flow is an indication of the viscosity of the thermoplastic resin.
Thickness of the laminate; the thickness is a measure for the state of impregnation or void content.
Ultrasonic damping; the laminates were checked by means of C-scan.
Mechanical performance; the laminates were subjected to three point bending testing, which is generally a good indicator for the consolidation quality.
The impregnation behaviour of the CNT/PPS appeared acceptable. From the thickness measurements and microscopic evaluation, it was observed that the flow of the CNT/PPS into the fibre bundles was the same as found with the pure resin. However, the intralaminar strength of the woven fabric CNT/PPS laminates was very poor: Individual plies could be separated manually. The poor intralaminar strength was confirmed by the response of the ultrasonic measurement. A large amount of damping with respect to a reference C/PPS laminate was measured. Subsequent analysis by optical microscopy revealed cracks, which were located within and around the fibre bundles.
Three point bending tests were performed to quantify the mechanical strength of the laminates which were found to be inferior to those using neat Forton PPS powder..
After discussion with the compounder it was discovered that the CNT/PPS also contained compounding aids, such as stearic acid. It was thought these additives might negatively influence the fibre-matrix adhesion. Subsequent compounds were manufactured without the processing aids.
The next stage of the work was to produce 35 m × 0.5 m wide powder coated prepreg with unfunctionalised CNT/PPS being deposited on both sides of the carbon fabric. The target resin content was 44%.
Laminates were produced using three consolidation cycles these being 10 bar @ 310°C, 20 bar @ 325°C and 30 bar @ 340°C.
The C-scan results indicated that good impregnation could be achieved but the microscopy showed significant crack formation between the fibre bundles. As the laminates cooled cracking noises were heard, the noise being related to the formation of the cracks shown in the micrographs.
The laminates were assessed by three point bending. Four laminates were tested, these being PPS 0205 and 0214 pressed with 30 bar at 340°C and PPS 0205 and 0214 pressed with 20 bar at 325°C.
The results were worse than those obtained previously and this is thought to be due to the crack formation, these being seen in SEM micrographs.
Four laminates were produced with functionalised CNTs, these being pressed with 20 bar at 325°C. The initial visual inspection indicated that the laminate was blacker than had been found with PPS 0205 containing unfunctionalised CNTs. The performance of the laminates containing functionalised CNTs is significantly worse than those containing unfunctionalised CNTs.
Further batches of PPS containing unfunctionalised and functionalised CNTs at varying loadings were converted into pre-preg with resin on both sides of the fabric and a 44% resin content. The laminates were manufactured using two pressing cycles these being 10 minutes at 325°C with 10 bar pressure and 75 minutes at 325°C with 31 bar pressure. The results indicated that whilst the impregnation of the fabrics with the CNT/PPS resin systems was possible, the mechanical results were similar and consistent with those obtained in earlier experiments.
Analysis of the fracture surfaces using SEM shows that following failure, the carbon fibres in the laminates which did not contain CNTs are well wetted with the PPS resin whereas in the case of the CNT containing laminates, the bond between the resin and the fibres appeared to be very poor although some wetting was observed.
The introduction of the unfunctionalised CNTs into the PPS results in crack formation within the laminates. Typically when adding reinforcing fillers to a thermoplastic it would be anticipated that the strength and modulus would increase but the elongation at break would be reduced. With a typical short-fibre-filled compound the fibres would be expected to affect the contraction of the material on cooling, the fibrous filler shrinking far less than the polymer. In the case of the CNT filled materials, the CNT concentration is only 0.5% by weight and so volumetrically it would not be expected to have any significant impact on the contraction of the material on cooling and solidification.
On solidification, the outer layers of the laminate are the surfaces which are directly cooled and so will probably cool and solidify first. Loss of heat from the interior of the laminate is via conduction and so heat loss from the centre of the laminate will be slower. The rate of cooling will also be affected by the heat of crystallisation which is released at the melt-to-solid transition, this slowing the cooling rate. The consequence of this is that the shrinkage due to crystallisation in the central regions of the laminate would be expected to be greater than that found at the surfaces of the laminate. As the central regions of the laminate cool, the outer layers will be cold and possible below the Tg of the polymer, this making the outer layers relatively stiff and inflexible. Inclusion of 0.5% CNT results in approximately 25% fall in elongation at break.
In injection moulding, the typical shrinkage values quoted for Fortron 0205 is between 1.2 and 1.8%. The extension at break found with the CNT reinforced material was 1.6% as compared to 2.1% for the unfilled PPS. The shrinkage of the CNT reinforced material on cooling is therefore of a similar order of magnitude to the extension at break as opposed to the unfilled PPS where the extension at break is greater than the shrinkage. In the case of an injection-moulded component there is usually the scope for the dimensional change to occur thus accommodating the shrinkage. In the case of the laminates, the structure surrounding the polymer will be fairly rigid due to the carbon fibre loadings and thus the surrounding material will be un-yielding and any shrinkage is unlikely to be accommodated by deflection of this material. It therefore seems likely that the result is that the CNT/PPS material fails and cracks appear in the laminate structure.
The mechanical performance of the laminates made with the functionalised CNTs is worse than that with the unfunctionalised CNTs. Reductions in elongation at break can arise for a number of reasons but the two main reasons are that the filler is poorly dispersed or the filler acts as a reinforcing agent for the polymer. In the case of this work, the extensive electron microscopy carried out examining the various CNT/PPS materials indicated that the level of dispersion was generally good and so it seems likely that the reduction in elongation at break is the result of the CNTs reinforcing the PPS. As the reinforcement increases, a reduction in the elongation at break would be expected. In the case of the laminates, the improvement in reinforcement would therefore be expected to result in more extensive crack formation and the poorer performance of the resulting laminate. The inclusion of functionalised CNTs results in a poorer performing laminate than found with the unfunctionalised CNTs and thus it could be construed that their reinforcing effect was improved over the unfunctionalised CNTs.
Micrographs of fracture surfaces suggest that the strength of the bond between the PPS and the carbon fibres has been reduced in the case of the CNT containing materials. The strength of a carbon fibre reinforced polymer is largely determined by adhesion of the polymer to the fibre. In the case of thermoplastic matrices, this adhesion is more of a mechanical nature than a chemical bond:
The thermoplastic resin shrinks around the reinforcing fibre. Thermoplastic resins solidify at high temperatures, so they show a significant volumetric shrinkage, typically around 2%. This large shrinkage results in a volumetric lock of the resin onto the fibres but this may also result in significant residual stresses.
It is assumed that the polymer molecules creep into the micropores on the fibre surface. It has been shown that an oxidised, rough fibre surface gives rise to better mechanical properties of a C/PPS composite.
As discussed, the reinforcing CNTs reduce the strain to failure. The effect of the CNTs on the shrinkage of the polymer was not determined for the current PPS and PEEK polymers. Whilst the CNTs are present at a small concentration it might be assumed that the CNT network decreases the volumetric shrinkage found as the polymer cools and solidifies. Hence, the shrinkage stresses on the fibre are reduced, and therefore the pressure which drives the development of the interface and the final mechanical adhesion of the polymer onto the fibre is also reduced. The effects of a weaker interface and a less ductile matrix reinforce each other as, as soon as the filament becomes detached from the matrix, the local stress concentration increases and microcracks will then grow from this stress concentration to release the residual stress.
A second effect might be that the CNTs reduce the mobility of the thermoplastic molecules in the fibre-matrix interface. The increase in the melt viscosity on the macroscale is limited, but the effects on the micro- or nanoscale are not known. If the CNTs hinder the diffusion of the PPS polymer into the micropores on the fibre surface, then the negative effect becomes stronger with improved dispersion of the CNTs.
Carbon woven fabric reinforced CNT/PEEK
The preparation of the functionalised CNTs resulted in them being attached to PEEK powder particles with a d50 of 10μm. This material was processed using a twin-screw extruder to produce a compound where the CNTs were bound into the polymer. The compound was ground down to produce a powder with a d50 of approximately 1000μm.
The two batches of material, one with functionalised CNTs the other with unfunctionalised CNTs were converted into pre-preg. The materials were coated onto a standard 5H satin carbon fabric, this being the same type of material that had been used with the PPS materials.
The laminates produced showed acceptable C-scan results and did not exhibit the cracking noise found with the PPS materials when cooling down in the press.
The mechanical properties of the CNT containing laminates were again inferior to that of the laminates which did not contain CNTs. The PEEK containing unfunctionalised CNTs produced laminates with a higher flexural strength than that containing functionalised CNTs but the flexural moduli of the materials was similar.
The inference drawn from these initial results is that a similar process is happening with the CNT/PEEK matrix as occurred with the CNT/PPS matrix. However, the ductility of the PEEK materials is better than that of the PPS materials and so cracks do not form in the laminate. However, due to the contraction occurring during cooling and solidification, significant residual stress is set up within the polymer rich zones of the laminate and this results in underperformance of the laminates containing CNTs. The performance of the PEEK containing functionalised CNTs may also be being affected by potential degradation of the resin as discussed earlier.
CNT/polymer UD tape
The manufacture of unidirectional (UD) reinforced CNT/PPS and CNT/PEEK tapes was investigated in parallel to the woven fabric reinforced CNT/polymers. Two approaches were used.
The first approach involved modifying an existing UD prepreg technology based on polymer powder. In this process fibre yarns are pulled through an aqueous dispersion of very fine polymer powder. The powder particles migrate into the yarns. After drying, the polymer powder is melted and a prepreg is consolidated. Any increase in the viscosity of the resin system due to the inclusion of CNTs would only make a small difference to the process as the polymer particles are already amongst the filaments and, therefore, polymer flow during impregnation and consolidation is very limited.
Whilst this approached appeared to have advantages, the principal condition for the dispersion impregnation is that the powder particle size and distribution meet a narrow specification. The grinding of tough polymers down to the required particle size is very difficult and this proved to be a barrier to the application of the process.
The second approach to producing UD tape with CNT/polymer was based on melt extrusion. An existing polymer extruder was retrofitted and equipped with a crosshead die. In this process fibre yarns pass through the die, where the polymer melt is forced in between the filaments. The level of impregnation is determined by the yarn line rate (or residence time), the polymer viscosity and the melt pressure in the die.
Reasonable impregnation was achieved. However, the throughput rate was low, of the order of 1-3 m/min and the residence time in the die at the impregnation pressure was too short. Significant redesign of the impregnation head was required and so this approach was discontinued.
Novel Dry-Powder Impregnation Technique
The development of a new impregnation technique included evaluating the use of an existing D-Preg technology and comparing the products manufactured using the new technology with ‘standard’ material derived through powder coating and film stacking technologies.
In the D-Preg process the polymer is applied in powder form and is forced through the reinforcement fibres via a high voltage alternating electric field. The process has the capacity to distribute powder form materials in all kind of porous structures (nonwovens, fabrics, papers, foams…). This technology works with insulating powders such as thermoplastic or thermoset and conductive ones as carbon black.
Process improvements were made but due to the electrical conductivity of carbon fibre reinforcement, the D-Preg process was found to be inefficient and a new process was developed. The conductive fibres create a local zone without any electric field and so inhibit the powder impregnation.
One other problem associated with the process was the requirement for fine powders with a d50 of around 20-25μm. In the case of granular PPS and PEEK materials, grinding down to this d50 was extremely difficult and expensive.
The carbon fibre conductivity prevents impregnation due to a Faraday Cage effect and the contact between carbon and powder doesn’t allow nebulisation of powder (cloud). The fibre fineness and sizing increase the difficulty of impregnation. The initial stage of the work involved a literature search based on:
dry technologies (no suspension, no solvation)
based on powder (no film, no fibres, no liquid)
suitable for 1, 2 or 3 D reinforcement (not only for UD or tape)
Of the possible routes considered, only the fluidised bed and mechanical vibration were deemed to be suitable techniques.
Trials were initiated with fluidised bed and mechanical vibration technologies. Fluidised bed impregnation resulted in a simple or double face coating without any migration of particle between fibres. As a consequence efforts were concentrated on mechanical vibration technologies.
An alternative technique was developed based on mechanical vibrations at laboratory scale. This process should allow the opening of the fibres and the impregnation by the powder. In a first stage, PA12 powders were used in order to determine the most suitable powder particle size. Particle sizes of 20 µm allowed higher final powder content, but this is still limited to 22% w/w, into carbon fibres.
Other parameters were investigated these including:
Time of impregnation
Gap between vibrators
Amplitude of vibration
All the parameters were optimised to produce a carbon fibre/PA12 pre preg, the properties of which were compared to product manufactured by the simple scattering technique. The flexural properties of samples impregnated mechanically were slightly improved in terms of moduli and strength.
All the parameters were reviewed and optimised to produce PEEK and PPS matrix pre-preg and plates for mechanical characterisation.
The following trends were highlighted:
The maximum powder content (PPS or PEEK in carbon fabric) was limited to 20% w/w
The main parameters were time and amplitude of vibration
Fluidity of the powder is important (too fluid = powder expulsion; not fluid = no impregnation )
After powder addition and compression moulding, composites based on powder coating and vibration were compared using flexural tests, as illustrated in Figure 90.
It was found that the modulus and maximum strengths were similar for all the impregnation technologies.
This result was however, still far from the target of 42% w/w of resin content for the woven carbon fabric.
The development work was divided into five steps:
Powder preparation: with and without anti caking agent (0 to 0.5% w/w in powder)
Impregnation with different parameters along with the characterisation of the materials
Thermal fixation
Compression moulding
Characterisation
The maximum powder content achieved was 36-37% w/w, this being better than in the previous trials. The powder content is proportional to the impregnation time and process parameters. The set-up of ‘opening system’ allowed the improvement of the powder content into the semi-preg. The use of fine powder, d50 = 20 µm, resulted in better impregnation and higher powder content.
It was concluded from the work that powders with a d50 up to 50µm had no influence on the result matrix/fibre ratio. However, PPS+CNT powder with a top cut of 255µm was too coarse to be impregnated in any significant quantity into the carbon fabrics.
The project objectives were not only based on the manufacture of woven fabric materials but also on unidirectional materials (UD Tape). Trials were carried out with the new technology in order to impregnate glass roving and carbon tow.
In order to supply samples to the partners and to validate the potential of the technology on an industrial scale, the impregnation process was upscaled. The new pilot line was supplemented with a tunnel furnace and peripheral devices.
New improvements were also implemented during these trials:
The new system allowed the impregnation of carbon tows at high speeds up to 20m/min
Multiple tows could be impregnated simultaneously
The powder content could be adjusted and was a function of the system parameters
A PEEK content of 35% wt could be reached easily with a very homogeneous distribution of powder between the carbon filaments.
Due to this high level of impregnation quality, the optimisation of moulding cycles in terms of time and pressure was carried out, the results being compared with standard commercial products. In all the cases, the semi-preg produced using the new technology had equal or better modulus than the commercial products.
A few technical limits were found with the new technology, these being:
The impregnation in the warp and weft directions in woven fabric was quite different. Higher resin levels were found in the warp fibres.
The equipment in the line could not achieve the dimensional tolerances required by other partners for tape laying with in –situ consolidation. This was not due to a fundamental problem with the new technology but more a limitation of the equipment that was available and so this could be resolved.
The next stage of the work was to carry out a study comparing existing pre-pregging technologies, such as scattering or film stacking, with the new impregnation technology. This work was based on an unbalanced carbon woven fabric (88% warp 3k carbon fibre, 12% weft glass fibre) impregnated with neat PA11, PPS and PEEK.
In the case of PA11, the new impregnation technology was compared to the double scattering. The products obtained were compression moulded into laminates and were characterised by 3 point bending tests. The influence of the pressure and time for the compression moulding process on mechanical properties was evaluated.
Irrespective of the moulding pressures or times, the semi prepregs produced using the new technology showed better mechanical properties for the same fibre content. At very short moulding times at very low pressures (1 or 3 bars), the improvement was linked to use of the new impregnation technology which resulted in an improvement of 50% for strength and 13% for modulus.
In the case of PPS and PEEK, the new impregnation technology was compared to the film stacking method. The same procedure used previously was followed. In the case of PPS, the new technology generated an improvement of 44% in terms of the maximum flexural strength and 14% in terms of flexural modulus for a short compression cycle of 5 min at 5 bar, the moulding temperature being 310°C.
In conclusion, the new impregnation technology has the following advantages:
Dry technology without the use of any solvents
Fast, one step processing
Better fibre opening and matrix distribution in composites compared to traditional technologies
Better control of resin flow due to the improved powder distribution between filaments
Higher mechanical properties at the same moulding conditions
The use of CNT reinforced materials for the manufacture of semi-pregs/UD tapes was not possible due to the problems related to grinding the compounded materials. Initial work was carried out using uncompounded materials but there was a significant health hazard due to the release of CNTs due to the vibrational nature of the process.
Novel Laminate Manufacturing Methods
Two techniques were investigated, these being internal and external bagging.
Internal Bagging
The internal bagging process has been evaluated using bags made from PEEK film, the bags being produced from flat film using a laser based welding process. In order to weld the bags a black film has been sandwiched between two ‘natural’ films, the black film acting as a radiation absorber.
An inflation adaptor was manufactured for the bags, this fitting into the inflation hole. The aim of the design was to have a system where bags could be supplied in a lay flat form and inflated at the point of use. Following inflation, the necks of the bags were welded and the inflation assembly cut off so that the adaptor could be re-used.
The mould was composed of two box sections with simple clamps to close and lock it. The individual bags were designed to fill one half of the tool in order to ease assembly of the two halves of the mould. The interior of the mould was coated with Frekote 55-NC to prevent the polymer sticking to the mould. The layers of 280 g m-2 5-Harness satin carbon fabric semi-preg were interlocked at the corners.
The layers of pre-preg were generally well bonded in the finished component although the external definition of the cube needed to be improved. The wetting of the fibres was very good when considering that the composite material was a semi-preg where there was virtually no impregnation of the fibre bundles prior to the bagging process. The wetting of the fibre, whilst not being perfect, is very good.
The limitation with the process appears to be the ability to pressurise the bags to a sufficiently high pressure without them failing. Currently levels above 1 bar internal pressure are difficult to achieve and pressures of 5 bar would be the ideal.
External Bagging
The external bagging technology uses a 200mm x 200mm hot plate tool from Surface Generation The hotplate has sixteen active heat/cool zones and uses an air heating system.
Laminate consolidation is good with UD tape but poor with fabric materials. This is due to the low forming pressure which is not sufficient to cause the impregnation of the fibre bundles with a semi-preg material.
Within the project there is a need to manufacture honeycomb panels which have skins. Experiments have been carried out to examine the possibility of in-situ bonding of PEEK film honeycombs to laminates as the laminates are being manufactured. The top surface of the composite component is the PEEK bagging film and when this melts it is possible to apply the PEEK honeycomb in a manner where the honeycomb melts locally at the film surface so forming a bond. Figure 107 shows the method being employed and the results have been very successful. The process was one where the laminate was formed as per the cycle described earlier with the exception that the honeycomb was mounted onto the panel when the PEEK was molten, a load of 0.001N/mm2 applied with the system held at temperature for 30s and then the system was cooled to room temperature.
The Evaluation of the Properties of the Laminates
The laminates evaluated were produced using a variety of techniques, these including:
Direct compression moulding of pre-pregs as discussed earlier in this report.
Compression moulding of the combination of dry carbon-fabric and polymer films, referred to as a film stacking technique.
Manufacture of Laminates by Film Stacking
Initial laminates were produced using 75μm PEEK film, the approach. The samples were produced by compression moulding at a pressure of 7 bar with a heating rate of 2°C/min and a 30 minute hold at 400°C.
The laminates produced using 8 plies with a +45/-45lay-up were of good quality as evidenced by ultrasound examination.
Laminates produced using 10 plies with a 0/90 lay-up and with a PTFE film between plies 5 and 6 were not as good, as measured by ultrasound. A PTFE film was inserted into the lay-up in order to prevent bonding through the thickness of the laminate so that samples could be cut to determine G1c.
Attempts were made to produce laminates using PEEK/CNT films pressed at Victrex but the initial batch of compounded material was degraded. This resulted in films which contained voids and the material viscosity that was too high, degradation involving cross-linking, for impregnation via the film stacking method.
Due to the lack of suitable PEEK/CNT film, suitable materials were sourced from outside the consortium. The films purchased were a reference PEEK film with no CNT which was around 110µm thick, a CNT containing film, NS10295, with 1.5% CNT type 1 which was around 75µm thick and a further CNT containing film, NS10309 film, with 3% CNT type 2 and a thickness of around 91µm.
Using Differential Scanning Calorimetry, the CNTs were found to have no effect on the melting point or the crystallisation of the PEEK. The inclusion of CNTs resulted in an increase in viscosity as shown in Figure 111.
Examination of the films using microscopy indicated that some CNT agglomerates existed.
Laminates
In addition to the materials mentioned previously, two further materials were purchased, these being:
UD tape based on IM7/PEEK, 220g/m² prepreg area weight, 34% resin content by weight, 145g/m² dry fibre area weight
UD tape IM7/PEEK with CNT, 216g/m² prepreg area weight, 34% resin content by weight, 143g/m² dry fibre area weight
A 5H fabric/PEEK without CNT, 280 g/m² dry fabric, 38% resin content was supplied from within the consortium.
Fracture Mechanics
G986/PEEK and IM7/PEEK Laminates
Testing was performed in accordance with Airbus specification AITM 1.0005.
The introduction of CNT into the G986/PEEK film composite slightly reduces the G1c values whatever the testing temperature. No increase was observed in the values of G1c following the addition of CNT as was originally expected. The G1c value for G986/PEEK film laminate is higher than that for the UD tape IM7/PEEK as expected when laminates made from UD tape and fabric are compared. The result obtained with semi-preg G986/PEEK is lower than would be expected and this may have been due to the presence of porosity in the composite.
PPS Matrix Samples and G986/PEEK – Samples 5381-5384
The interlaminar behaviour of laminates based on three thermoplastic matrices was investigated using double cantilever beam (DCB) , Mode I, test and the end notch flexure (ENF) , Mode II, tests. The laminate lay-up was a 12-ply [(0,90)3,(90,0)3]S construction and the matrices were:
PPS: Fortron 0214 [3];
PEEK: Poly Ether Ether Ketone, Victrex 150P [4];
PEEK+CNTs: PEEK Victrex150P reinforced with 0.5 wt% of non-functionalised (“naked”) Nanocyl NC7000 CNTs
G1c values were determined from the peak values from the force-displacement tests.
Compression After Impact Tests
The drop weight hemispherical impactor used was 16 mm diameter and had a mass of 5.5 kg. The specimen is clamped on a base plate with a cut-out of 75mm by 125mm. The specified ratio of impact energy to specimen thickness defined in the relevant standard is 6.7 J/mm, the drop height being calculated based on the average thickness of the samples.
The extent of the damage resulting from the impact was determined using ultrasonic examination and the results are given in Table 22.
The compressive strength data was normalised to remove any effects related to minor changes in the volume fraction of the matrix. The results indicate that, in general, a smaller damaged area results in improved compression performance. However, the data for the matrix containing the CNTs does not fit this pattern and it is thought that this is due to local delamination in the outer layers of the laminate.
Tensile Properties
In the warp direction failure happens perpendicular to the loading direction. PPS 0205 is a higher flow [hf] resin that PPS 0214. The CF/PPS(0205)/CNT delaminates extensively on the way to failure.
Acoustic Emission (AE) measurements and digital strain mapping were also made during the tensile tests.
Shear Properties
Shear properties of the laminates was evaluated using a specimen with a tensile +/-45°, tests being carried out according to the Airbus standard document AITM 1.0002.
There is no difference in the shear modulus generated with different pre-preg materials. However, the shear strength does show a strong dependency on the type of pre-preg used. The results are ranked with deceasing properties from UD tape, to semi-preg and G986/PEEK film.
The addition of 1.5% of CNT in the G986/PEEK composite significantly increases the shear strength of the laminates but the addition of 3% of CNT in the G986/PEEK results in no improvement in the shear strength when compared to the results without CNT. There is however an improvement in shear modulus following the addition of 3% CNT but the increase is less than that found following the addition of 1.5% CNT.
In the case of the IM7/PEEK laminates, the addition of CNT significantly reduces the shear strength but does not affect the shear modulus.
Peeling Tests
The influence of the temperature and time of consolidation for the test specimens was investigated in order to optimise the processing parameters.The addition of CNT to the IM7/PEEK prepreg has only a limited effect on the peeling load.
The consolidation time does not influence the peeling load level. The heating and cooling rate in the autoclave is 2°C/min, the complete heating and cooling cycle taking approximately fifty minutes. The zero minute cycle has undergone the heating and cooling process, the zero referring to the time the autoclave was held at 400°C prior to the cooling stage being initiated. The data clearly indicates that the heating/cooling cycle in itself is sufficient to generate a good bond between the plies. As with the consolidation temperature, the addition of CNT does not result in any significant change in the peeling loads measured.
The Development of Sensor Technologies for Monitoring the Health of the Multi-Scale Materials
A structural health monitoring system based on the swept-wavelength coherent interferometry technique has been successfully demonstrated. Several methods for the surface application on composite materials and also techniques for the integration of distributed fibre optic sensors (DFOS) were tested and evaluated. For surface applied DFOS the investigation of suitable materials mainly focused on the adhesive and the fibre coating material, but also extended to appropriate surface preparation of the composite structures in order to ensure a reliable strain transfer. In order to evaluate the performance of the sensors, DFOS with polyimide coating were applied on the surface of VICTREX® PEEKTM 450G test samples. For integrated DFOS, the adhesive is no longer necessary, but the optical fibre has to withstand the process conditions during the manufacturing of the composite structures. In case of Poly-Ether-Ether-Ketone (PEEK) this means the DFOS have to withstand a temperature of 385 °C and a pressure of 10 bars which makes high demands especially on the fibre coating due to the fact that it is the sole protection of the sensitive optical fibre. Therefore, silica optical fibres with copper-nickel coating CU1300 and high-temperature polyimide coating were chosen for the integration into TenCate Cetex® TC1200 PEEK AS-4 test samples.
The DFOS and the measurement technique have been shown to provide excellent results during fatigue tests containing 10 million load cycles for DFOS applied on the surface and integrated into composite structures. The DFOS were able to detect crack formation as well as determining local strains along the axis of the fibre used. As a result, it has been demonstrated that DFOS are reliable structural health monitoring sensors during their estimated service lifetime for the application on the surface of composite structures and the integration into composite structures under process conditions.
Furthermore, the DFOS were applied to composite engine stiffeners that were developed and manufactured within this project. Based on the FEM-simulation data from CIMNE, the maximum load for the mechanical tests was determined as 200 N. In order to monitor the engine stiffeners, DFOS and displacement sensors were used in the test setup.
The position of the DFOS was determined at the position where the highest strain was expected according to the FEM-simulation.The derived displacement and strain values differ from the expected values from the FEM-simulation which is mainly caused by a significant difference in thickness of the real composite stiffeners compared to the simulated stiffeners.
Carbon fibre reinforced polymers (CFRP) are multifunctional materials and they can be also used as sensors themselves for strain and damage for structural health monitoring. This self-sensing concept has the advantages of being low-cost and applicable to a large volume of structural material with absence of any mechanical property loss. The concept is based on monitoring the changes in electrical conductivity in order to detect the onset, nature and evolution of damage in advanced polymer-based composites. Carbon fibres are highly conductive making the composite also a conductor. Electrical conduction in CFRP can occur by two means:
By current flow in the longitudinal direction along the carbon fibre
By conduction in the transverse and through thickness direction due to contacts between the fibres.
In the longitudinal direction, the conductivity is influenced by the number of conducting fibres while in the transverse direction conducting properties are related to the density of fibre to fibre contacts.
The composites were prepared using UD carbon fibres (Hexcel IM7) at a volume content of 38% in a matrix of PEEK. Measurements were performed on samples with length 60mm, width of 10mm and thickness of 0.15mm.
Three different configurations were used for the measurement of resistance and voltage change. For the measurement of longitudinal resistance a four-wire method was used by applying a fixed current through the two outer electrical contacts and measuring the voltage between the two inner electrodes. The configuration for through thickness resistance measurement involves a current applied in the two opposite contacts and the voltage drop is measured from the same wires. The configuration for oblique resistance measurement has two contacts in the oblique direction that are used for applying a fixed current and the voltage drop is measured from the same wires. In this configuration a second voltmeter is used in order to measure the voltage drop between two opposite contacts in the through thickness direction in the middle of the sample.
Resistance increases monotonically with strain from the minimum strain. The increase of ΔR/Ro is more than 10% at fracture. In the curve corresponding to the second sample a steep increase of resistance at 150s could be observed, this being an indication of fibre breakage.
A slight increase of ΔV/Vo in the oblique direction is observed up to 0.6 % of strain which becomes more profound for higher strain up to fracture. As concerns ΔV/Vo of through thickness in the middle of the specimen a negligible change is observed up to 0.8% strain followed by a decrease up to the fracture.
In all the experiments using different configurations an increase of resistance change upon loading was observed when voltage drop was measured by the same contacts which served for current injection. In general, the increase in the resistance during tensile loading is ascribed to:
The load-dependence of the surface resistivity
Geometrical change
Destruction of the conductive paths such as fibre breakage and the reduction in the contact area of each fibre due to deformation and/or damage.
At low strains the effect of (i) and (ii) are dominant while (iii) is more pronounced at higher strains. The results indicate that measurement in the longitudinal direction is more sensitive to fibre breakage while measurement in the through thickness direction leads to higher ΔV/Vo upon tensile loading.
For the oblique measurement an increase of the ΔV/Vo up to about 5% is observed. After the first load cycle the viscoelastic deformation results in an increase of resistance change at the unloaded state. The above finding implies that there is an irreversible damage during the first cycle that could be attributed to the permanent damage of the matrix. However, the ΔR/Ro is reversible in the consecutive cycles. The resistance after the second cycle is recovered back to the level after the first cycle and a slight decrease of resistance level at the minimum strain is observed. The above behaviour is also observed when the maximum displacement of every cycle is set to higher values. Contrary, the ΔV/Vo measured at the through-thickness direction shows a decrease upon increase of stress. Furthermore, a slight decrease of ΔV/Vo at the minimum loading is observed that may be attributed to relaxation processes.
Summarizing, the results reported above suggest that longitudinal resistance is more sensitive to fibre breakage while through thickness resistance gives higher sensitivity for strain sensing. In addition, oblique resistance changed irreversibly during tensile cyclic loading at various stress levels, which could be attributed to matrix damage.
Development of the DIRIS Honeycomb Structures
The DIRIS (Directionally Reinforced Integrated Stretched Single yarn) honeycomb (European Patent 1950034) is a composite, isogrid sandwich panel reinforced (up to this project) with continuous CFs, along all of its principal directions. The main function of the sandwich core is to provide resistance against in-plane and out-of-plane shear loading.
The preform, which provides the panel with resistance to in or out-of-plane shearing forces and also tensile resistance to forces acting in the plane of the preform, is formed by winding a single UD reinforced tape (typically 2mm x 0.5 mm thick) in successive layers in the directions of 0°,+60°, -60°, thus forming equilateral triangles as shown in Figures 144 and 145. Triangular prismatic cells, made of the same reinforced material, are placed into the triangular gaps created within the preform this being done such that the fibres run in a direction inclined at 45° to the edges of the prism. The triangular prismatic cells are fabricated by winding and simultaneous bonding of the UD tape, using temperature and pressure, with the fibres running at an angle of ±45°. Afterwards the prismatic cells are cut at their precise length and folded at one end so that they form one of the bases of the triangular prism. The bonding of the adjacent layers which are in contact is achieved through the simultaneous heating and pressing of the contact area, causing local melting and bonding of the thermoplastic polymer. Ultrasonic melting-bonding of the tape and of the cells is also used.
The CNT/PEEK tape used was reinforced with 0.5% wt CNTs and 40% wt AS-4 carbon fibre. The CNTs were ‘bonded’ on the PEEK grains as described previously and the long CF and the CNTs/PEEK grains were mixed and placed in the channels 750 mm long and compressed into tapes using the tape thermo-compression machine as described previously.
SEM images indicated that the fibre impregnation was almost complete in some places but in other parts voids were detected. The high pressure and the limited melt shearing along the fibre direction were clearly not sufficient to fully impregnate the fibres.
The multi-reinforced tapes were used to prepare three preforms and then three honeycomb cores. The component parts were joined by ultrasonic welding. The cores were 160.5 mm long, 69.8 mm wide and 12.6 mm thick. They weighed approximately 33 g (equivalent to a density 0.22 g cm-3), the dimension and weight being the same as those used in the previous projects in order to provide a comparison with the earlier results.
Shear testing of the specimens followed the ASTM C 273-00 standard test method. The loading rate was 0.2 mm/min. All experiments exhibited a linear elastic behaviour up to about 3.7 MPa and then yield started. After reaching a well-defined peak, the stress decreases with clear evidence of internal honeycomb destruction.
The observations suggest that the multi-reinforced core shows a higher resistance to in in-plane shear than those reinforced only with carbon fibre. Since the carbon fibre impregnation was partially satisfactory, firm conclusions about the extent of the improvement of the core performance, arising from the additional CNT reinforcement, cannot be reached.
The use of the dry impregnation technology developed within the project was investigated as a route for the manufacture of DIRIS honeycombs by preparing a preform in a suitable plastic mould and impregnating it with PEEK powder. The electrostatic repulsion of the charged fibres results in the fibre network opening, thus making the penetration of the powder particles relatively easy. The distribution of the powder on the fibrous structure is rather homogeneous and the impregnation process rapid (60 m/min). In practice, it was found that the powder penetrated into the preform structure, preferentially at the cross links of the fibre network, which is advantageous for the core strength.
In general the powder covered all fibres but several drawbacks were observed:
The powder diameter should be much smaller than the filament diameter (~ 10 μm) in order to ensure that the filament surface is covered with enough powder and the coverage is homogeneous. However, reducing the powder size to a few microns is an expensive procedure and time-consuming procedure.
The powder was loosely bonded on the fibres and therefore a consolidation process should follow.
The technology could only be applied to glass fibre and was unsuitable for carbon fibre.
This procedure could only be applied to the preform in several steps, this leading to an increased cost. The difficulties becomes more severe when the semi-closed cells are incorporated.
All these observations led to the conclusion that dry impregnation is not yet a suitable method for application in the manufacturing the DIRIS core/panel.
Following the limited results using the dry-powder impregnation technology the concept of toothed rolls was examined. The idea was the production, by two counter-rotating heated rolls, of a sheet with an anaglyph with a periodic geometry like the DIRIS core. The concept was one where the rollers acted as two toothed wheels that never come in contact with each other but instead left a small gap between them such that molten matrix could penetrate around a spreading sheet of fibre filaments placed between the wheels. Obviously the thickness of the core produced could not be constant at the tolerance of the flat thermoplastic sheet but the aim was to attempt to vary within a small acceptable range. The rotation of the rolls was initially numerically simulated from 0° to 9° in order to examine whether a core production similar to the one of the DIRIS architecture would succeed.
Two toothed geometries were initially examined. The first core geometry investigated was ellipsoidal and the calculations indicated that 52 teeth were required on each roller.
Melt impregnation technologies have been developed over recent years and Ticona offer a commercial material, Celstran, impregnated in this way. Molten thermoplastics have a high viscosity which tends to make impregnation difficult and normally the fibre tow is spread in order to ease the difficulty of impregnation. Tow spreading is often brought about using a series of bars in a chamber filled with melt, the fibre running over the bars and so spreading.
The problems associated with the various techniques include:
They are restricted in their ability to spread the fibres.
The process involves the use of relatively high tensions and this, coupled with the mechanical interaction of the fibres with the bars, can result in fibre damage.
There are restrictions in the width of the sheet of fibres.
The analysis indicated that the toothed-rolls could sufficiently deform a thin sheet of fibres and can generate an anaglyph resembling the DIRIS core architecture. The next stage was to examine whether the structure of a sheet of fibres sheet can follow the steepness of the proposed anaglyph and how the filaments can be spread to facilitate impregnation by melt. This requires:
Each filament to bend at a slightly different angle (upwards or downwards), according to the anaglyph.
To fully impregnate each filament with the molten matrix (PEEK or CNT/PEEK).
The first requirement can be met if:
Each filament can slide independently to the adjacent one.
The total contraction (in order to follow the anaglyph) of each filament over a specific length (a period) is equal for all filaments. This means that although the adjacent filaments may bend locally to a different extent, at the end of a specific length they should be equally contracted.
In general, the adjacent un-coated filaments of an un-twisted fibre can each slide upon one another.
A mathematical model was developed which maximised the permissible differential sliding along the axis of filaments in order to enhance the vertical gap between adjacent filaments. The main conclusions from the mathematical model are:
It is feasible to treat a filament in a toothed-rolling process without inducing normal differential strains in both spatial directions of an orthogonal fabric. Therefore the filaments can slide independently in the X or Y directions and the compressed sheet will be in full contact with the teeth anaglyph.
High contraction values can be achieved through this procedure, depending on the period and the amplitude. The filament contraction ratio is a metric of the relative sliding of the filament-resin system with respect to the inertial frame of reference of the die.
High relative shearing can result in both spatial dimensions, depending on the geometrical form of the 2D periodic function. Local shear strains can attain very high values (up to 1200%) in many areas depending on the steepness of the slope. The mean absolute shear strain along the surface can also attain high values (up to about 800%).
The design and production of the tooth-rollers, their incorporation in the pultrusion chamber and the calibration process would require prolonged and extended work. In order to investigate the concept and its numerical interpretation the generation of the anaglyph on two flat surfaces of a mould was evaluated during the present project.
The anaglyph was produced on two prismatic pieces of stainless steel by the use of a CNC OKUMA cutting machine. After polishing the surfaces, a thin orthogonal sheet of glass yarns was placed on the one part and compressed by the other. Although the yarns slid differentially along their axis (X or Y) and were also separated, the filaments within each yarn were slightly affected. Since the aim is to separate the filaments, the steepness (the gradients) of the anaglyph needs to become greater and the process would also be helped by using orthogonal sheets with a smaller number of filaments per yarn.
An alternative is to use two thin UD sheets, placed in an orthogonal manner, one pretensioned along the axis of the (pultrusion) chamber.
This work is continuing and involves a German University.
Recycling of Multiscale-Reinforced Thermoplastics
Most recycling is ‘open-loop’ in that the recycled material does not go back directly into the same end-use. However, in plastics factories it has been common practice for many years to practice closed-loop recycling where such things as sprues and edge trims are ground and fed directly back into the manufacturing process. Recycling within a factory can result in significant cost savings both in terms of raw materials but also in respect of disposal.
The materials being used within the M-RECT project have added value over and above the value of the raw materials. It therefore makes sound environmental, technical and commercial sense to reprocess and reuse these advanced materials.
The polymeric component of the composite is as valuable, if not more valuable, than the carbon fibre and so it is logical that the materials would be recycled into injection mouldable compounds. Because thermosets cannot be reprocessed, the normal route with these materials is to degrade and pyrolyse the thermoset resin to leave the carbon fibre.
Separation of the CNTs and polymers is excluded as a method of recovery on the bases:
PEEK and CNTs are approximately equivalent in terms of cost/kg. Thus there is no financial advantage in separation to recover one component which involves the destruction of the other. PPS has a significantly lower value than PEEK so there may be some commercial advantage to removing the PPS.
Cleaning PEEK and PPS contaminated machinery is best done by thermally degradation. In the case of PEEK, temperatures of the order of 600-650°C are used for this process. If thermal degradation were to be used to recover the CNTs then the temperatures being used would almost certainly result in damage to the CNTs. Thus, the value of the residual CNTs would be very small and recycling the CNTs would result in low value materials which potentially had very poor properties.
The PEEK and PPS could be removed by chemical means but both materials exhibit good chemical resistance and so this approach would not be easy. In the case of PEEK the route would probably be to dissolve the PEEK in concentrated sulphuric acid. This would involve safety issues plus the residue of sulphuric acid would need to be recovered. In principle this would be possible but the costs would be very high.
Filtration to remove the CNTs from the polymers would be very difficult given the dimensions of the CNTs. Extremely fine filters would be required and these would be expected to block quite quickly so making the process very difficult and expensive.
The materials produced in the project are suitable for recycling into injection mouldable compounds. Previous work has shown that APC-2 (unidirectional carbon fibre tapes impregnated with PEEK) can be readily re-processed to produce good quality short-fibre reinforced materials4.
Separation of the individual components would be very difficult to achieve and so it seems logical to reprocess the CNT containing materials directly into mouldable materials. The basis of this document is that the materials produced in the project are suitable for recycling into injection mouldable compounds.
Principles of Waste Collection
Recycling of post-consumer waste introduces two significant problems, those of separation and cleaning. Ideally, any in-factory process should be clean and should also offer the benefit of not having to separate as waste can readily be segregated at source. However, there will potentially be cases where cross-contamination of waste streams does occur and so measures must be put in place to prevent any problems resulting.
Cleanliness
Normally, lay-up of composite pre-pregs occurs in a clean-room facility and so contamination of the materials up to the point of pre-preg lay-up should be minimised. Collection of any waste needs to be in appropriate sized containers fitted with lids to stop subsequent contamination.
Segregation
In the case of pre-pregs there are a number of variants possible in respect of the resin, for example PEEK 90 and 150 grades, and fibres, for example AS4 and T700S, used, and where possible, the waste from each variant should be stored separately.
The elevated processing temperatures for engineering thermoplastics such as PPS and PEEK mean that it is vital not to cross-contaminate them with other thermoplastics and thermosets. Cross-contamination will result in the lower performance polymers degrading and black residues being formed. Good quality recyclate can only be produced from good quality waste.
The basic principle should be to ensure that measures are in place to keep waste stream separate in such a manner as to limit any potential cross-contamination. Effective segregation at source produces a high quality waste stream which in turn produces high quality recyclate.
Transportation
Transportation is an important cost in the recycling process. It is unlikely that any one factory will use sufficient quantities of the thermoplastic composite to warrant the installation and running costs for the conversion of the waste into injection moulding compounds.
Any recyclate supply chain must be sustainable in that potential customers will require stability of supply. It therefore would seem to make commercial sense to have a limited number of sites for compound manufacture, these sites also forming the sales and marketing organisation for the materials.
Size Reduction
Size reduction in the plastics industry is usually carried out using granulators/grinders which essentially consist of sets of rotating and stationary blades between which the material is trapped. Size reduction of the material is usually achieved not by cutting the material as the two blades pass each other but rather by fracture arising from the impact as the material is trapped between the blades.
The styles of machine used for films and for thicker solid materials, such as composite laminates, do differ. Generally film type materials are often converted into compressed pellets following grinding in order to improve their ease of use in subsequent stages of the recycling operations.
In the case of large composite components, the primary size reduction would probably be via water jet or laser cutting followed by size reduction in a standard granulator system. Smaller components could be fed directly into the granulation stage.
Economics
Whilst the quantities of material will probably be relatively small, certainly in the early stages the adoption of the technology, the value of the materials and, more specifically, the value of the product derived from them makes it worthwhile recycling any waste as long as the waste streams are free of contamination. In the case of the PPS composites the base resin cost is slightly less than the cost of the carbon fibre and so recycling may be more marginal than with PEEK where the resin cost exceeds that of the carbon fibre.
Reprocessing of CNT-filled Polymers
Material
Materials containing functionalised and unfunctionalised CNTs were evaluated.
The CNTs were compounded into the polymer and the product ground, the result being PEEK powder particles, d50 of 1000μm, with the CNTs being bound into the polymer.
Rheology
The rheological properties were determined using a Rosand RH 10 capillary rheometer.
It had been intended to measure the rheological properties of the material containing the functionalised CNTs before and after reprocessing. Unfortunately, it proved impossible to determine the rheological characteristics of the as-supplied material as there was evolution of significant quantities of volatiles as the material was loaded into the Rosand RH10 rheometer. This caused the plug of material to extrude out of the filling aperture, so leading to an extended loading time, this in turn resulting in the degradation of the material.
The material containing unfunctionalised CNTs was tested successfully and the results were compared to unfilled Victrex 150G and 450G samples. The CNT compound was based on a Victrex 150 material. It can be seen that whilst the viscosity of the CNT containing material is higher than that of the base resin it is still significantly lower than that of a 450-base material.
Recycling Procedure
The recyclability of the material was evaluated via injection moulding, the material being passed through a Boy 12A injection moulding machine.
The sample moulded was an ISO 1BA test bar. The limited amount of material available necessitated the use of such a small test bar. In the gauge length of the bars, the dimensions were 2mm thick by 5mm wide.
It was noted that there was a slight chemical smell within the machine hopper during processing, the intensity of the smell reducing as the number of recycling passes increased. The chemical could not be identified from its smell but it was not typical of those found when normally processing PEEK.
Mechanical Properties
The injection moulded 1BA bars were tested using an Instron 5969 with a cross-head speed of 2 mm/min. A 25mm extensometer was used for strain measurement and the tests were carried out at room temperature. All the samples failed without yielding and necking.
The results suggest that multiple passes through the moulding machine tends to improve the properties of the material which could suggest an improvement in CNT dispersion.
Incorporation of the CNTs has resulted in an increase in modulus although the tensile strength and the elongation at break has reduced.
Electrical Properties
The electrical characteristics of the samples were measured using equipment supplied by Electro-tech Systems Inc. A model 844 two-pin probe was attached to a model 880 meter. Surface measurements were taken at three points on each specimen, near the gate, in the narrow waisted central region and at the end furthest away from the gate. In all cases the samples were shown to be totally insulating.
Validation of Recycling Procedure in Commercial Components made from Recycled Compounds
It was originally anticipated that sufficient waste CNT containing product would be available for the manufacture of commercial components, the properties of which could be compared against existing components made from existing materials. The components intended for this work were gears and the testing was to have taken place on two state-of-the-art gear test rigs at Victrex.
Unfortunately, the amount of PEEK-like functionalised CNTs manufactured during the project was low and delivery of the materials was late in the project. The consequence of this was that all the functionalised CNTs available had to be used for the evaluation components being manufactured and tested in the project. Limited amounts of PEEK-like-functionalised-CNT reinforced PEEK were made available for the recycling activities but this amounted to less than 500g of material.
A gear weights around 50g and the minimum quantity of material required to mould a set of gears is around 5-10kg. Thus, there was insufficient material to manufacture the test gears and the work was not carried out.
However, previous work has indicated that recycled PEEK based composite materials do perform well when recycled into injection moulding compounds. Thus, it is certain that the materials produced within M-RECT could be recycled successfully into finished products.
Costs of Recycling
The main cost associated with the conversion of scrap material into a moulding compound will be the resin costs associated with reducing the fibre content of the scrap down to ca 30% by weight in order to produce the moulding compound.
As mentioned earlier 150kg of scrap would produce something of the order of 300kg of moulding compound. To produce 1T of recyclate would require an addition of approximately 500kg of virgin resin which, in the case of PEEK would result in a cost of the order of £50k. The final price would depend on the cost of collection, transport, size reduction and compounding but this is likely to be relatively small when compared to the costs of the virgin material. Costs could be reduced by, given that the composite material is scrap, utilising reworked polymer although this may complicate issues around product quality.
Potential Applications for the Recycled Material
The potential applications for these materials are limited by a number of factors:
In order to develop a market, sufficient scrap material has to be available to ensure a continuous supply in the quantities required by the end user. In the initial stages of manufacture this may well not be the case and the only apparent option, that of supplying premium grade non-recycled materials as an alternative at the same cost to cover the shortfall, would be very expensive and incur significant losses.
The material is recycled and initially the on-going quality of the recyclate may be difficult to guarantee with time and as the operation grows, more data will be available on quality and larger supplies of scrap material will be obtainable and so the quality questions should be overcome.
There will probably be a reluctance, certainly initially to use the materials in situations where the mechanical performance is at a premium.
It may be difficult to penetrate markets such as aerospace with these recycled materials due to the extensive qualification processes that are required.
These problems will probably create issues initially with the reuse of the materials and it is likely that they will be ‘downcycled’, that is used in less demanding applications. There will be properties of the materials that will be unaffected, for example the chemical resistance of the PEEK and PPS. Thus there may be applications where the properties of the materials are actually superior to the incumbent materials and, due to the lower cost, a share shift will be possible from the incumbent to the recycled PEEK and PPS materials.
Suitable applications may therefore be as replacements for acetal and nylon in gears, under-hood clips, brackets and cowlings in the automotive business and in cooking utensils.
Numerical Modelling of the Multi-Scale Materials
The software development was at two levels. The first dealt with the numerical modelling of the damping behaviour of CNT reinforced thermoplastics and the second the numerical modelling and optimization of multiscale reinforced thermoplastics.
Numerical Modelling of the Damping Behaviour of CNT Reinforced Thermoplastics
The stages of this development were:
Derivation of the mechanical properties of SWCNT.
Parametric studies of the influence of tube wall thickness, diameter and chirality.
Modelling the Representative Volume Element (RVE) for CNT reinforced PEEK.
Modelling viscoelasticity from frequency dependent test data derived from Dynamic Mechanical Analysis (DMA).
Implementation of a frictional-type bond slip model simulating the interface characteristic behavior between the matrix and CNT.
Simulation of the damping behavior of an RVE subjected to cyclic load.
A SWCNT was been modeled as a space-frame structure using 3D Bernouli beam elements to simulate the C-C bonds, as shown in Figure 159. The mechanical properties of the beams were derived through the Modified Molecular Structural Mechanics (MMSM) approach using values of Young’s Modulus = 5.59TPa and Shear Modulus = 0.871TPa.
The exact size of the wall thickness of SWCNTs was ambiguous and is under investigation as it seems to affect the mechanical properties of carbon nanotubes, as shown in Figure 160. There is a wide scatter of values for it in the literature, ranging from 0.066 nm to 0.69 nm. In the majority of studies, it has been assumed that the wall thickness of CNTs is equal to the interlayer spacing of graphite which is 0.34 nm.
In order to estimate the effect of nanotube diameter and chirality on the elastic moduli of SWCNTs, a series of simulations were conducted in tension and torsion loading. The mean diameter of the carbon nanotubes ranges from 1 nm to 20 nm and two types of chirality were examined, armchair (n,n) and zigzag (n,0). All models had an approximate aspect ratio about 10 (length/diameter) and the value of the interlayer spacing of graphite for the wall thickness simulations was used.
The simulations revealed that the effect of the diameter on Young’s and shear modulus is very weak, especially for large diameter and the chirality as far as armchair and zigzag is concerned has no significant effect.
The C-C bond in the atomistic scale is represented by a 3D Bernoulli beam which is the structural element for the space-frame CNT. The elastic properties of this beam were derived through an MMSM approach. After a series of simulations, where the space-frame CNT was subjected to tension and torsion loadings, the surrogated 3D beam was derived with elastic properties calculated by FE analysis. Using this constitutional element any geometry of the longer (wavy or straight) micro-scale CNT can be modeled. This CNT is embedded into the matrix, so building the Representative Volume Element. The matrix of the RVE is a rectangular solid which is modeled with 20-nodes quadratic brick elements. The material used is the viscoelastic thermoplastic PolyEtherEtherKetone (PEEK) by VICTREX.
The viscoelasticity of the PEEK was done using a Generalised Maxwell model, as shown in Figure 163, which is a spring connected in parallel with Ν Maxwell elements. A Prony series expansion is used to define the relaxation modulus of the viscoelastic material.
In order to investigate the damping behavior of CNT-reinforced thermoplastics, a series of simulations were conducted in RVEs with CNT volume fractions of 1.5 3.3 and 5.8 % when subjected to a cyclic load at frequency 1 Hz. Two geometrical shapes of CNTs were examined, a straight and a wavy-spiral CNT, the latter being the norm for the CNTs as shown in Figure 166. The geometry of the matrix was a rectangular parallelepiped with a constant length of 1µm for all RVEs, and dimension of 30, 20 and 15 nm.
A spectral technique was developed to take into account the waviness of the CNTs and to determine the mechanical performance of the CNT/matrix system. The model indicates that the effect of waviness is to reduce the damping capacity by around 20%, straight CNTs performing better.
An interfacial bond-slip model was developed where the interfacial load transfer is taken into account through a friction-type bond-slip model at the interface between the polymer and the CNT. A critical interfacial shear strength (ISS) is used to define initialization of slippage of the CNT. A procedure was incorporated into a Finite Element code to compute shear stresses, as shown in Figure 168, from normal stresses acting at the end nodes of each CNT beam element. Assuming constant shear stress along the element, the equilibrium imposes:
The use of large RVEs that contained many CNTs imposes computational requirements that are outside the scope of the current project. Relatively simple models would have something of the order of 100000 degrees of freedom (DoF) and more accurate models could have up to five million DoF. The generation of large RVEs would involve random spatial distribution of the CNTs, random CNT axis orientation and random Gaussian length distributions. The use of large RVEs was being explored within the complementary Master project (Mastering the Computational Challenges in Numerical Modelling and Optimum Design of CNT Reinforced Composites, project number 291239, start date 01/03/2012).
The overall damping capacity of RVEs was found to be due to the viscoelastic nature of the PEEK matrix and slippage of CNT inside the matrix, this occurring when a critical interfacial shear stress is exceeded (τc=40MPa). With an RVE with vf≈1.5% the damping was almost entirely due to viscoelastic effects, the slippage having a minimal contribution. The hysteresis effect in the stress-strain loop is increased with increasing CNT volume fraction due to a greater contribution arising due to slippage. The stiffness of the reinforced PEEK is increased with increased concentration of CNTs in the matrix.
The modelling indicated that the stiffness of RVEs with an embedded wavy-spiral CNT was less than that found with RVEs with a straight CNT.
The process of solution involves:
Compute the incremental displacements
Loop over all beam elements and check each element for slippage. When slippage occurs, the axial stiffness is reduced to zero resulting in a local modified tangent stiffness matrix.
Correction of internal forces of the elements and update global force vector.
The microscopic behaviour of the heterogeneous medium is projected to an homogenous equivalent continuum model at an upper scale. The heterogeneous medium is assumed to be statistically homogeneous so that an RVE can be recognised. Micro-macro relations have to be defined and in this work, macroscopic quantities are formulated as volume averages of the corresponding microscopic state. The average of a quantity at the microstructure level is defined as its integral over the corresponding RVE volume V. The averaging theorem requires the strain energy of the micro-structured RVE to be equal to that of the homogenised macro-continuum.
In defining a representative constitutive model for the aligned and oriented CNTs-RC material, which will be assigned to the homogenised model, the following observations have to be taken into account:
Unidirectionality of CNT-reinforcement induces anisotropic stiffness to the CNT-RC RVE material.
Slippage is observed at the CNT-polymer interface region leading to anisotropic energy dissipation mechanism. Hill’s orthotropic plasticity model combined with Prager’s nonlinear kinematic hardening law is capable to accurately predict damping
Stress-strain behaviour of the CNT-RC is subjected to cycling loads.
The six initial yield stress values can be calculated by imposing the corresponding macro-strain to the RVE model with increasing magnitude up to the point where the ISS is violated. Equivalent yield stress value is calculated from the multiaxial stress state. The nonlinear kinematic hardening data are derived by calculating the equivalent yield stresses as a function of the accumulated plastic strains. Tabular data of equivalent yield stresses and accumulated plastic strains are graphically defined from the nonlinear stress-strain curve derived after tension or shear of the RVE model. Note that the nonlinearity is attributed solely to the CNTs slippage.
The macro-strain is applied in the form of imposed displacements on the boundary nodes of the RVE micro model. The macro-stress is computed from the calculated boundary nodal reaction forces in a volume average manner.
The models were employed in the design of a shear testing jig manufactured during the project, an inlet fan blade and the DIRIS honeycomb.
Numerical Modelling and Optimization of Multiscale Reinforced Thermoplastics
This activity aimed to produce numerical tools for predicting the mechanical behaviour of sheets as well as honeycomb core-panel and complex-shape parts produced with long CF reinforced CNT/PEEK and CNT/PPS matrices.
The numerical model was developed and implemented into an in-house finite element code and integrated with the pre and post processor GiD making a single package called COMPACK . The code was applied to the analysis of the mechanical behaviour of straight beams and DIRIS honeycomb panels made of multi-scale reinforced thermoplastics (PEEK and PPS).
The code developed by was validated against experimental results and then applied for the analysis of complex-shape components proposed by the industrial partners in the project.
The approach to the modelling used an “Advanced Serial-Parallel Model” where each phase (MWCNTs, matrix, long fibres) of the composite material has its own constitutive law (elastic, elasto-plastic, visco-elastic, damage…). The final material behaviour depends on these constitutive laws, volumetric participation and morphological distribution of each component and, eventually, on the inter-phase performance. The model is based on the mixing theory. It can be considered as a constitutive equation manager capable of combining two different performances:
Parallel behaviour Components have the same strain.
Serial behaviour Components have the same stress.
The composite is divided in several layers; each one with a different orientation of reinforcement.
The Parallel/Serial Theory assumes parallel behaviour for the fibre axis and serial behaviour for the other axes.
The new formulation divides the composite into the matrix and a new element, shown in Figure 179, resulting from the coupling of the CNTs with the interface material:
Matrix and the CNTs-interface are related with the parallel mixing theory
CNTs and the interface are bonded together with a combination of parallel and serial mixing theories
The stresses in the interface and the CNTs, at both ends, are assumed to be equal (serial behaviour). The load is transferred from the interface to the nanotube in these regions. The central section has a parallel performance: the strains are equal and the CNTs are capable of developing their full strength.
The model was validated for the thermoplastic materials, PPS and PEEK, used in the project. In the case of PPS, the constitutive model was validated with experimental results taken from the literature.
In the case of PEEK, the constitutive model was validated with experimental results taken from the literature.
The effect of temperature and viscosity on the matrix was included in the code.
The finite element software was written in FORTRAN and compiled into the framework of MS Visual Studio 2010. The Code, made available to M-RECT Consortium, was distributed in the form of executable files for 32 and 64 bit computer operating systems. Graphical Pre and Post Processor and the corresponding interface to the Code have also been offered to the M-RECT Consortium.
The Code, PLCD , is an implicit finite element code, which allows quasi-static and dynamic analysis of solid mechanical problems, bi and tri-dimensional, linear and non-linear, in small or large deformations. It also possible to analyse thermo-mechanical and coupled problems.
Pre-processing of the problem to be analysed is done in GiD by definition of the geometry, meshing, materials and lay-up, assignation to structural parts, boundary conditions, imposed displacements, loading history and control parameters. As soon as the model is complete (GiD), a calculation file is generated for PLCD; input data for PLCD are introduced as a text file. Once the analysis is finished, the results are imported into GiD post-processing. Different options for results output are implemented into PLCD: after iterations, converged load steps, displacements, stresses, internal variables (damage, plasticity), etc.
The library of elements available in PLCD was enlarged to include a three-node triangular shell element with nine degrees of freedom (three per node). This new element joined an optimal membrane element with drilling freedoms and a discrete Kirchhoff (DKT) plate-bending element, counting the membrane-bending coupling effect.
Different solvers (Frontal, LU, GMRES, Cholesky) have been programmed into PLCD. Despite that and in order to accelerate the solution process for large three-dimensional systems, the solver PARDISO was included. The package PARDISO is high-performance, robust, memory efficient and easy to use software for solving large sparse symmetric and asymmetric linear systems of equations on shared-memory and distributed-memory multiprocessors.
Concerning non-linear large deformations, there are different approaches that can be used for analysis. In the Total Lagrangian (TL) approach, equations are formulated with respect to a fixed reference configuration that is usually the initial configuration. In the Updated Lagrangian (UL) approach, the reference configuration is the last converged solution. Both approaches are offered in PLCD when analysis is performed using solid (3D) elements. Details on the different methodologies can be obtained consulting classical references Bathe K-J. (1996), Finite Element Procedures. Prentice Hall and Crisfield M.A. (1991), Non-linear Finite Element Analysis of Solid and Structures, Wiley.
The co-rotational (CR) approach is the most recent formulation developed for geometrically non-linear structural analysis. In this method, the finite element equations are referred to two systems: a fixed configuration and a co-rotated configuration. The main advantage of CR formulation is its effectiveness for problems with small strains (1-4%) and large rotations. The co-rotational (CR) approach has been implemented into PLCD for shell elements.
The Code was applied for the analysis of the mechanical behaviour of straight beams and the DIRIS honeycomb panels made of multi-scale (CNTs and LCF) reinforced thermoplastics (PEEK and PPS).
Complex-shape components proposed for industrial partners have been analysed.
Manufacture and Testing of Components
Five components were manufactured to assess performance of the multi-scale materials and to validate the numerical models. These were:
An engine stiffener
An engine oil pan
An air-supported floating beam
An inlet fan blade
Honeycomb
A satellite structure
Engine Stiffener
The objective of this work was to design and manufacture a medium duty engine stiffener for a 8l EU6 engine using multi-scaled thermoplastic composite material in order to improve the engine’s overall NVH characteristics (by improving the damping characteristics compared to the current aluminium design) while guaranteeing functionality of the part (increasing the engine first bending mode frequency).
Several different stiffener designs were developed and analysed using Finite Element Analysis (FEA) methods, based on a PEEK-based composite material, for:
Static stiffener analysis
Road load analysis
Powertrain modal analysis
The final stiffener concept is an 8,5mm thick compound of different layers of both materials with optimised composite layer orientations.
Fabric semi-preg was used in the core of the stiffener in order to ensure drapeability and the main orientation (0°) is in the principal stiffener direction.
A tool for press-forming the stiffener was developed using ProEngineer CAD software.
The stiffeners were fitted to a test engine located at AVL in Graz and the NVH performance evaluated. The engine was connected to an external dynamometer capable of developing a static or dynamic brake torque on the engine in order to simulate engine loads. It was also capable of driving the engine without combustion (motor mode), this being important to analyse the noise emissions of moving parts decoupled from the combustion process.
The engine tests have been performed in the following order for all three configurations:
- Motored engine run-up (500rpm-2500rpm) without combustion
- Engine run up with 400Nm constant torque (500rpm-2500rpm)
- Full load engine run up (500rpm-2500rpm)
The composite stiffeners resulted in a general increase of sound pressure of around 2dNB over the whole engine speed range. Tests involving the insulation of the composite stiffeners clearly showed that they were the cause of the increase in noise when compared to the performance of the existing aluminium design of stiffener.
To investigate the damping behavior of the stiffeners and to identify the local modes of the stiffeners, hammer tests were performed with the stiffeners installed on the engine and for the stiffeners alone, the stiffeners being attached to elastic bands in order to approach free boundary conditions.
The results showed that in the case of the installed stiffener, the damping of the composite is small in comparison to the aluminium whereas in the free test, the damping is much higher. The reasons for this could be the smaller flexural stiffness of the composite stiffener compared to the aluminum stiffener and/or the decreased screw contact forces due to reduced static friction values at the composite /steel interface. The combination of decreased damping and the resonance of a local stiffener mode with a local powertrain mode has been identified as main reason for the increased noise emission using the composite stiffeners.
Numerical analysis indicated that if the material properties (Young’s modulus, shear modulus) were decreased by 10% and the weight of the stiffener was by 10% the modeled properties are comparable to those measured in the engine test. Thus, it seems evident that small changes in material properties do lead to significant changes in the dynamic behaviour of the stiffeners.
The stiffeners were manufactured by the stamping of a pre-consolidated laminate. This process involves re-melting the polymer matrix and it seems likely that the orientation of the fabric plies was changed during the stamping and the thickness may also be affected in specific regions. This would result in a non-optimise orientation the plies and the possible reduction of the stiffness of the stiffener in specific directions. As noted earlier, there were differences in the results of tests carried out on the stiffeners and those determined by FE analysis.
Whilst the noise levels of the composite stiffeners were higher than those found with the existing aluminium stiffeners the mass of the stiffeners was reduced by more than 50% and the functional requirements were fulfilled. The modelling work suggests that increasing the thickness of the stiffeners to 10mm (ca. 25% increase in weight) will improve the performance so that it exceeds that of the aluminium components.
Engine Oil Pan
The oil pan design developed was that from a Volvo MD8 engine, this being the same engine model to that used for the stiffener evaluation.
A number of alternative materials for the oil pan were modeled and compared with the existing steel oil pan which is 2mm thick. The steel oil pan has a weight of 10.5kg.
The oil level in the pan influences the vibrational performance of the pan. Two oil levels were considered, the levels relate to the maximum when the engine is not running and the normal running level. The oil level had a significant impact on the performance of the oil pan.
The use of a standard Victrex injection moulding compound 90HMF40 was investigated. The moulding of the oil pan was simulated using Moldex 3D and it was found that the pan needed to be 3.5mm thick in order for the cavity to fill. The effect of 24l and 17l of oil in the pan was to reduce the mode one frequency from 390Hz to 137Hz and 147Hz respectively.
The use of stiffeners in the pan structure was investigated and found to reduce the mode one and two frequencies.
Two composite constructions were analysed, these being based on Ten Cate R5690/1 PEEK-matrix semi-preg. The two lay-ups were:
[0/+-45/90]s. A total of eight plies and a thickness of 2.48mm
[0/90/+-45/0/90]s. A total of twelve plies and a thickness of 3.72mm.
The production of a composite oil pan used an existing steel oil pan as a mould. Following manufacture, the oil pan was fitted to a test engine at AVL and its function compared to the existing steel oil pan.
Measurements of the airborne sound from the engine showed that there was an increase in noise with the composite oil pan when compared to the steel version.
Insertion loss analysis of the characteristics of the steel and composite oil pans using forced vibration to excite the pans showed an increase of around 10dB in the airborne noise generated by the composite pan. A speaker was mounted onto the oil pans to generate the noise and the oil pan was attached to a steel plate, the noise being measured one meter away from the oil pan surface.
Comparison of the results from the steel and composite oil pans indicates:
The 1m Airborne sound level increase by about 2 dBA with the composite oil pan at almost all engine speeds and load conditions.
The differences can be observed over a broadband frequency range from 800 Hz to 4kHz, this range being the dominating contributor to the overall noise.
The decoupling of the oil pan showed a positive effect at the engine speed range above 1800rpm.
Insertion loss testing was carried out and the measured noise levels were up to 10 dB higher at some frequencies for the composite oil pan with the same noise speaker excitation.
The difference of the insertion loss was due to the differences of the area weight with a factor of approximately four. (sheet steel 2mm = 15,7 kg/m2 / composite with 8 layers = 3,8 kg/m2)
The structural analysis showed that the material damping loss factor is about twice as large for the composite material but this did not compensate the lower insertion loss of the material.
Air-Supported Floating Beam
The beam developed weighed 35kg and could support a load of 300kg with an air pressure of 100mbar.
The beam was analysed computationally and the results compared to experimental loadings. The two give results which are in close agreement. When inflated with a pressure of 220mbar with a point load applied in the center of the air beam, the vertical deflection in this point is 116 mm, this being in good agreement with the model.
The compression element made of PPS/carbon fibre works very well. However, to increase the size of the air-bridge to 15 m long, a new design of composite construction would be required as this length of span would need a very thick plate. A honeycomb construction would also be a possible candidate for increasing stiffness without significantly increasing weight.
Inlet Fan Blade
The primary objective of this work was to design, build and test a multi-scale reinforced fan blade that reduces blade weight by 15%. The blade should have similar quasi-static and dynamic behaviour to real fan blades.
The two main generic styles of blade are:
Civil airliner : >1m length, 0.4 m width and10 kg mass
Military: » 0.4 m length, 0.3 m width and 4 kg mass
The key dimensions of these blades are the aspect ratio, the camber, the twist and the fan diameter. Maximum blade tip velocities are approximately Mach 1.3 and the blade thickness varies depending on the local loading.
Initial test work was carried out using a PEEK-matrix quasi-isotropic laminate and, as the fan blades can reach 150°C in use, testing was carried out at 20°C and 150°C. Testing was be focused on tensile, shear and CTE performance.
Several designs of tab were investigated for the manufacture of tensile test specimens. However, none of the designs yielded test data where the samples yielded and so the design was altered to one which featured a v-notch, where the notch had a 1mm radius. In the v-notch specimens, the failure occurs where it is expected but it is not possible to measure strain or elastic modulus. It is however, possible to determine the mechanical data using FE analysis to model the tests.
Interlaminar shear tests were carried out by introducing notches in the specimens to make them similar to a lap shear joint. Tabs were used to reduce the loading on end of composite and minimise any crushing effects and a guide fixture was used to prevent any tendency for buckling. As with the tensile tests, the strain cannot be measured directly with this type of specimen and so FE analysis was used to estimate the shear modulus
The CTE was estimated from test data to be 28.8µm m-1 ºC-1.
Work was also carried out to evaluate free vibration, fatigue, creep and high speed impact. The latter evaluation was carried out using a piston driven by compressed air to launch a projectile that contained a force sensor. A high-speed camera was used to record the impact and could be used later to determine the displacement time history and velocity of the projectile. The results indicated that the composite based on PEEK with fCNTs had an increase in flexibility just before rupturing.
Using the data measured, a blade was modeled and designed.
Comparing the natural frequencies of the composite blade to a solid titanium blade, analysis using FE indicates that the natural frequencies of the composite blade are relatively close to those of the equivalent titanium blade. However, the lower damping ratio and lower frequency for mode 3 for the titanium blade would result in higher stresses from the harmonic loading.
The total mass of the existing configuration using solid titanium would be ~1.637kg whereas with the composite configuration the mass is ~0.525kg. It should be noted however, that the titanium blade would normally be hollow so the mass would be less than that calculated.
Three blades were manufactured as listed below:
Plain PEEK film with woven carbon fibre (film stacking technique)
PEEK/fCNTs film with woven carbon fibre(film stacking technique)
Plain PEEK/woven carbon fibre pre-preg
The pre-preg blade was found to be stiffer than the others and the damping levels were slightly lower. This could be because the pre-preg results in fewer voids and gives better wetting of the carbon fibres. The expected mass of the blades was 547 grams based on nominal fibre fraction of 50% and accounting for the PEEK core at the root. This is slightly higher than the actual mass of the blades and suggests that some voids may be present.
The two blades made using film stacking had very similar properties although the one with nanotubes was marginally stiffer and in most cases had slightly higher levels of damping.
From the work carried out it is possible to make the following observations:
The manufacturing process produced blades with reasonably repeatable properties. There was some uncertainty regarding the level of wetting of fibres when following the film stacking process.
The mode shapes obtained were exactly those that were predicted from finite element models while the natural frequencies were somewhat lower than expected.
A model with isotropic properties came surprisingly close to predicting the dynamic properties accurately.
Adding nanotubes to the material appeared to increase the stiffness by a small amount.
Damping levels were higher in the composite blades than would be seen in equivalent metal ones. Some additional damping may be due to incomplete wetting of the fibres.
Properties of the blades were very similar around 150°C to those measured at room temperature.
The blade behaviour was relatively linear with respect to vibration amplitude and exposure to extended high level vibration did not cause damage.
In general therefore, it could be concluded that the blade design and manufacturing process was successful although some improvements could be made to the method used for consolidation.
Honeycomb
The objective of this work was to develop SILICOMB cellular materials to allow for the structures to morph for use in space antennas.
The work necessitated the development of mechanical, thermal and dielectric models for the SILICOMB cellular configuration. The mechanical models were formulated through a rigorous Finite Element homogenisation technique based on the theory of periodic structures. The dielectric and thermal models for the SILICOMB honeycomb were developed using variational principles. Direct application of the use of PEEK Aptiv films in the models made it possible to predict the multiphysics properties of PEEK-based SILICOMB configurations.
PEEK SILICOMB cellular plates have been manufactured and tested for in-plane loading conditions. The plates were manufactured through laser-jet and water-jet cutting of thermo-formed PEEK Aptiv films with two different thicknesses (0.25 mm and 0.6 mm). The SILICOMB PEEK plates were tested under uniaxial tensile loading along the two orthogonal directions, and their elastic and failure modes measured. Good comparison has been observed between the experimental and numerical models developed.
The method of manufacture of the honeycombs, based on Kirigami techniques.
Thermal ageing was been carried out on APTIV PEEK films and they were shown to have a stable
performance.
The compressive behaviour of the Silicomb geometries was found to be nearly twice as good as that of the equivalent commercially available Nomex honeycombs. It was found that the normalised shear modulus was comparable with HRH-10 and Flexcore and that the normalised shear strength was 3-4 times higher than the competing honeycombs.
Panels were made by bonding the honeycomb onto PEEK composite skins and the properties of these panels was evaluated.
Three of the SILICOMB cores were produced with PEEK inserts and the other three with PEEK+CNTs inserts. The inserts are all 10x5 mm and the average thickness of the PEEK and PEEK+CNTs films used for the inserts were 0.22 mm. The nominal size of the SILICOMB cores in plane was 76x76 mmx15 mm thick with the skins being larger than this. The flatwise compression tests were performed according to the ASTM standard [3]. The loading was in displacement controlled mode, at a rate 0.5 mm/min.
The results show that the maximum compressive stress for the sandwiches with plain PEEK inserts is a slightly higher than for the sandwiches with PEEK+CNTs inserts: 1.18±0.015 MPa against 1.11±0.04 MPa. However, the residual strain characteristic the sandwiches with PEEK+CNTs inserts is better with a lower residual strain than for the sandwiches with plain PEEK inserts: 0.031±0.0075 for sandwiches with PEEK+CNTs inserts against 0.042±0.020 for the reference group.
The honeycomb system was further developed with SILICOMB corrugations for multistable applications. Finite Element nonlinear elastic models to simulate the snap-through and bistable behaviour of the plates for different curvature radiuses were developed and plates with SILICOMB corrugation and curvature being manufactured the corrugation and curvature being generated during a thermoforming process. A rig was built to investigate snap through on curved panels and a rig was developed for the manufacture of such curved panels, typical panel dimensions being 200mm x 168mm x 15mm thick with a radius of curvature of 90mm.
Curved structures were tested and the flattened structures showed a recovery to within 0.2% of the original vertical dimensions on release of the compressive loading. The behaviour of the curved panels was modelled and the results were in close agreement with the experimental measurements although the experimental samples showed a higher linear response and a lower stiffness hysteretic return.
Numerical analysis showed that:
The equivalent effective height was inversely proportional to non-dimensional radius and was independent of the rib lengths.
Radial stiffness followed a similar dependency over internal cell angles as per flat SILICOMB configurations.
Radial stiffness was approximately inversely proportional to density, variations over radius and internal cell angles.
Work was also carried out investigating curved corrugated plate structures.
Numerical analysis, performed through parametric analysis of these types of structures had identified the fundamental characteristics in nonlinear deformation of such corrugated PEEK plates.
A prototype curved SILICOMB PEEK honeycomb with an elastomeric zero-ν skin for morphing applications was manufactured. The skins were attached with a 3D knitting technique.
The panel showed large hysteresis during quasi-static cyclic loading and it was found that the energy dissipated per unit volume was strain dependent.
Copper skinned sub-reflectors were constructed for testing and evaluation in the Electronics Department at the University of Bristol.
The snap-through resulting from the use of bistable honeycomb systems in antenna meant that the antenna was easily reconfigured. A number of antenna designs were modeled using the performance of the bistable honeycombs developed and it was found that configurations of antenna with R=2500mm and γ = 13° and 15° had displacements at snap-through which was compatible with Ka bands
The significant results arising from this work are:
Production and testing of cellular structures in PEEK. To the best of knowledge, no other cellular structures in the thermoplastic material have been developed/disclosed in open literature.
Development of multiphysics models for SILICOMB PEEK structures which predict, in a robust manner, the overall performance of the structures
Development of the first case of PEEK-based plates with multistability
Development of a manufacturing technique to produce SILICOMB honeycombs at industrial prototype scale in PEEK.
The overall flatwise properties of the SILICOMB PEEK honeycomb, when normalised against Young’s modulus and compressive buckling stress, are significantly higher compared than other advanced honeycombs for high-end performance.
In the final stages of the project, the first SILICOMB honeycomb was made entirely out of CNT reinforced films although it did not prove possible to evaluate this honeycomb.
Satellite Structure
The support structures that were designed and manufactured were modular structures where a module, a panel-resembling the DIRIS panel, is used repeatedly. For each structure, the size and the distribution of the cells were optimised numerically, attempting to address the requested structural characteristics.
Only structures of one and two modules were experimentally investigated, since it was assumed (and numerically confirmed for compressive loading) that key parameters of their performance are representative for structures with several modules
It was found from the experiments that the reinforcement of PEEK by CNTs is addressing only the requirement of increasing the buckling load by increasing the E-modulus and the bonding strength as a result of improving the thermal conductivity of the material.
Another observation associated with the compressive testing of the module, is that its yield in compression occurs at a much lower stress level than the anticipated strength derived by adding the contribution of all the cells of a module. It should be noted that the bonded preform laterally supports the cells from all sides and therefore the buckling load for each cell should be higher than the one without support from the preform. An explanation of this observation could be related with the cell geometrical imperfections which may cause premature buckling to a few cells, which then leads the surrounding cells to buckle and finally the module to yield.
The structure with two modules was subjected to a dynamic, multiaxial, compressive–tensile fatigue loading up to 4.2 x 106 cycles. The minor changes to the actuator motion recorded during the testing support the argument that the mechanical degradation of the structure was limited. The structure was also subjected to cyclic compressive tests up to the yield, before and after the fatigue testing. The stiffness was not reduced during fatigue testing. Also, the measured stress-strain relations confirmed visco-elastic behavior.
The logarithmic decrement and the hysteresis loop methods were used to measure the damping parameters of the structures. Single-layer modules and two-layer modules were tested in a cantilever beam configuration by applying an impulse force (always in the X direction). The structure response at the free end was measured using two accelerometers.
The loss factor measured using the hysteresis loop method was very similar to that determined via the logarithmic decrement method. There was a small difference between the results and this could be attributed to the differences in the generated stress-strain amplitudes at micro-structural level. Damping increases with increasing frequency due to the viscoelastic behavior of the PEEK/CNTs material.
The measured natural frequency (in the X direction) of the DIRIS panel and the core are high enough and outside the usual range of vibration of transport systems. When the cell size increases and also the cell distribution in the structure is changed such that there are more gaps, the natural frequency decreases. This offers the possibility to tailor the structure design in order its natural frequencies to coincide with the vibration of the system and therefore maximising the energy absorption.
Life-Cycle Assessment of a Multi-Scale Reinforced Engine Stiffener
Life Cycle Assessment (LCA) has been used to compare the current design of engine stiffener used today with an alternative produced in composite materials to see if the environmental impact could be reduced. The function is to stiffen the engine-gearbox assembly, i.e. increasing the first bending modes to a level where prop shaft excitation is avoided. The current version is made from die-cast aluminium and the composite version was developed in polyetheretherketone (PEEK) with woven carbon-fibre fabric with and without CNTs.
In this study, the LCA results were characterized by Global Warming Potential (GWP), Acidification Potential (AP), Ozone Depletion Potential (ODP) as well as environmental indices expressed in the
EPS 2000 method (Environmental Priority Strategies).
The LCA method adopted was based on ISO 14040:2006 and based on a medium duty truck with an 8 litre engine which is expected to cover 750000km in a lifetime of 15 years.
The production phase consists of raw material production and parts manufacturing. For the raw material production, the inventory data (in- and outflows of various resources, emissions and energy flows) as well as the transportation of the raw materials were included.
For the manufacturing phase, the available dataset from the GaBi software was used.
The use phase included production of diesel and urea which are consumed during the truck’s lifetime, the exhaust emissions and maintenance. Maintenance covers the materials consumed such as transmission oil, engine oil, coolant etc. and replacements of spare parts over the truck’s lifetime. However, no maintenance is expected for the engine stiffeners themselves. Hence, the maintenance phase was excluded from the study.
Generally, as regards the EoL of the truck, 100% of the steel, aluminum, copper and 97% of the scarce metals (e.g. platinum and palladium) are considered to be sent for recycling; 80% of gold, silver and stainless steel are considered to be sent for recycling due to lack of data for recycling process. Others are set as landfill. For the engine stiffeners only the recycling of the parts themselves will be considered.
The following activities were excluded (cut-off criteria) in this study due to the activities being assumed to be insignificant or no data was available.
Surface treatment of the engine stiffeners
Production, use and EoL of packaging materials;
Manufacturing data of the metals and polymeric materials, which are not included in GaBi and where no other data was available.
No maintenance of the engine stiffeners are expected.
Waste water treatment and waste treatment are not included in the materials’ inventory data.
Transports of raw materials and between suppliers and sub-suppliers as the part is still in a pre-industrialisation phase and such parameters are unknown at this stage.
The results show that the weight of the engine stiffeners has a significant effect on the use phase because of the reduction in fuel consumption. This means that the lower weight of the composite engine stiffeners significantly reduce the overall environmental impact. The use of carbon fibre in the composite has a significant impact comparable to the use phase for most environmental impact categories as they significantly reduce the weight of the composites in comparison to systems based on such materials as glass fibres.
Thus the engine stiffeners with a reduced weight would be advantageous as long as the other property requirements of the part(s) are fulfilled, for example those concerning mechanical, physical, chemical, noise, vibrations, fire resistance, crash resistance performance.
The following recommendations may be applicable for vehicle applications where the fuel consumption is very important
Reduce the weight of the parts as much as possible without compromising the requirements with regard to other properties (mechanical, physical, chemical, noise, vibrations, etc)
Is it possible to find alternative raw material sources if the impact of the current raw material is high? For example, the production of carbon fibre.
Is it possible to redesign part(s) so that more parts can be integrated in a larger part that may be considered for recycling?
Risk Analysis
Risk assessment is usually based on a structure where the ranking of the level of overall risk being based on the product of the likelihood and the impact of an event.
The different risks which occur in this type of project include:
Technological risks
Partnership risks
Market risks
Legal risks
Management risks
Environmental/regulation/safety risks
The risks were assessed by the participants and the major risks were determined and are presented in the Table below.
Major Risks Solutions and Actions Comments
Significant dependency on other technologies. The materials developed within the project depend on the successful functionalisation and incorporation of the CNT into the laminates.
To improve the incorporation of the CNTs within the matrix, laminates were manufactured with a novel impregnation technology (patent application was proposed). Problems were faced in the manufacture of the PPS laminates when CNT were incorporated due to the formation of cracks.
The functionalisation of CNTs do not reach the expected target Technologies for the successful functionalisation of CNTs were developed. Patent application for a route to the chemical functionalisation of the CNTs
Industrialisation at risk: an industrial partner leaves the market This risk was considered to be a high risk during the project. But the correct decisions were made by the participants which minimised the risk. The project started using Baytubes supplied by Bayer as these were in a “granular” form and hence they were advantageous from the health and safety perspective. At 6M of the project one partner started to use the Nanocyl CNTs. After agreement of the consortium the Nanocyl materials were chosen even if more precautions were required for the handling. 6M before finishing the project Bayer announced that they are no longer to produce Baytubes, so the coordination actions to switch to Nanocyl material was, retrospectively, be considered to have been a good decision.
Cultural differences and mode of operating of partners cause delays Coordinating actions were successfully developed to improve the communication between the partners.
Worthless result: performance lower than market needs The benchmark performance of current materials in the market is known.
This risks show that the tensile properties of the CNT containing laminates exhibit inferior properties to those without CNTs; but the fracture toughness of the PEEK/CNT laminates show promising results. For the new technologies to be acceptable in the market, their cost-benefit ratio must be good enough but the market must feel the need for them or must be able to see new regulations being prepared
Nobody buys the product. Too expensive. Development of a Life Cycle Cost Benefit Analysis to determine the benefits of the implementation at industrial level. It was shown by the life cycle assessment carried out that the engine stiffeners made out PEEK 37% Carbon fibre and 0.5% CNT had 56% less weight than the engine stiffeners made out of die-cast aluminium. The reduction of weight by using composite materials will imply a decrease of the fuel consumption during the life time of the vehicle showing a positive environmental and social impact.
Inadequate or ineffectual communication among partners. Coordinating actions were successfully developed to improve the communication between the partners
Lack of awareness in the use of personal protection when handling nanomaterial The handling of nanomaterial in a safety lab should follow the recommended practices for laboratories.
Lack of awareness in regards to the stability of the nanomaterial contained in a solid matrix, during use and mobility. Engineering controls should be applied when handling the final product in the laboratories and further studies should be carried out for analysing the lifecycle of the nanomaterial in the solid matrix. However it is known that once the nanomaterial are embed in the solid matrix the health risks are low.
The environmental hazards of the final product are unknown Preparation of the SDSs according to the format provided by REACH and CLP regulations. The preparation of SDSs will imply the full characterisation of the product as well as the identification of the health and physical hazards.
Main Identified Risks
The information developed during the project included information concerning:
Practical approach for managing safety of nanomaterials in a lab
Material Safety Data Sheets (MSDS)
Possible Hazards for the life cycle of Nano-Materials
Scientific research about Nanomaterials
Guidelines for Carbon Nanotubes, Nanofibers
Recommendations
Exposure Assessment
Recommended Exposure Limit (REL) for CNT
Engineering Controls
Respiratory protection for CNT
The different risks posed in M-RECT project were evaluated by the consortium and the partners evaluated 11 out of 52 (21%) risks as high or very high. The high risks faced during the project were successfully managed by the coordination activities, different solutions and actions were identified.
One of the highest concerns raised by the partners was the lack of awareness in regards to the safety aspects, for example in the health and physical hazards of CNT, in the use of personal protection devices and safety controls and in the hazards of nanomaterials embed in a solid matrix. As a consequence, guidelines were developed for managing safety with nanomaterials, especially for CNT
It is important to note that several studies have investigated the toxicity of carbon nanotubes (CNTs) in experimental animal studies. The results from these studies indicate potential respiratory health risks from exposure to CNTs, including granulomatous pneumonia and fibrosis. Evidence also indicates that when multi-walled carbon nanotubes (MWCNTs) are administered intraperitoneally to mice, the MWCNTs have asbestos-like pathogenicity. Although a causal link has not been established, there is concern about possible cancer hazards in addition to potential for fibrosis/non-malignant respiratory disease.
Assessment of Uncertainties
All measurements are affected by errors coming from different sources e.g. the measurand, the measurement instrument, the measurement procedure, the environmental conditions, etc. Therefore, it can be assumed that the result of any measurement is affected by an uncertainty, which is, an interval of values inside of which will be a true value of a quantity.
The work involved the calculation of uncertainties in material characterization and a sensitivity analysis dealing with the possible aberrations from nominal product characteristics and included a web based resource dealing with the test methods for the determination of such things as the tensile properties of plastics.
Data from partners was collected and initial uncertainty calculations developed by following the steps:
Identify the measurand, and the quantities to be reporting
Identify the influence factors: sources of uncertainty i.e. test piece, testing system, environment, test and measurement procedure.
Classify the uncertainty types as below:
Type A:
Evaluation is by calculation from a series of repeated observations, using statistical methods.
Type B:
Evaluation is by means other than those used for Type A evaluation. For example, by using data from: calibration certificates, manufacturer’s specifications, previous measurement data, experience with the behaviour of the instruments, and all other relevant information.
Estimate the standard uncertainty and sensitivity coefficient for each source of uncertainty
Estimate the standard uncertainty and sensitivity coefficient for each source of uncertainty
Calculate the combined uncertainty
Calculate the expanded uncertainty, U, by multiplying the combined uncertainty uc calculated in Step 5, by a coverage factor, k. For a normal probability distribution k=2; this corresponds to a confidence interval of 95%. The expanded uncertainty U is defined in the GUM as “the interval about the result of a measurement that may be expected to encompass a large fraction of the distribution of values that could reasonably be attributed to the measurand”.
Report the value
Data from a shear test (+/- 45° tensile test) on composite laminates underwent preliminary analysis, and the uncertainties calculated provided an overview of the calculation for 6 samples of the Type A uncertainty. The calculation of the uncertainty Type A is done by the following steps:
Determination of the mean value and the standard deviation of the measurand
Determination of the confidence region of the mean value by using t factor of student’s distribution at 95% of confidence and n-1 degrees of freedom
For the combined uncertainty of the maximum stress (σ= F_max/(2*w*t) ) a sensitivity analysis of the parameters was performed by the use of tornado diagrams. The calculations indicate that for the Type B uncertainties of the σmax, the parameter which has the highest influence is the maximum force (Fmax) in comparison to the width (wo) and the thickness (t0). This is also confirmed by consideration of the influence of each variable in percentiles to the σmax, each percentile indicating the value below which a given percentage of observations fall. For example, the random data from Fmax disturbs the final uncertainty for σmax, in a way that, 95% of the uncertainty falls below 2.16 MPa and 5% of the uncertainty falls below 2.12 MPa.
Estimation of uncertainty allows the comparison of equivalent results with other laboratories, and with values given in specifications or standards. A standard methodology for determination of uncertainties in materials characterisation was applied within the project.
It is important to note that when developing new materials if the mechanical properties are worse than the standard materials and if the certainty of the results is lower, a low benefit is expected. In the case, when properties are better compared to the standard materials, a high benefit is expected only when the certainty of the results is also high.
Surface Coatings
Most polymers are susceptible to degradation as a result of exposure to UV radiation. The chemical structures of PEEK and PPS result in significant absorption within the UV spectrum and so measures are needed to prevent/inhibit/reduce any degradation mechanisms that may result from this absorption.
Whilst additives can be used in the polymers, this involves compounding with the subsequent problems related to powder manufacture.
Thus, the chosen route for UV protection of structures was by applying coatings, the coatings needing to have high levels of adhesion to PEEK and PPS. This was difficult due to the nature of the polymers, the difficulties being compounded by highly fluorinated systems having very limited adhesion capability. The research carried out was directed at introducing a suitable primer layer or using a suitable polymer matrix to support the aluminum platelets used to generate the desired coating properties. In the case of the aluminum platelets they are automatically orientated parallel to the surface during solvent evaporation and this builds up a uniform reflective mirror.
It was envisaged that the total coating would be thin (< 50 μm) and the bonding to the substrate would be characterized as chemical, rather than physical, bonding.
The test specimens were cut out of selected standard PEEK films and composite laminate supplied by partners in the consortium.
Various developments and tests were carried out to assess the coatings developed for the protection of the PPS laminates and these included:
Optimized polymer / hardener ratio
Optimized additive content – new formulation
Qualified layer thickness
Qualified drying / storing and curing data
Qualified vibration test results
Qualified industrial application procedure (spraying)
Bonding qualified (grid cut and pull off)
Chemical resistance qualified
Protective properties of coating qualified
The optimised formulations described as numbers RK 022 and RK 023 passed all tests, especially chemical resistance and bonding. Most important in this application is resistance to gasoline and benzene, as these were considered to be challenging reference chemicals for motor/truck applications.
Various developments and tests were carried out to assess the coatings developed for the protection of the PEEK films and laminates and these included:
Optimized polymer / hardener ratio
Optimized additive content aluminum pigment – new formulation
Qualified layer thickness
Qualified drying / storing and curing data
Qualified bending test results
Qualified industrial application procedure (spraying)
Bonding qualified (grid cut and pull off)
Chemical resistance qualified
Protective properties of coating qualified after special 3000 hrs UV impact and 100 KGray gamma radiation impact
0% transmission of light UV and 70 % of Reflection after application of intended load including UVC load
The complete PEEK –coating testing “Phase II” resulted with formulation V 12 showing the best technological and application performance. The aluminium-pigmentation is a mixture of a so called conventional “non-leafing Alupigment” and a high reflective leafing “bronze” pigment, which achieved a reflection of ~70 %. The transmission measured between 250 nm and 700 nm was 0%. Formulation V 12 / V 12.2 showed the best behavior in bonding, flexibility, UV resistance and gamma resistance.
Durable UV and heat resistant coatings were successfully produced and evaluated for both types of composite matrix.
Potential Impact:
The main outcomes to the project are related to potential economic and environmental impacts.
The life-cycle-analysis carried out in the project clearly indicates that the major factor in respect of environmental impact is the ‘use’ phase. Reducing the weight of structures through the use of composite materials generates significant reduction in the impact of the ‘use’ phase on CO2 production and resource usage.
Fuel economy is directly related to the weight reduction and the reduction in fuel usage will have both environmental and financial impacts. Current designs of composite bodied aircraft are prone to increased levels of NVH due to their construction. The use of additional materials for damping purposes within these applications leads to increase in weight and so the development of smart multi-functional materials is important.
Information from the aircraft industry on the effect weight savings suggests the following average yearly fuel burn per kg weight carried in the table below. These savings have a considerable effect on the running costs of an aircraft and the environmental impact that the aircraft makes.
Aircraft Type Annual Fuel Savings per kg
Airbus A310 200kg
Airbus A330-200/300 145kg
Boeing 777 160kg
CNT Functionalisation
The development of the functionalisation processes for the CNTs has resulted in:
• Large increases in the efficiency of functionalisation both in respect of the times required and the quantities of chemicals used and their recovery.
• The development of novel reactor and mixing equipment that is likely to also have potential applications in pharmaceutical manufacture.
• Developed technologies where CNTs, which are initially loose and fluffy and so present a health hazard, are converted into functionalised CNTs which aggregate, so reducing the potential health hazards. This aggregation does not hinder the dispersability of the CNTs in the polymer matrix as would normally be expected.
• The chemical route developed for the PEEK-like functionalisation of CNTs is the subject of a patent application WO 2013/190162 filed on the 20th June 2013.
Once combined with the matrix material, the CNTs are fully encapsulated and so pose no safety hazards.
The outcome of the project in relation to the use of functionalised CNTs was not as expected, the materials containing functionalised CNTs tending to exhibit poorer performance when compared to those containing unfunctionalised and no CNTs. However, the results do indicate a clear impact of functionalisation and has indicated the route forward in order to improve their performance.
Pre-Preg Manufacture
The novel impregnation technology developed has significant impacts on the economics and safety of the impregnation process. The benefits include:
• Higher manufacturing outputs for UD tapes. Typical line speeds with dispersion based processes are <10m/min. The new impregnation technology has successfully run at 50m/min, the line speed being limited by ancillary equipment, not the fundamental nature of the process. This increase in speed was not to the detriment of the quality of the tapes produced.
• Fabric pre-pregs are usually produced by powder scattering technologies and so the level of impregnation of the matrix into the fabric is relatively poor. This necessitates the use of high pressure processes to convert the pre-preg into finished products. The development work in the project has generated fabric pre-pregs which are very much better impregnated than the normal materials produced by powder scattering. Work is still required to produce a fabric which is fully impregnated but this technology would offer:
o The ability to use low-pressure processes such as oven consolidation for the manufacture of parts.
o This in turn would reduce any capital expenditure required to establish production lines for the conversion of these materials.
o The outcome would be faster production at lower costs, both revenue and capital.
A patent application, FR1361523, was filed on the 22nd Nov 2013 for the impregnation process and the beneficiary is currently having discussions with seven companies for validation trials (three are French and four are European/worldwide) concerning the exploitation of the technologies developed.
Component Manufacture
Novel methods of component manufacture were investigated during the project which could lead to:
• The manufacture of hollow components without the need for extractable bladders using oven heating. This would result in:
o The reduction in capital expenditure required to initiate such processes
o Faster uptake of such technologies and the materials being used in the project
o Stronger, stiffer component structures that will result in the ability to use lighter structures so improving efficiency of aircraft and automobiles.
• The possibility of bonding elements together during the primary manufacturing process, for example bonding honeycombs to skin panels. This will reduce the production steps required and so speed up the manufacturing process and reduce the cost of manufacture.
The technologies developed will result in the faster manufacturing of composite panels this resulting in higher output rates from lower cost equipment so reducing the overall cost of manufacture.
Recycling
Minimising waste and maximising the use of resources is important both economically and environmentally. The materials can be re-melted and reprocessed as many times as it is required without any deterioration of their physical and chemical properties. The project has clearly indicated the recyclability of the materials developed, identified potential markets for the recycled material and the methodology for waste collection. In most cases, it appears that the waste material could be ‘up-cycled’ into high-end applications yielding significant economic, as well as environmental, advantages.
Recycling was not considered during the life-cycle-analysis and so this would further reduce the environmental impact of the materials.
The recycling process did not involve the degradation of one or more components of the waste material and hence is far more environmentally friendly than the processes used for thermoset composites. In the case of the latter materials, the waste is chopped up and pyrolysed to degrade and drive off the resin component so as to leave the carbon fibre. This process involves significant energy usage in order to elevate the temperature of the waste to the point where the matrix degrades and also results in the evolution of significant quantities of waste gaseous by-products from the reaction which may be toxic. The proposed recycling methodology did not result in any similar issues.
A similar situation arises at the end of the service life of the components in that they can be readily broken up and recycled, possibly ‘up-cycled, and so they should not be treated as waste but as a valuable resource.
As with the recycling situation, the CNTs contained within the multi-scale composites pose no health hazard as they are bound into the matrix material. The matrices used are ductile in behaviour so the likelihood of the release of CNTs on failure of the components is highly unlikely.
Development of Numerical Models
The numerical models developed to predict the behaviour of the multi-scale composite materials being developed were shown to be very accurate.
The ability to model the materials and structures accurately will result in:
• Shorter design cycles so allowing industry to respond more quickly to any required design changes
• Reduce the development time for new components and also, by coupling systems, finished structures such as automobiles and aircraft.
• Reduced cost of development and manufacture by being able to design on a ‘right-first-time’ basis. Many problems are designed into structures and good design tools are essential in avoiding these sorts of problems.
The model development work has resulted in a further EU project, Mastering the Computational Challenges in Numerical Modelling and Optimum Design of CNT Reinforced Composites, project number 291239, which started on 01/03/2012.
Software for the determination of the performance of multi-scale reinforced composite materials is now commercially available.
Validation Components
Engine Stiffener and Oil Pan
It has been observed that most of the engine noise in a modern automobile comes from internal engine covers such as oil pans and valve covers. These covers are excited in the acoustic relevant frequency range between 500 Hz and 5 kHz. Passive and active damping control solutions have been already tried on these components but the results have been poor so far since their cost-effectiveness, despite their technical merit, is usually low.
The engine stiffener and oil pan manufactured and tested during the project did not perform as well as expected in terms of noise, vibration and harshness (NVH). Whilst the components developed during the project did not lead to the anticipated reductions in NVH, the results obtained provide a good engineering basis and understanding for the further development of components which will result in such reductions. This should result in a direct social impact resulting from noise control and pollution, particularly in automotive applications, especially in urban areas and highways.
It should be noted though that the reduction in weight of these components as compared to the aluminium and steel counterparts is clearly advantageous in terms of the environmental impact of the solutions.
Air Supported Floating Beams
Market demand for new light-weight and highly resistant bridge structures based on flexible light-weight materials is constantly increasing and related manufacturing processes and technologies are rapidly changing. In 2002, the South African Parliament dealt with a budget of over 3 million € for the acquisition of a temporary stock of emergency bridges of various sizes, in order to cover 14 of more than 262 path destroyed by floods.
The new adaptive and functional inflatable bridge structures developed in the project are applicable to a wide array of sectors in civil and construction industries, ranging from fast deployable road bridges for emergency situations, to temporary bridges for auxiliary tasks in the construction of highways and railways.
Manufacturing of the air bridge structures considered in the project is possible but an integrated and automated process chain has not yet been developed. Work in the project has resulted in the definition of the production equipment and the related manufacturing system for the massive industrial production of such highly resistant and adaptive air-beam/bridge structures. A fully automated and adaptable manufacturing system is currently being developed which will address the vertical integration of the process production chain including the definition of cutting pattern, assembly of patterns by sewing or gluing, transport, deployment, layout, dismantling and storage, among other specifications.
The air bridge will be applicable for facilitating emergency evacuation tasks, communication, provision of resources and rebuilding of devastated areas due to natural disasters. Indeed, many other applications of the air-bridges developed in this project exist in surface transport engineering and in the building construction sector, among others.
In summary, the growing need for emergency civilian bridging systems clearly establishes a need for a new generation of unique light weight and easily deployable bridge systems as the air-sandwich bridge.
Fan Blade Aero-Engine
It has been estimated that a reduction in the mass of a fan blade by 15%, a project objective, would be translated to a reduction of the weight of a large twin engine aircraft by 350kg. General Electric claimed a weight reduction of nearly 350kg by switching to a composite fan system for their new GEnx engine for the Boeing 787. These weight savings are multiplied when one considers the reduced mechanical performance requirements in terms of bearings and the fan containment casing due to the lower mass of the fan blade.
The main factors to consider in the design of a inlet fan blade are the strength, the weight and the damping behaviour.
Analysis using FE indicates that the natural frequencies of the composite blade are relatively close to those of the equivalent titanium blade. However, the lower damping ratio and lower frequency for mode 3 for the titanium blade would result in higher stresses from the harmonic loading and so the composite blade would have an improved performance in this respect..
The total mass of the blade configuration developed during the project when using solid titanium would be approximately 1.64kg whereas with the mass of the composite blade is approximately 0.53kg this being a reduction of approximately 70%. It should be noted however, that the titanium blade would normally be hollow so the mass would be less than that calculated but the projected weight saving of 15% seems very feasible.
From the work carried out it is possible to make the following observations:
• The mode shapes obtained were exactly those that were predicted from finite element models while the natural frequencies were somewhat lower than expected.
• A model with isotropic properties came surprisingly close to predicting the dynamic properties accurately, this taking some complexity out of the analysis.
• Adding nanotubes to the material appeared to increase the stiffness by a small amount.
• Damping levels were higher in the composite blades than would be seen in equivalent metal ones and so the performance in respect of vibration should be improved over the metal equivalents.
• Properties of the blades were very similar around 150°C to those measured at room temperature and so the blades should perform in a satisfactory manner in an engine under load.
• The blade behaviour was relatively linear with respect to vibration amplitude and exposure to extended high level vibration did not cause damage.
The blade design developed shows great promise in respect of reducing the mass of the engines on an aircraft and so improving fuel performance and reducing environmental impact from emissions and also giving better damping and so improved performance of the blade.
Morphing Honeycombs
The use of honeycomb structures offers advantages in respect of the noise and thermal insulation as well as the ability to manufacture light stiff structural components.
SILICOMB honeycombs, in the form of honeycombs and laminate structures, have been made and evaluated. The properties of the honeycombs were in many cases superior to those of the current commercial materials both mechanically and in terms of thermal performance.
The morphing structures developed proved to be very suitable for deployment as transformable space antenna, numerical modelling approaches being developed for the assessment and design of such structures.
The work has resulted in a further project being started, FP7-AAT-2012-RTD-LOMORPHELLE, which should result in exploitation of the technologies developed.
Space Antenna Assembly
The support structures designed and manufactured using PEEK film reinforced with CNTs, were modular structures where a module, a panel-resembling the DIRIS panel, is used repeatedly. For each structure, the size and the distribution of the cells were optimised numerically, attempting to address the requested structural characteristics. Only structures of one and two modules were experimentally investigated, since it was assumed, and numerically confirmed for compressive loading, that key parameters of their performance are representative for structures with several modules.
The performance of the modules was assessed using dynamic, multiaxial, compressive–tensile fatigue loading, cyclic compressive tests up to the yield and the logarithmic decrement and the hysteresis loop methods were used to determine the loss factor.
The measured natural frequency (in the X direction) of the DIRIS panel and the core were high enough and outside the usual range of vibration of transport systems. Changes in the cell size and the cell distribution in the structure can be used to tune the natural frequency system. This offers the possibility of tailoring the design of the structure such that its natural frequencies coincide with the vibration of the system, therefore maximising the energy absorption and so the design offers great potential for satellite applications and similar applications where stable platforms are required.
Surface Coatings
Most polymers are susceptible to degradation as a result of exposure to UV radiation. The chemical structures of PEEK and PPS result in significant absorption within the UV spectrum and so measures are needed to prevent/inhibit/reduce any degradation mechanisms that may result from this absorption. Whilst additives can be used in the polymers, this involves compounding with the subsequent problems related to powder manufacture and so the chosen route for UV protection of structures was by applying coatings.
The coating development work and associated testing for the PPS laminates resulted in optimised formulations described as numbers RK 022 and RK 023 which passed all the tests including vibration tests, bonding tests, and chemical resistance and could be applied by spraying, a qualified industrial application procedure. Chemical resistance included resistance to gasoline and benzene, as these were considered to be challenging reference chemicals for motor/truck applications.
Similar activities resulted in Formulation V12 for the PEEK laminates, the adhesion of the coatings to PEEK being more difficult than to PPS. Testing for the PEEK laminates also included subjecting the coated laminates to 3000 hours of UV radiation and 100 KGray of gamma radiation impact, the coatings performing well.
The coatings contained aluminium pigmentation, this being a mixture of a so called conventional “non-leafing Alupigment” and a high reflective leafing “bronze” pigment, which achieved a surface reflection of approximately 70 %. The transmission measured between 250 nm and 700 nm was 0%. Durable UV and heat resistant coatings were successfully produced for both types of composite matrix and are now commercially available.
Structural Health Monitoring
One major problem with composite structures is damage detection. The ideal approach is to be able to mount a sensor system into a composite structure such that the behaviour of the structure can be measured and analysed in real time.
A structural health monitoring system based on the swept-wavelength coherent interferometry technique has been successfully demonstrated in this project. Initial work involved the successful surface mounting the fibre on the composite materials and also the development of techniques for the integration of distributed fibre optic sensors (DFOS). Later work involved moulding the fibre into PEEK composite which showed that the DFOS was capable of withstanding temperatures of 385 °C and pressures of 10 bar.
The DFOS and the measurement technique have been shown to provide excellent results during fatigue tests containing 10 million load cycles and also the ability to detect crack formation as well as determining local strains along the axis of the fibre.
As a result, it has been demonstrated that DFOS are reliable structural health monitoring sensors during their estimated service lifetime for the application on the surface of composite structures and the integration into composite structures under process conditions.
Thus, sensors can be embedded into composite structures and used to measure the performance of the structure in real time over the full lifetime of the composite structures. This technology can be used to ensure the safety of composite structures, for example in aircraft, and provide early warning of potential problems associated with the structures whilst in use. The results can also be used as the basis for the future development of guidelines or standards which, in safety conscious industries such as aerospace, are very important.
In conclusion, the developments within the project should stimulate the introduction of such composite materials into applications within the aerospace and automotive markets with the components produced forming good case studies. There will be economic benefits of adopting the new materials in manufacturing, usage and end-of-life as well as significant environmental benefits particularly in usage. Whilst the NVH benefits were not fully met during the project, clearly defined routes to achieving this have been developed.
List of Websites:
http://mrect.risk-technologies.com/
Executive Summary
The broad aims of the M-RECT project were several fold:
• Functionalisation of carbon nanotubes (CNTs) to aid the dispersion of the CNTs in the polymer matrices, which were PPS and PEEK.
• Manufacturing of compounds where the CNTs were dispersed and disentangled within the polymer matrix.
• Incorporation of the CNT loaded matrices into long-fibre composite materials.
• The development of computer based models to allow for the simulation of the materials and evaluation of the components produced as part of the project. The intention was to provide designers with the tools to be able to accurately model components.
• Development of tools for the risk analysis of the project and the materials being developed during the project.
• Manufacture of validation components to test the technologies developed and comparison with the output from the computer models being developed. These included automotive components, satellite applications, inflatable structures and jet engine inlet fan blades.
• The project sought to develop surface coatings to improve the UV resistance of the polymer matrices and to control heating effects due to solar radiation.
The project involved the development of a novel technique for damage detection within composite structures.
The project was therefore, very wide ranging and was aimed at covering and developing the complete supply chain from CNT functionalisation through to finished component design, manufacture, testing and use. Where possible the project used industry-standard approaches to operations and also innovative alternative approaches.
Technologies for the successful functionalisation of CNTs were developed which were found to aid the dispersion of the CNTs and scalable chemical processes for manufacture of the functionalised CNTs were developed. The development of the chemical functionalisation has resulted in one patent application being filed. The CNTs were incorporated into the matrix polymers although the anticipated improvement in simple mechanical properties was not observed. This was thought to be related to the orientation of the CNTs as the dispersion appeared to be good.
Composite laminates were manufactured from the materials developed in the project both via standard approaches and novel impregnation and bagging technologies. The development of a novel impregnation technology has resulted in one patent application being filed. Problems were experienced in the manufacture of the PPS laminates due to the formation of cracks within the matrix of the laminates when CNTs were incorporated.
The mechanical and damage tolerance properties of these materials were investigated. The results indicated that whilst the tensile properties of the CNT-containing laminates exhibit inferior properties to those without CNTs, the fracture toughness of the PEEK/CNT laminates showed promising results.
The computer models developed showed good correspondence with the performance of the validation components and the development of one of the modelling approaches is continuing within a follow-on project.
A range of novel morphing honeycomb systems have been developed based on a SILICOMB geometry. Whilst the addition of CNTs to these honeycombs did not produce the anticipated improvement in mechanical properties, the recovery of the honeycombs following compression was superior. A further follow-on project has been initiated around the novel honeycomb structures developed as part of the project.
Durable UV and heat resistant coatings were successfully produced for both types of composite matrix.
The tools relating to the evaluation of risk were fully developed and have been deployed within the consortium.
A technique for the assessment of damage within a composite structure was developed and shown to be robust enough for moulding into a thermoplastic composite material, this being vital for use in the health monitoring in large composite structures.
Project Context and Objectives:
The M-RECT project aimed to create multiscale – reinforced semi-crystalline thermoplastics, which would outperform all reinforced polymers in terms of strength, stiffness and damping, by developing and upscaling state of the art production procedures, within cost-efficient manufacturing routes.
The use of composite materials in engineered components has increased significantly within recent years due to the need to the reduce weight and the environmental impact of artefacts, particularly in the case of transportation. The desire for more fuel efficient and environmentally friendly aircraft and the regulations concerning emissions from road vehicles is driving both the aerospace and automotive industries to adopt composites as materials of construction. Examples of this trend are:
• The BWM i3. In order to maximise the range of electric vehicles, the body weight must be reduced and so composites are being used.
• Boeing 787 where approximately 35T of composite are used per aircraft. The use of composites results in significant weight savings for the aircraft which in then yields better fuel efficiency this giving longer range and better environmental impact.
The majority of composite materials are currently based on thermosetting matrices but thermoplastic matrices offer many advantages. The potential advantages include recyclability, faster and more energy efficient production methods, improved fatigue resistance and better mechanical performance, in particular impact performance.
It was in this context that the project was developed in order to seek to:
• Improve the performance of the composite systems by the inclusion of carbon nanotubes (CNTs)
• Provide scalable manufacturing technologies for the raw materials and products made from these multi-scale reinforced composites
• Fully characterise the performance of the materials
• Develop modelling software to aid the design of components made from the multi-scale materials
• Develop UV and heat resistant coatings to extend the performance of the materials
• Develop health monitoring systems for use in multi-scale composite structures
• Develop novel honeycomb systems
It was anticipated that multiscale reinforcement of PEEK or PPS in aircraft manufacturing would greatly enhance composites durability for long-term utilisation at intermediate temperatures (100-150°C). On top of the strength and damping enhancement, the crack-bridging capability of the CNTs was expected to result in a reduction in micro-cracking under thermal-cycling ageing or isotropic ageing with oxidation.
The semi-crystalline thermoplastics selected for the project were Poly-Ether-Ether-Ketone (PEEK) and Poly-Phenylene-Sulphide (PPS). The matrix materials selected were very-high performance thermoplastic resins which are used extensively in the aerospace and allied industries due to their properties. The products derived from these materials are also high value.
However, whilst these matrix materials were specified, the proposed project sought, primarily, to develop generic technologies for the production, modelling and design of high-performance, light-weight reinforced thermoplastics. The multiscale reinforcement envisaged, would comprise of disentangled, dispersed, straightened and aligned multi-walled CNTs and also fully impregnated long carbon fibres (CFs).
It was envisaged that, depending on the extent of reinforcement and manufacturing process, four products would emerge with a PEEK or PPS matrix:
i. Polymer sheets reinforced with CNTs.
ii. Laminates reinforced with CNTs and long fibres (multiscale reinforcement).
iii. Honeycomb cores with bonded panels as skins (multiscale reinforcement).
iv. Laminated parts with complex shapes (multiscale reinforcement).
The effective utilization of CNTs in a polymer matrix depended strongly on the development of processing technologies that would ensure their disentanglement, dispersion, straightening, alignment and good interfacial bonding with the thermoplastic matrices used.
The risks associated with failure of any of the newly developed technologies would be mitigated by always developing processes in parallel with the use of the existing state-of-the-art technologies in order to ensure delivery of the objectives.
The main objectives of the proposed project are the following:
1. To develop CNT-reinforced sheets from semi-crystalline thermoplastic PEEK and PPS at low CNT concentrations in order to obtain a nanocomposite material showing improved performance over the matrix material. The primary objective of this work was to functionalise the CNTs by directly synthesizing polymer-specific molecular chains on their surface. This functionalisation was intended to promote de-aggregation of the CNT bundles with minimal destructive effects on their structure. Their functionalisation with polymer specific organic groups would promote bonding with the matrix. As part of the project, the chemical procedures being used would be modified to increase the rate of the functionalisation and also to simplify it.
2. To develop novel technologies for the disentanglement, dispersion and straightening of the CNTs.
3. To investigate the use of microwave heating to reduce heating times and improve the energy efficiency of the processing of the CNT containing materials. Microwave heating of the CNTs and eventually of the polymer should be a very efficient process since the material is heated rapidly at the molecular level and therefore volumetrically. It reduces material degradation (surface micro-cracks), while the targeted localized melting along the CNT-polymer interface should minimise heat losses and improve the control of the melting and bonding processes.
4. To use the envisaged CNT-PEEK and CNT-PPS nanocomposites as matrices for continuous carbon fibre reinforcement, achieving high levels of impregnation and fibre distribution.
5. To use multiscale-reinforced PEEK sheets to manufacture strong honeycomb cores and panels based on the Directionally Reinforced Integrated Single-yarn (DIRIS) architecture.
6. To develop novel morphing honeycomb structures using CNT reinforced films.
7. To improve the quality and further develop a novel dry powder impregnation technique in order to impregnate performs for laminates and honeycombs.
8. To integrate the multiscale reinforced polymers for the production of representative aerospace and automotive components by developing smart out-of-autoclave manufacturing processes.
9. To develop on-line monitoring tools for rapid measurement of selected electrical properties and almost real-time quality assessment of the outcome of the production line.
10. To verify and measure the mechanical properties of the multiscale reinforced materials and to assess the quality of the consecutive stages of manufacture through testing at nano-scale and macro-scale at a wide range of testing conditions covering the full spectrum of anticipated operating environments.
11. To determine the damage tolerance of the materials developed.
12. To develop suitable numerical tools for modelling and designing structures based on Finite Element Analysis techniques and validate the developed models and software by comparing them with reference to the mechanical test results.
13. Using the numerical tools, simulate and validate any damping mechanisms observed.
14. To demonstrate the feasibility of readily incorporating the new products in large European industrial sectors such as aerospace, automotive, building and space structures. This will include the design, manufacture and testing of real-scale prototypes, which will be characteristic of the specific applications.
15. To ensure through optimized design that the cost benefit achieved through use of the high-performance CNT and multiscale-reinforced thermoplastics developed is effectively transferred into the envisaged structures.
16. To incorporate low-cost smart health monitoring optical fibre sensors in this new-generation of structures for continuous assessment of progressing structural damage.
17. To greatly enhance the strength of the materials to chemical attack, weathering and UV rays though the use of improved novel fluorocarbon-based coatings. The fluoropolymer based coating will be bonded, through a suitable primer, onto the reinforced PEEK and PPS components, the coating protecting the structure against gamma radiation and heating up to 260°C.
18. To increase the cost effectiveness of the materials developed through automation and optimisation of the manufacturing procedures (particularly the CNT functionalisation process), to increase the strength-to-weight and stiffness-to-weight ratios in order to maximize fuel and energy savings, to minimize the use of active protection systems, to maximize safety and also to promote recyclability.
19. To improve techniques for the recovery and EOL disposal of waste materials to ensure that the cost of the fabrication process is minimised.
Project Results:
Introduction
This project dealt with:
The functionalisation of carbon nanotubes (CNTs), incorporating microwave heating
The incorporation of the CNTs in thermoplastic matrices and assessment of the quality of the resulting material
The manufacture of composite pre-pregs and laminates using these CNT filled matrices, including novel consolidation technologies
The evaluation of the properties of the laminates
The development of sensor technologies for monitoring the health of the multi-scale materials
Development of DIRIS honeycomb structures
Recycling of multiscale-reinforced thermoplastics
Numerical modelling of the multi-scale materials
Manufacture and testing of components
Life-cycle assessment of a multi-scale reinforced engine stiffener
Risk assessment
Assessment of Uncertainties
The development of surface coatings for the laminates to prevent attack of the laminates by UV and to control temperature build-up within the laminates due to solar radiation
This structure will be used to detail the development and results obtained within the project.
Functionalisation of CNTs
Two high performance thermoplastics were chosen for the development work within the project, these being poly(ether ether ketone) [PEEK] and poly(phenylene sulphide) [PPS].
Two types of CNTs were considered for use in the project, both coming from European manufacturers. These were:
Baytube C70P. These CNTs are supplied in a pelletised form. This has the advantage that they are better from a health and safety perspective as regards handling. However, the pelletised form makes them more difficult to disperse and good dispersion is important in respect of functionalizing the CNTs and also in the properties they confer on the polymer once the materials are combined.
Nanocyl NC7000. These CNTs are supplied as loose nanotubes and thus there are increased hazards in handling them but they are easier to disperse. For the reasons of ease of dispersion, the Nanocyl CNTs were chosen for this project.
The intention of the functionalisation process was to attach oligomers to the CNTs which would improve the compatibility and hence dispersability of the CNTs in the chosen polymers. In the case of PPS, the oligomers were short PPS chains and in the case of PEEK the oligomers were poly (ether ketone) [PEK] oligomers. PEK oligomers were chosen due to the simpler electrophilic chemistry associated with the polymerisation and its compatibility with PEEK.
The original intention had been to develop a functionalisation process based on the use of a baffled continuous-flow meso-reactor. However, the functionalisation chemistry initially chosen utilised polyphosphoric acid [PPA] and this had a high viscosity which was not compatible with the chosen process. As a consequence, two alternative approaches were adopted in order to ensure a solution to this problem, which was fundamental to the project, and these were:
Development of new design of reactor which could accommodate the high viscosity PPA reaction mixture. This route was used for the PEEK-like functionalisation of the CNTs.
Development of alternative chemical routes by Nanozar, these involving the use of low- viscosity solvents such as NMP. This was used for the PPS-like functionalisation of CNTs and also investigated as a route to the PEEK-like functionalisation.
Details of the various chemical routes investigated are given below.
PEEK-like Functionalisation
Synthetic route #1
This route involved functionalising the CNTs by following a Friedel–Crafts acylation, initially proposed for a PEEK matrix by Baek et al (Polymer 44, 2003, 4135-4147). In the reaction PPA and phosphorus pentoxide [P2O5] are used as drying reagents but also act as a solvent, Friedel-Crafts catalyst and dehydrating agent. The first stage of the process involves 4-phenoxybenzoic acid reacting and bonding onto the CNTs, this being followed by the in-situ polymerisation of 4-phenoxybenzoic acid to form the PEK oligomers. The published method was carried out at 130°C for 48 hours followed by water extraction over three days and a further three days extraction utilising methanol.
Synthetic route #2
This involves the oxidation of the CNTs and anchoring them to the PEEK backbone. A patent has been applied for relating to this route.
Synthetic route #3
This route involved the functionalisation of CNTs with PEEK monomers and the in-situ polymerisation of PEEK, the CNTs being introduced during the synthesis of the PEEK. This route involves the addition of phenolic groups onto the surface of the CNTs, the phenolic groups featuring hydroxyl groups which can take part in the PEEK polymerisation process. The initial stage of this process takes approximately twelve hours at 60°C with DMF as the solvent.
All three routes proved to yield functionalised CNTs.
Route #1 was used for the manufacture of the PEEK-like functionalised CNTs. This route initially involved several steps and several days for completion. An extensional flow reactor was developed, this reducing the timescale for the functionalisation process to thirteen hours. The reciprocating extensional flow reactor featured 10µm filters through which the reaction mixture flowed, this promoting disentanglement and dispersion of the nanotubes, this being evidenced by the changing rheology during processing. Micrographs from an SEM and images from an AFM indicated that the nanotubes were no longer clustered. The process developed:
Had the capability of splitting a volume of more than 1lt into millions of micro-cylinders, with lengths a few times greater than the length of the CNTs.
Produced elongational flow rather than shear flow using convergent – divergent flow geometries.
Had the capability to provide good control of the flow parameters.
Exhibited upstream and downstream flow to re-open blocked channels.
Was energy efficient.
Exhibited limited re-aggregation and temperature increase.
Resulted in limited attrition of the CNTs.
Was capable of being scaled–up.
Reduced the reaction time for functionalisation by a factor of ten from that originally used by Baek.
The productivity of the extensional flow mixer (from CNT disentanglement to homogenisation) was 1kg/24h (1kg of functionalized CNT/PEEK slurry).
The procedure was validated periodically by the use of FTIR examining the spectrum for the loss of the peak at 2922 cm-1, this indicating no free defect sites or open ends on CNTs, and a peak at 1650 cm-1 for the C=0 bond, this indicating the functionalisation. Analysis using an SEM indicated that entangled CNTs were found in <1% of all the images examined.
If the CNT disentanglement is 100% successful every defect site is available for reaction with the monomer. If the CNTs are fully functionalised, the C-H stretching band disappears. [Jasco, Ir application note 02-03 http://www.jascoint.co.jp/asia/literature/application_notes.html ]
Numerical simulation, using a Molecular Mechanics Method, and also observations from SEM images of functionalised CNTs indicated that the degree of functionalisation was small (< 10 chains per CNT) but that the PEK polymerisation around the functionalised CNT increases the CNT diameter by, on average, 3nm.
Upscaling the PEEK-like Functionalisation Process
Following the validation of the suitability of the extensional flow mixer in performing the tasks of CNT disentanglement, dispersion, distribution and functionalisation and their homogenisation with PEEK grains, the up-scaling to a pre-industrial model was carried out by improving the efficiency and increasing the capacity of the equipment. The improved mixer configuration included a simultaneous translation and rotation of an assembly of a great number of micro-channels, these being part of the piston this process resulting in:
Flow splitting, recombination and rearrangement
Flow twisting, folding and reorientation (shearing of the stream lines)
Transition from laminar to turbulent flow at the exit of micro-channels
It should be noted that the mixer operation was changed from disentanglement to functionalisation by simply changing the filter unit.
The new extensional cross-flow mixer, performed all the functions for a mixture over a 24 hour period.
Microwave Heating of Nanocomposites
Microwave heating of materials is preferred to conventional heating, as microwaves heat the material at the molecular level, i.e. volumetrically, and there are no temperature gradients when heating homogeneous materials. In contrast to that, only the material surface is heated in conventional heating, and heat has then to be transferred to the inside of the material by conduction, which is a very inefficient process for materials with low thermal conductivity.
A detailed investigation was carried out and it was concluded that the use of microwave heating in combination with the given scheme of mechanical and thermal processing of the nanocomposite had several severe disadvantages:
The need for simultaneous heating of the high pressure chamber.
High temperature gradients caused by inhomogeneity of the energy absorbed in the cylinder, both along its axis and along its diameter.
The high cost of microwave generators.
The sequence of the proposed mechanical and thermal processing which made the microwave method difficult to apply in a continuous flow production line.
The harmful effects of microwaves on the human organism which make it necessary to take additional measures for cutting off the electromagnetic energy from the exit slot from the equipment.
Thus, the microwave method proved to be incompatible with the proposed scheme of mechanical and thermal processing of the nanocomposite material, since the material is permanently in contact with the walls of the chamber. The material had to be heated by conventional means, which were much less expensive as well as being more reliable.
PPS-like Functionalisation
Two routes were studied in the case of functionalisation of CNTs with PPS-like compatible groups.
Synthetic route #1
This route involved a first stage of reacting CNTs with PPA and 4-chlorobenzoic acid in the presence of phosphorous pentoxide and, in a second step, reacting the product with 4-methylbenzenethiol in NMP/toluene in the presence of sodium carbonate. This mechanism introduced 4-(4- methylbenzenethio)benzoyl groups onto the CNTs, these being highly compatible with the PPS. The functionalised CNTs obtained in the first step were then reacted with chlorobenzenethiol or dichlorobenzenethiol to give anchored linear poly(phenylene sulphide) (LPPS) or hyperbranched poly(phenylene sulphide) (HPPS).
In the second step PPS compatible groups were attached to the grafted monomers. The first reaction uses methyl thiol as reactant and 2,4-dichlorothiol as reactants and the products obtained were characterized by TGA and FTIR.
The first intermediate product was synthesised in the laboratory with a good yield. The reaction was conducted at 130ºC over a period of 72h under mechanical stirring. The product obtained was soxlhet extracted with water and methanol for 3 days with each solvent to remove the non-grafted molecules. The product obtained was dried under vacuum and characterised by FTIR and TGA.
Synthetic route #2
The first route had problems associated with it due to the high viscosity of PPA and so up-scaling was difficult and thus an alternative was considered.
In a second step the grafted molecules react with sodium sulfide in the presence of dichlorobenzene resulting in PPS compatible groups grafted onto the surface of the tubes.
All the products were characterised by means of TGA and FTIR in order to determine if the reaction had occurred and to estimate the amount of grafted polymer. With both strategies the results obtained where optimised and characterised.
Upscaling the PPS-like CNT Functionalisation Process
An oscillatory flow reactor (OFR) was developed and validated within The Polymer Fluids Group at The University of Cambridge.
The initial functionalisation process, synthetic route #2, was a two-step reaction with long reaction times. Tests were carried out using a small lab scale stirred 250 ml reactor to establish if reaction times could be shortened and whether performing a single stage reaction was viable. It was found that it was possible to reduce the reaction time to 6 hours (4 hours for the first reaction and 2 hours the second reaction) compared with the 24 hours originally proposed. In addition it was discovered that it was possible to move from a two-step reaction to a single-step sequential reaction taking 6 hours using DMF as a solvent, thereby saving reaction, washing and separation times.
Key factors in the scale up process were the initial dispersion of the CNTs and the concentration of CNTs in the reactor. Without surfactants, CNT aggregation occurs even at low concentrations and so good dispersion is difficult to achieve. However, aggregation greatly enhances the separation time of CNTs from the solvent after reaction and also has benefits in handling the functionalised CNTs on completion of the reaction.
The initial CNT concentration used at lab scale was around 0.15 %, which was considered far too low for economic up-scaling. In order to increase the initial concentration and to optimise the up-scaling process the possibility of increasing the initial concentration to 1 % was assessed at lab scale. Despite CNT aggregates being present in the bulk of the reaction, there was clear evidence that functionalisation yields were in the same range as those obtained using lower concentrations.
Scale up of the functionalisation process was carried out using an OFR. By combining oscillation with periodically spaced baffles in a tube, vigorous eddy mixing can be achieved between the baffles and this leads to excellent mixing (Mackley MR and Ni X. Experimental fluid dispersion measurements in periodic baffled tube arrays. (1993) Chemical Engineering Science, Vol. 48, No 18, 3293-3305 and Ni X, Mackley MR, Harvey AP, Stonestreet P, Baird MHI and Rama Rao NV. Mixing through oscillations and pulsations. A guide to achieving process enhancements in the chemical and process industries. (2003) Trans IChemE, Vol. 81, Part A, March 2003). These baffles are usually low-constriction orifices plates where the fractional open cross-sectional area is based on a compromise between minimising frictional losses and maximising the mixing effect. The baffles are typically spaced at a uniform distance apart, in the range 1-2 times the tube diameter, with a spacing of 1.5 being the most common. A linear actuator causes the oscillation, moving the set of baffles up and down from the top of the reactor.
The OFR consisted of a jacketed reaction vessel of 104 mm internal diameter and 3 litres capacity although only 2.5 litres of reaction fluid were used. The vessel was operated vertically and connected to a linear actuator which caused the oscillation of the set of baffles, in this case in a reciprocating plate column. During oscillations, it was possible to specify both the frequency and the amplitude of the movement, giving a wide range of mixing conditions. The oscillation conditions were a frequency of 1 Hz and the amplitude centre to peak was 10 mm giving a moderate oscillation in the eddies formation region. The temperature was controlled by a Huber CC302 oil circulator which was able to reach the temperature needed for the reaction. Once the reaction had finished, the CNT aggregates were allowed to settle for a few minutes under gravity and then washed and filtered using Omnipore Membrane filters 1.0 µm coupled with a vacuum pump.
Following all the reactions TEM results showed a layer coating the CNTs. IR spectroscopy was also indicated that the CNT functionalisation was successful in both the one and two-step reactions.
TGA analysis was performed to quantify the amount of functional groups attached and to compare the yield of surface functionalisation between the one-step and two-step reaction process.The functional groups generated using the OFR with 1 % CNTs present showed high thermal stability and the yield was higher than in the initial lab scale experiments. The yield was slightly higher (35% compared to 30 %) if the reaction was carried out in two steps but the additional yield was considered not large enough to justify the longer processing time associated with the two-step process. In particular, the additional washing and filtration in the two-step reaction slowed production significantly and utilised a large volume of solvent.
Once the reaction protocol in the MRECT OFR was established and validated for a single step reaction with an initial concentration of 1 %, another trial was carried out with an initial concentration of 2 %. The TGA analysis showed encouraging results with a yield of grafted functional groups higher than 25 % and with high thermal stability. The output of functionalised nanotubes from the OFR was around 500g/month.
The OFR produced aggregated CNT dispersions at all concentrations tested. Initially there were concerns that the aggregates would inhibit functionalisation, however it was realised that this was not the case and the aggregation provided a strategic advantage in that on completion of a reaction fluid oscillation could be stopped and the CNT aggregates would settle in the bottom of the reactor within minutes. This feature meant that it was not necessary to use very long settling times or high pressure small pore size filters and had advantages in respect of handling the materials at the compounding stage.
Incorporation of CNTs into Thermoplastic Matrices
Incorporation of CNTs into PEEK
Problems were experienced incorporating the functionalised CNTs in to PEEK and two routes were adopted to overcome the problems:
Developing and scaling-up a melt mixer which operates on similar principles to the equipment used for the functionalisation process as discussed previously.
Industry standard process using a twin-screw extruder.
Novel Melt-Mixing Device
The construction of the initial equipment comprised of an outer static chamber, a symmetrical piston and a double rod system which rotated and translated simultaneously within the chamber.
Techniques were developed to determine the uniformity of the CNT dispersion by analysing the density of CNTs in systems of concentric rings of constant area on SEM images. The product from the melt mixer was subsequently processed by hot-compaction in order to produce a mechanically-stable tape which could be handled and fed into the next stage of the process. The compaction conditions were 250 MPa biaxial pressure at 370°C, this producing a micro-void free tape 600mm long.
The hot-compaction system increased the density of the material from 0.91g/cm3 to 1.38g/cm3 which is higher than normal crystalline PEEK. The tapes produced in the hot-compaction process were then fed into a calender system to further enhance the dispersion, distribution and orientation of the CNTs.
The tapes were then subject to sequential iso-stretching and annealing aimed at increasing the levels of orientation. The localised heating system on the equipment developed during the project ensured local uniform deformation of the tape at the minimum acceptable temperatures. This process also sought to reduce the natural waviness of the CNTs, aligning them in the direction of stretching.
Due to damage sustained in the mixing of the CNTs with PEEK, the melt-mixing device had to be rebuilt such that it was fed via a single screw extruder and featured a simple slit die at the exit to produce a film. The film was cooled rapidly with water sprays so that it was amorphous. This film was then subjected to stretching as before. Samples of the film evaluated indicated that there may be some residual undispersed and unbound carbon nanotubes and suggested that some further work may be needed to perfect the design. Outputs from the equipment were quite low when compared to the capacity of the alternative route, process of mixing and dispersing the CNTs in the machine taking several hours and so the process did not really constitute a viable commercial process with PEEK.
A CNT concentration of 0.5 wt% was adopted for all the materials investigated in the project on the basis of:
It was found that even with a concentration of 0,5 wt% CNT, the dispersion of the CNTs was not feasible following processing for several hours in the melt-mixing equipment.
Examination of the electrical conductivity of aligned PEEK/CNT tapes with CNT loadings from 0.15 to 1.5 wt% indicated that the percolation threshold occurs at about 0.2 wt% after processing on the melt-mixer.
The optimum MWCNT loading in PEEK composites for thermal stability is 0,5 wt%, according to P. Patel et al [P.Patel et al. Flammability properties of PEEK and CNT composites, Polymer Degradation and Stability (2012) 1-11].
Twin-Screw Extrusion
The second route to producing a 0.5% CNT loaded PEEK was by compounding of the functionalised CNTs with PEEK using a Coperion ZSK32MC twin-screw extruder. The compound was then ground down to produce a powder with a d50 of approximately 1000μm for subsequent use.
Initial trials to process the PEEK/CNT powders using a twin-screw extruder resulted in the PEEK degrading. Elemental analysis indicated that the material contained a high level of phosphorus which can be an issue when processing PEEK in the molten state. The phosphorus was assumed to originate from the polyphosphoric acid used in the functionalisation process. Compounding PEEK with unfunctionalised CNTs did not result in any similar degradation-related problems and the elemental analysis results indicated that the phosphorus levels were very low in comparison. Analysis of the material using DSC, TGA and thermal desorption with GCMS showed that the polymer was degraded.
A second sample of PEEK/CNT material was supplied for compounding, elemental analysis indicating that the residual phosphorus levels were significantly lower than in the first sample, 210ppm as opposed to 800ppm. This material was compounded successfully.
It was unclear as to whether some aggregation and attrition of CNTs would occur via this route. However, subsequent results indicated that this was not the case.
Incorporation of CNTs into PPS
Initial work involved processing and testing PPS neat resin as a reference, PPS with 2% CNTs and
3% CNTs and PPS with pre-dispersed CNT masterbatch as a reference for CNT dispersion. The CNTs used in this work were BaytubeC70P and the compounding was carried out using a Coperion ZSK32MC machine with the low-shear screw profile.
The properties of the compounds were assessed by:
Surface Resistivity: correlated to the dispersion of CNTs creating a network inside the polymer matrix
Tensile elongation at break and Impact Resistance: suitable to reveal the presence of agglomerate of CNTs in the compound
As a consequence of the high surface resistivity found with the Baytubes, a higher-shear screw profile (VHS) was examined. The work indicated that better dispersion was obtained if the CNTs were fed into the extruder part way along the barrel. It was also found that feeding the CNTs part way along the barrel whilst using the HT screw profile gave better, more consistent results. This approach produced the optimum shear and residence time for the materials. The optimum screw profile was found to be a compounding length of 20/1 L/D with the CNTs being added half way along the extruder barrel. The PPS is added at the rear of the machine so that it is molten before the addition of the CNTs. This is to prevent the CNTs sintering together as this makes dispersion more difficult. The CNTs are added part way along the machine in order to limit the amount of work carried out on them and so limit any possible attrition effects.
The optimised process resulted in moulded specimens with a resistivity of 104 ohm at a CNT loading of 3.25% using Baytubes C70P.
Baytubes are supplied in a pelletised form and so tend to present less of a potential health hazard than Nanocyl NC7000 CNTs which are very dusty. Techniques had to be developed as part of this work to ensure good levels of safety.
The work resulted in the manufacture of 3.0% loaded Nanocyl NC 7000/PPS compounds with a surface resistivity of 104 ohms. Further optimization of the processing parameters was carried out, the result being CNT/PPS compounds with a surface resistivity of 104 ohms at a loading of 2.5% of Nanocyl NC7000.
Attempts were made to produce masterbatches with CNT loadings of 17.5% but it was found that the CNTs tended to be packed together and the dispersion was poor. This approach was abandoned.
The initial screw profile used with functionalised CNTs, the High Shear profile, was found to be the best profile for the unfunctionalised CNTs. The results obtained with this configuration indicated that the product was sub-optimal and so the material was re-processed on the extruder to determine if this would generate an improvement. Reprocessing did produce improved results and so this indicated that process modifications were required to improve the quality of the compounds produced.
The revised screw configuration was the same basic screw configuration but the CNTs were added earlier in the process. Some mechanical properties showed an improvement over the compound produced with the initial machine configuration. The effect of the second loop through the machine was less pronounced in this case.
A third configuration of the machine was evaluated where the compounding length for the CNTs was maximised.
Following decisions within the consortium, the CNT loading was reduced to 0.5%. Several further iterations were carried out using:
• Functionalised CNTs
• Powder and pellet PPS
• Different molecular weights of PPS
The results indicated that:
The functionalised CNTs were less susceptible to sintering than the non-functionalised CNTs.
The dispersion of the functionalised CNTs was best with the maximum compounding length.
The process of dispersing the functionalised CNTs was different to that found with the unfunctionalised CNTs.
It was noticed that the theoretical reinforcing action of CNTs was not observed in the injection moulded samples produced.
Compounding work had been carried out on a small scale at another partner and this indicated an improvement in mechanical properties and this did not correlate with the results obtained on commercial scale equipment.
It was noted that there were some inconsistencies in the various data sets for the polymer and the compounds:
The base PPS resin is Fortron 0205 but the data produced on the lab scale was different from those reported in the technical data sheet of the manufacturer.
The mechanical data for PPS with unfunctionalised CNT compounds generated on the lab scale appeared to be much worse than those generated on the commercial scale at the same CNT concentration.
In contrast, the data for PPS with functionalised-CNT compounds generated on the lab scale appeared to be much better than those obtained on the commercial scale at the same f-CNTs concentration.
The reason for the differences was unclear and so extensive evaluation of the dispersion of the CNTs in the PPS was undertaken using SEM. The SEM work indicated that the CNTs were well dispersed in all the materials as well as in composite laminates produce using the materials.
There were a number of differences in the sample preparation methods used, these being:
Lab scale processing involved premixing the materials before processing them, in a molten state, on a twin screw extruder. The materials were milled as part of the premixing process.
The lab scale compounding was carried out using a Haake Minilab which is a small laboratory twin-screw extruder.
The lab scale materials were converted into test samples by compression moulding whilst the commercial scale materials were converted using injection moulded.
In injection moulding it is well know that with short-fibre reinforced polymers the cross-section of the component has a skin/core structure. In the skins, the fibres are predominantly oriented parallel to the flow whilst in the core they are transverse to the flow. The processing conditions and rheology of the polymer have a significant impact on the thicknesses of the skin and core layers. Thus it is possible that the core layer of the injection moulded samples was significantly large and so most of the CNTs were oriented perpendicular to the test direction and thus had little effect on the properties.
The flow occurring during compression moulding is limited and as a consequence, the CNTs were more likely to adopt a random orientation in the plane of the moulding. A random orientation would result in a significant number of CNTs being aligned in the direction of test and thus one would expect the improvement in performance that was observed.
It is believed that these flow-related effects were the primary reason for the differences in the test results.
For the impregnation of the carbon fabrics to form a pre-preg, the material had to be ground into a powder. The target particle sizes for this work was a d50 of 1000μm for Ten Cate and a d50 of between 50 and 100μm for Fibroline.
A d50 of 1000µm powder was relatively easy to achieve by normal polymer processing techniques. Particle sizes below 100µm were, however, found to be more difficult to achieve and a number of methods were investigated.
Initial work was carried out using a pin mill and this produced powder which had a particle size of under 1000µm but the particle sizes required by Fibroline could not be achieved.
Following this an underwater pelletiser system was investigated using pure PPS. This produced very regular sized spherical particles around 400µm in diameter. Whilst the process was capable of easy industrialisation, the particle size was still too large for Fibroline.
The next process investigated was particle size reduction by impact at very high speed. A laboratory device was used where the powder was fed tangentially at subsonic speeds (approximately 50 m/s) into the flat cylindrical milling chamber through a Venturi system using pressurized air or nitrogen. Once inside the milling chamber the particles were then accelerated by a series of jets around the perimeter to supersonic speeds (300 m/s), in a spiral movement. The micronising effect occurs when the slower incoming particles and the faster particles, in the spiral path, collide. While centrifugal force retains the larger particles at the periphery of the milling chamber, the smaller particles exit with the exhaust gas from the centre of the chamber.
Tests on small sized samples showed that 100µm could be achieved but unfortunately in the parallel work on impregnation, it had been found that powder with a particle size of 100µm was unsuitable for their process, the process requiring a particle size of around 20µm.
A Ball Mill machine was developed and some preliminary tests on small samples of PPS with a 15% loading of CNTs and PPS with a 17,5% loading of CNTs were by IMMG, the materials being processed for 4, for 8 and 12 hours. The outcome of these tests was not satisfactory as the particles became flat, a few millimeters in size, with a thickness of around 0.3 mm.
A ring mill was constructed, this possessing superior grinding performance when compared to ball mills as the energy density is around a thousand times greater. The equipment featured a chamber which contained a revolving shaft that had sets of rolls attached to it. As the shaft rotated, the rolls were forced radially out against the wall of the chamber. This type of mill was reported to have worked well with minerals but unfortunately was unsuccessful in meeting the particle size requirements specified.
External trials did indicate that grinding down to a d50 of 20μm was possible, one design of grinder successfully manufacturing a powder with a d50 of 23µm, but the production rate was so low that it would not be a commercially viable process.
Subsequent work on the novel impregnation technology resulted in powders of 50μm being suitable for the process. However, the problem of grinding the CNT reinforced materials down to a d50 of 50μm was not resolved.
Quality Measurement
Techniques need to be available for the quality assessment of the compounds manufactured and also, ideally, for the on-line assessment of the quality of product.
Differential Scanning Calorimetry (DSC)
Data obtained using DSC indicates that the levels of crystallinity achieved with a CNT loaded PEEK is less than that obtained with pure PEEK. It is believed that this arises due to the restricted chain mobility arising from the interaction of the functionalised CNTs and the PEEK molecules. This indicates that the functionalisation has been successful and that there is significant interaction between the CNTs and the PEEK.
DSC was used to investigate the glass transition region. Experiments were performed at a heating rate of 5°C/min. In all experiments, a weak glass transition is obtained, which is expected for a semicrystalline material. Figure 46 shows heating thermograms obtained with PEK/CNT nanocomposites at various compositions as indicated on the Figure. The slight Tg reduction in the nanocomposites, obtained after evaluation following standard procedures, is attributed to the increment of polymer mobility [Ash, B.J. Siegel, R.W. Schadler, L.S. 2004. Glass-transition temperature behaviour of alumina/PMMA nanocomposites. J. Polym. Sci. B 34, 4371-4383], indicating that the interface between matrix and nanotubes must be poor and mechanical properties are not improved [Zax, D.B.,Yang D.-K. Santos, R.A. Hegemann, H., Giannelis, E.P. Manias, E., 2000. Dynamical heterogeneity in nanoconfined poly(styrene) chains. J. Chem. Phys. 112, 2945-2951].
AC and DC conductivity measurements
Conducting networks formed within an insulating matrix can be utilised as highly sensitive sensors for detecting the onset, nature and evolution of damage in advanced polymer-based composites. The samples firstly had gold stripes sputter coated onto them, copper electrodes were attached to the stripes using a conducting silver-loaded epoxy, the samples then being mounted on the tensile testing machine.
DC techniques are mainly sensitive to fibre failures whereas AC measurements provide information on the development of both intra-ply and inter-ply matrix cracks, the latter being associated with delamination.
The results showed a normalised resistance change, this indicating that there is a significant increase in (ΔR/R0) before the plastic deformation of the sample. It is also noted that (∆R/Ro) changes faster than strain in the elastic region.
Various configurations of electrode have been evaluated, the work showing that 4-wire sensing was superior to 2-wire sensing and it was noted that the electrical resistance linearly increases with strain until fracture and the resistance changes immediately from minimum strain. Changes in resistance were also observed when cyclic loading was applied. The results showed a reversible change of resistance on every cycle, the cycles being repeatable. A small increase in resistance was observed at zero stress after many cycles.
Dynamic Mechanical Analysis (DMA)
Dynamic mechanical analysis at 1Hz has shown significant differences in the behaviour of PEEK and CNT loaded PEEK materials around the glass transition temperature.
DMA measurements for the evaluation of glass transition and the assessment of improvement of mechanical properties were performed using the TA Instruments DMA Q800 in the tension mode with a pre-strain equal to 0.7%, a strain amplitude of 0.5 %, a temperature range 20-230°C, at five frequencies of 1, 5, 10, 20 and 40 Hz, and a heating rate of 3°C/min. The storage and loss modulus curves versus temperature were evaluated, as well as the corresponding master curves, applying the time-temperature superposition (TTS) principle.
Scanning Electron Microscopy (SEM)
Various samples have been examined using a Nova NanoSEM 230, Fei-Innova Nanoscope, and micrographs are indicate the presence of the nanotubes on the surface of the polymer powder particles and the nature of the compacted samples prepared using the hot-compaction system discussed previously. All samples were coated with a thin film of gold (~7 nm).
Attempts were made to observe the CNTs below the surface of the tape. Chemical etching with phosphoric acid [Olley, R. H., Bassett, D. C., Blundell, D. J., & Division, P., 1986. Permanganic etching of PEEK. Polymer 27, 344–348] did not reveal the CNTs in the matrix. Further attempts were made using laser ablation at 213 nm, using a Nd:YAG laser. The results were poor and this method would need optimisation in order to improve the results. The use of a laser at 355 nm was attempted [Romoli, L., Fischer, F., Kling, R., 2012. A study on UV laser drilling of PEEK reinforced with carbon fibers. Optics and Lasers in Engineering 50, 449–457] but the results were worse due to thermal diffusion in the polymer. There is also a question as to whether laser ablation results in redistribution and/or damage of CNTs [Qian, D., Dickey, E.C. Andrews, R., Rantell, T., 2000. Load transfer and deformation mechanisms in carbon nanotube-polystyrene composites. Appl. Phys. Lett. 76, 2868-2870].
Assessment of the attrition of the CNTs
SEM and AFM observation at the different stages of processing and ac electrical conductivity measurements on the CNT reinforced PEEK sheets were used to assess possible attrition of CNTs caused by processing.
SEM images were used to provide evidence regarding possible attrition of the CNTs at the different stages of processing.
PEK/CNT samples of varying CNT content prior to calendering were selected for ac conductivity measurements. Two electrodes configuration measurements were performed using an Alpha Analyzer (Novocontrol). Measurements were performed using gold sputtered circular electrodes on opposite sides of the sample in order to assure good electrical contact between the sample and the electrodes of the apparatus. The sputtering process was performed using an EMS 550 Sputter Coater operating with sputtering current of 25mA and sputtering time of 4min. From the ac conductivity plots values of dc conductivity were determined, following standard procedure [Logakis, E., Pandis, Ch., Pissis, P., Piontek, E., Poetschke, P. 2011. Highly conducting poly(methyl methacrylate)/carbon nanotubes nanocomposites: Investigation of thermal, dynamic-mechanical, electrical and dielectric properties. Comp. Sci. Technol. 71, 854-862]. These values were used to construct the percolation plot and the following equation from percolation theory was fitted to the data and percolation threshold was determined to be φc = 0.13 % (corresponding to pc=0.17 wt%).:
σdc ~ (φ-φc)t
where σdc is the dc conductivity, φ is the volume fraction of the filler, and t is a critical exponent related to the dimensionality of the investigated system. The extremely low percolation threshold provides strong evidence in support for a fine dispersion of CNTs in the PEK matrix.
A model from the literature, the Li model [Li, J., Ma, C., Chow, W. S., To, K., Tang, B. Z., Kim, J. K.2007. Correlations between percolation threshold, dispersion state, and aspect ratio of carbon nanotubes. Adv. Funct. Mater. 17, 3207-3215] was used to calculate the aspect ratio of CNTs in the nanocomposite from the measured percolation threshold and from the mean length of the CNTs which was taken as1.2 µm this being close to the value of 1.5 µm given by the manufacturer. Thus, the results of analysis suggest that the initial length of CNTs is preserved in the nanocomposites.
Determination of PEEK crystallinity
The main experimental technique employed for the assessment of crystallinity in this project was DSC as discussed earlier. Additional information on crystallinity has been provided using X-Ray Diffraction (XRD).
Three different samples were measured, one pure PEEK as reference (the sample was commercial product from Victrex), one PEEK+0.5wt%CNTs and one PEEK+0.5wt%CNTs which had been strained and annealed. The first sample was in the form of sheet and the other two were in the form of tapes of 0.5-1.0 cm width. An X'Pert Pro X-Ray Diffraction System (PANalytical) was used for the measurements which were in the range of 2θ=10°-35°.
The results show that the position of peaks in the pure PEEK sample agrees with the literature [Goyal, R. K., Negi, Y. S., & Tiwari, A. N. 2005. Preparation of high performance composites based on aluminium nitride/poly(ether–ether–ketone) and their properties. Eur. Polym. J. 41, 2034–2044]. Some of those peaks also appear in the PEEK/CNT nanocomposites. A new peak at 27.3° suggests that the incorporation of the CNTs and processing create new interfacial phases and affect the crystallisation [Wang, L., Weng, L., Song, S., Zhang, Z., Tian, S., Ma, R., 2011. Characterization of polyetheretherketone–hydroxyapatite nanocomposite materials. Mater.Sci. Engin. A 528, 3689–3696]. The degree of crystallinity can also be determined from XRD measurements, but the results are less reliable when compared to DSC, and the main power of XRD is in the study of crystalline structures.
Alignment of CNT in the Nanocomposites
One of the goals of the project was to design, develop and employ special preparation/processing methods resulting in an alignment of CNTs in the thermoplastic matrix. Thus, there was a need from the point of view of characterisation to design and develop methodologies for assessing the alignment of CNTs in the thermoplastic matrix. The methodology used in this work was based on conductivity measurements before and after annealing of samples. Other techniques were used with the electrical measurements in an attempt to confirm the observations made.
The methodology of using electrical conductivity measurements before and after annealing for assessing alignment of CNTs in the nanocomposites is based on the observation that conductivity increases after annealing [Cipriano, B.H. Kota, A.K. Gershon, A.L. Laskowski, C.J. Kashiwagi, T., Bruck, H.A. Raghavan, S.R. 2008. Conductivity enhancement of carbon nanotube nanofiber-based polymer nanocomposites by melt annealing. Polymer 49, 4846-4851], this probably being due to redistribution of CNTs.
Values of conductivity are significantly higher with no calendering. Moreover, the increase of conductivity after annealing is significantly more pronounced in the case of calendering. Three main reasons could explain the remarkable decrease of conductivity after calendering:
With increasing alignment electrical conductivity in the nanocomposites is lost, in other words the percolation threshold increases.
The calendering process may lead to breakage of the CNTs, which leads to a decrease in their aspect ratio. The result of the above is again to increase the percolation threshold.
Mechanical mixture of PEEK and CNT powders may create a structure with PEEK particles covered by the powdered nanotubes. Thus, the conductive phase of the CNTs is located on the surface of the polymer particles. Hot pressing deforms polymer particles and results in formation of a compacted continuous polymer phase, where conductive paths of CNTs are located on the boundaries between particles. After calendering, these paths are destroyed and conductivity is lost.
Thus, the results of ac measurements provide additional evidence that CNTs are aligned in the polymer matrix after calendering and that alignment is partially destroyed by annealing, which leads to an increase of conductivity.
Raman spectroscopy is a well-established technique to study and quantify alignment of CNTs. Repeated attempts to make use of this technique in the case of the PEEK/CNT nanocomposites manufactured during this project were unsuccessful due to the high fluorescence of the PEEK used. Change of the wavelength of excitation could not solve this problem.
In order to establish this technique for the needs of the project, PPS/CNT nanocomposites prepared within the project were investigated. Based on the method of preparation, no alignment of CNTs was expected in these nanocomposites. No differences at the intensity of the G mode at about 1590 cm-1 have been recorded, which provides evidence that the CNTs are not aligned [Haggenmueller, R., Gommans, H.H. Rinzler, A.G. Fischer, J.E. Winey, K.I. 2000. Aligned single-wall carbon nanotubes in composites by melt processing methods. Chem. Phys. Lett. 330, 219-225] as was expected from the conditions of preparation/processing.
Manufacture of Composite Pre-pregs and Laminates
State-of-the-Art Process
The normal approach to the manufacture of fabric pre-pregs is via a powder coating process. This process consists of:
Finely ground polymer powder is first sprinkled onto a fabric.
The powdered fabric then runs through an infrared heated oven, which melts the polymer powder.
Finally, a set of calendering rolls are used to fuse and quench the polymer onto the fabric.
Double sided powder coating is achieved by reversing the single sided coated fabric and repeating the process.
Carbon woven fabric reinforced CNT/PPS
Fortron 0205 PPS reinforced with CNTs was evaluated, together with virgin Fortron 0205 powder which was used as the reference material. Four types of carbon woven fabric reinforced PPS prepreg were initially produced:
Carbon fabric with Fortron PPS 0214 powder. The PPS 0214 is the standard polymer used by Ten Cate Advanced Composites [TCAC] for prepregs and laminates that are applied in aerospace structures. An extensive database has been built up at TCAC, so the PPS 0214 was used as a reference against which the processing of alternative PPS grades could be compared.
Carbon fabric with Fortron PPS 0205 powder, which was used as a reference to compare the CNT/PPS 0205 with.
Carbon fabric with CNT/PPS 0205 on both sides.
Carbon fabric with CNT/PPS 0205 on one side and with PPS 0205 on the other side. It was anticipated that the CNTs would considerably increase the viscosity of the PPS, which might result in poor impregnation of the carbon fabric during consolidation. By adding virgin PPS the chance of achieving acceptable impregnation would be increased. If the CNT/PPS does not migrate into the fabric, then it might improve the interlaminar properties of the woven fabric composite. The CNT/PPS acted as an interply reinforcement in this case.
Laminates were produced using the different prepregs. The quality of the laminates was assessed on the basis of the following aspects:
Visual appearance; the surface of the laminates that were based on the CNT/PPS showed whitish cracks. The original CNT/PPS was black. A pronounced relief of residual stresses was observed in the form of noise during cooling.
Flow of the matrix; the laminates were consolidated between heated platens which allow free flow of the melted resin during the consolidation. The amount of flow is an indication of the viscosity of the thermoplastic resin.
Thickness of the laminate; the thickness is a measure for the state of impregnation or void content.
Ultrasonic damping; the laminates were checked by means of C-scan.
Mechanical performance; the laminates were subjected to three point bending testing, which is generally a good indicator for the consolidation quality.
The impregnation behaviour of the CNT/PPS appeared acceptable. From the thickness measurements and microscopic evaluation, it was observed that the flow of the CNT/PPS into the fibre bundles was the same as found with the pure resin. However, the intralaminar strength of the woven fabric CNT/PPS laminates was very poor: Individual plies could be separated manually. The poor intralaminar strength was confirmed by the response of the ultrasonic measurement. A large amount of damping with respect to a reference C/PPS laminate was measured. Subsequent analysis by optical microscopy revealed cracks, which were located within and around the fibre bundles.
Three point bending tests were performed to quantify the mechanical strength of the laminates which were found to be inferior to those using neat Forton PPS powder..
After discussion with the compounder it was discovered that the CNT/PPS also contained compounding aids, such as stearic acid. It was thought these additives might negatively influence the fibre-matrix adhesion. Subsequent compounds were manufactured without the processing aids.
The next stage of the work was to produce 35 m × 0.5 m wide powder coated prepreg with unfunctionalised CNT/PPS being deposited on both sides of the carbon fabric. The target resin content was 44%.
Laminates were produced using three consolidation cycles these being 10 bar @ 310°C, 20 bar @ 325°C and 30 bar @ 340°C.
The C-scan results indicated that good impregnation could be achieved but the microscopy showed significant crack formation between the fibre bundles. As the laminates cooled cracking noises were heard, the noise being related to the formation of the cracks shown in the micrographs.
The laminates were assessed by three point bending. Four laminates were tested, these being PPS 0205 and 0214 pressed with 30 bar at 340°C and PPS 0205 and 0214 pressed with 20 bar at 325°C.
The results were worse than those obtained previously and this is thought to be due to the crack formation, these being seen in SEM micrographs.
Four laminates were produced with functionalised CNTs, these being pressed with 20 bar at 325°C. The initial visual inspection indicated that the laminate was blacker than had been found with PPS 0205 containing unfunctionalised CNTs. The performance of the laminates containing functionalised CNTs is significantly worse than those containing unfunctionalised CNTs.
Further batches of PPS containing unfunctionalised and functionalised CNTs at varying loadings were converted into pre-preg with resin on both sides of the fabric and a 44% resin content. The laminates were manufactured using two pressing cycles these being 10 minutes at 325°C with 10 bar pressure and 75 minutes at 325°C with 31 bar pressure. The results indicated that whilst the impregnation of the fabrics with the CNT/PPS resin systems was possible, the mechanical results were similar and consistent with those obtained in earlier experiments.
Analysis of the fracture surfaces using SEM shows that following failure, the carbon fibres in the laminates which did not contain CNTs are well wetted with the PPS resin whereas in the case of the CNT containing laminates, the bond between the resin and the fibres appeared to be very poor although some wetting was observed.
The introduction of the unfunctionalised CNTs into the PPS results in crack formation within the laminates. Typically when adding reinforcing fillers to a thermoplastic it would be anticipated that the strength and modulus would increase but the elongation at break would be reduced. With a typical short-fibre-filled compound the fibres would be expected to affect the contraction of the material on cooling, the fibrous filler shrinking far less than the polymer. In the case of the CNT filled materials, the CNT concentration is only 0.5% by weight and so volumetrically it would not be expected to have any significant impact on the contraction of the material on cooling and solidification.
On solidification, the outer layers of the laminate are the surfaces which are directly cooled and so will probably cool and solidify first. Loss of heat from the interior of the laminate is via conduction and so heat loss from the centre of the laminate will be slower. The rate of cooling will also be affected by the heat of crystallisation which is released at the melt-to-solid transition, this slowing the cooling rate. The consequence of this is that the shrinkage due to crystallisation in the central regions of the laminate would be expected to be greater than that found at the surfaces of the laminate. As the central regions of the laminate cool, the outer layers will be cold and possible below the Tg of the polymer, this making the outer layers relatively stiff and inflexible. Inclusion of 0.5% CNT results in approximately 25% fall in elongation at break.
In injection moulding, the typical shrinkage values quoted for Fortron 0205 is between 1.2 and 1.8%. The extension at break found with the CNT reinforced material was 1.6% as compared to 2.1% for the unfilled PPS. The shrinkage of the CNT reinforced material on cooling is therefore of a similar order of magnitude to the extension at break as opposed to the unfilled PPS where the extension at break is greater than the shrinkage. In the case of an injection-moulded component there is usually the scope for the dimensional change to occur thus accommodating the shrinkage. In the case of the laminates, the structure surrounding the polymer will be fairly rigid due to the carbon fibre loadings and thus the surrounding material will be un-yielding and any shrinkage is unlikely to be accommodated by deflection of this material. It therefore seems likely that the result is that the CNT/PPS material fails and cracks appear in the laminate structure.
The mechanical performance of the laminates made with the functionalised CNTs is worse than that with the unfunctionalised CNTs. Reductions in elongation at break can arise for a number of reasons but the two main reasons are that the filler is poorly dispersed or the filler acts as a reinforcing agent for the polymer. In the case of this work, the extensive electron microscopy carried out examining the various CNT/PPS materials indicated that the level of dispersion was generally good and so it seems likely that the reduction in elongation at break is the result of the CNTs reinforcing the PPS. As the reinforcement increases, a reduction in the elongation at break would be expected. In the case of the laminates, the improvement in reinforcement would therefore be expected to result in more extensive crack formation and the poorer performance of the resulting laminate. The inclusion of functionalised CNTs results in a poorer performing laminate than found with the unfunctionalised CNTs and thus it could be construed that their reinforcing effect was improved over the unfunctionalised CNTs.
Micrographs of fracture surfaces suggest that the strength of the bond between the PPS and the carbon fibres has been reduced in the case of the CNT containing materials. The strength of a carbon fibre reinforced polymer is largely determined by adhesion of the polymer to the fibre. In the case of thermoplastic matrices, this adhesion is more of a mechanical nature than a chemical bond:
The thermoplastic resin shrinks around the reinforcing fibre. Thermoplastic resins solidify at high temperatures, so they show a significant volumetric shrinkage, typically around 2%. This large shrinkage results in a volumetric lock of the resin onto the fibres but this may also result in significant residual stresses.
It is assumed that the polymer molecules creep into the micropores on the fibre surface. It has been shown that an oxidised, rough fibre surface gives rise to better mechanical properties of a C/PPS composite.
As discussed, the reinforcing CNTs reduce the strain to failure. The effect of the CNTs on the shrinkage of the polymer was not determined for the current PPS and PEEK polymers. Whilst the CNTs are present at a small concentration it might be assumed that the CNT network decreases the volumetric shrinkage found as the polymer cools and solidifies. Hence, the shrinkage stresses on the fibre are reduced, and therefore the pressure which drives the development of the interface and the final mechanical adhesion of the polymer onto the fibre is also reduced. The effects of a weaker interface and a less ductile matrix reinforce each other as, as soon as the filament becomes detached from the matrix, the local stress concentration increases and microcracks will then grow from this stress concentration to release the residual stress.
A second effect might be that the CNTs reduce the mobility of the thermoplastic molecules in the fibre-matrix interface. The increase in the melt viscosity on the macroscale is limited, but the effects on the micro- or nanoscale are not known. If the CNTs hinder the diffusion of the PPS polymer into the micropores on the fibre surface, then the negative effect becomes stronger with improved dispersion of the CNTs.
Carbon woven fabric reinforced CNT/PEEK
The preparation of the functionalised CNTs resulted in them being attached to PEEK powder particles with a d50 of 10μm. This material was processed using a twin-screw extruder to produce a compound where the CNTs were bound into the polymer. The compound was ground down to produce a powder with a d50 of approximately 1000μm.
The two batches of material, one with functionalised CNTs the other with unfunctionalised CNTs were converted into pre-preg. The materials were coated onto a standard 5H satin carbon fabric, this being the same type of material that had been used with the PPS materials.
The laminates produced showed acceptable C-scan results and did not exhibit the cracking noise found with the PPS materials when cooling down in the press.
The mechanical properties of the CNT containing laminates were again inferior to that of the laminates which did not contain CNTs. The PEEK containing unfunctionalised CNTs produced laminates with a higher flexural strength than that containing functionalised CNTs but the flexural moduli of the materials was similar.
The inference drawn from these initial results is that a similar process is happening with the CNT/PEEK matrix as occurred with the CNT/PPS matrix. However, the ductility of the PEEK materials is better than that of the PPS materials and so cracks do not form in the laminate. However, due to the contraction occurring during cooling and solidification, significant residual stress is set up within the polymer rich zones of the laminate and this results in underperformance of the laminates containing CNTs. The performance of the PEEK containing functionalised CNTs may also be being affected by potential degradation of the resin as discussed earlier.
CNT/polymer UD tape
The manufacture of unidirectional (UD) reinforced CNT/PPS and CNT/PEEK tapes was investigated in parallel to the woven fabric reinforced CNT/polymers. Two approaches were used.
The first approach involved modifying an existing UD prepreg technology based on polymer powder. In this process fibre yarns are pulled through an aqueous dispersion of very fine polymer powder. The powder particles migrate into the yarns. After drying, the polymer powder is melted and a prepreg is consolidated. Any increase in the viscosity of the resin system due to the inclusion of CNTs would only make a small difference to the process as the polymer particles are already amongst the filaments and, therefore, polymer flow during impregnation and consolidation is very limited.
Whilst this approached appeared to have advantages, the principal condition for the dispersion impregnation is that the powder particle size and distribution meet a narrow specification. The grinding of tough polymers down to the required particle size is very difficult and this proved to be a barrier to the application of the process.
The second approach to producing UD tape with CNT/polymer was based on melt extrusion. An existing polymer extruder was retrofitted and equipped with a crosshead die. In this process fibre yarns pass through the die, where the polymer melt is forced in between the filaments. The level of impregnation is determined by the yarn line rate (or residence time), the polymer viscosity and the melt pressure in the die.
Reasonable impregnation was achieved. However, the throughput rate was low, of the order of 1-3 m/min and the residence time in the die at the impregnation pressure was too short. Significant redesign of the impregnation head was required and so this approach was discontinued.
Novel Dry-Powder Impregnation Technique
The development of a new impregnation technique included evaluating the use of an existing D-Preg technology and comparing the products manufactured using the new technology with ‘standard’ material derived through powder coating and film stacking technologies.
In the D-Preg process the polymer is applied in powder form and is forced through the reinforcement fibres via a high voltage alternating electric field. The process has the capacity to distribute powder form materials in all kind of porous structures (nonwovens, fabrics, papers, foams…). This technology works with insulating powders such as thermoplastic or thermoset and conductive ones as carbon black.
Process improvements were made but due to the electrical conductivity of carbon fibre reinforcement, the D-Preg process was found to be inefficient and a new process was developed. The conductive fibres create a local zone without any electric field and so inhibit the powder impregnation.
One other problem associated with the process was the requirement for fine powders with a d50 of around 20-25μm. In the case of granular PPS and PEEK materials, grinding down to this d50 was extremely difficult and expensive.
The carbon fibre conductivity prevents impregnation due to a Faraday Cage effect and the contact between carbon and powder doesn’t allow nebulisation of powder (cloud). The fibre fineness and sizing increase the difficulty of impregnation. The initial stage of the work involved a literature search based on:
dry technologies (no suspension, no solvation)
based on powder (no film, no fibres, no liquid)
suitable for 1, 2 or 3 D reinforcement (not only for UD or tape)
Of the possible routes considered, only the fluidised bed and mechanical vibration were deemed to be suitable techniques.
Trials were initiated with fluidised bed and mechanical vibration technologies. Fluidised bed impregnation resulted in a simple or double face coating without any migration of particle between fibres. As a consequence efforts were concentrated on mechanical vibration technologies.
An alternative technique was developed based on mechanical vibrations at laboratory scale. This process should allow the opening of the fibres and the impregnation by the powder. In a first stage, PA12 powders were used in order to determine the most suitable powder particle size. Particle sizes of 20 µm allowed higher final powder content, but this is still limited to 22% w/w, into carbon fibres.
Other parameters were investigated these including:
Time of impregnation
Gap between vibrators
Amplitude of vibration
All the parameters were optimised to produce a carbon fibre/PA12 pre preg, the properties of which were compared to product manufactured by the simple scattering technique. The flexural properties of samples impregnated mechanically were slightly improved in terms of moduli and strength.
All the parameters were reviewed and optimised to produce PEEK and PPS matrix pre-preg and plates for mechanical characterisation.
The following trends were highlighted:
The maximum powder content (PPS or PEEK in carbon fabric) was limited to 20% w/w
The main parameters were time and amplitude of vibration
Fluidity of the powder is important (too fluid = powder expulsion; not fluid = no impregnation )
After powder addition and compression moulding, composites based on powder coating and vibration were compared using flexural tests, as illustrated in Figure 90.
It was found that the modulus and maximum strengths were similar for all the impregnation technologies.
This result was however, still far from the target of 42% w/w of resin content for the woven carbon fabric.
The development work was divided into five steps:
Powder preparation: with and without anti caking agent (0 to 0.5% w/w in powder)
Impregnation with different parameters along with the characterisation of the materials
Thermal fixation
Compression moulding
Characterisation
The maximum powder content achieved was 36-37% w/w, this being better than in the previous trials. The powder content is proportional to the impregnation time and process parameters. The set-up of ‘opening system’ allowed the improvement of the powder content into the semi-preg. The use of fine powder, d50 = 20 µm, resulted in better impregnation and higher powder content.
It was concluded from the work that powders with a d50 up to 50µm had no influence on the result matrix/fibre ratio. However, PPS+CNT powder with a top cut of 255µm was too coarse to be impregnated in any significant quantity into the carbon fabrics.
The project objectives were not only based on the manufacture of woven fabric materials but also on unidirectional materials (UD Tape). Trials were carried out with the new technology in order to impregnate glass roving and carbon tow.
In order to supply samples to the partners and to validate the potential of the technology on an industrial scale, the impregnation process was upscaled. The new pilot line was supplemented with a tunnel furnace and peripheral devices.
New improvements were also implemented during these trials:
The new system allowed the impregnation of carbon tows at high speeds up to 20m/min
Multiple tows could be impregnated simultaneously
The powder content could be adjusted and was a function of the system parameters
A PEEK content of 35% wt could be reached easily with a very homogeneous distribution of powder between the carbon filaments.
Due to this high level of impregnation quality, the optimisation of moulding cycles in terms of time and pressure was carried out, the results being compared with standard commercial products. In all the cases, the semi-preg produced using the new technology had equal or better modulus than the commercial products.
A few technical limits were found with the new technology, these being:
The impregnation in the warp and weft directions in woven fabric was quite different. Higher resin levels were found in the warp fibres.
The equipment in the line could not achieve the dimensional tolerances required by other partners for tape laying with in –situ consolidation. This was not due to a fundamental problem with the new technology but more a limitation of the equipment that was available and so this could be resolved.
The next stage of the work was to carry out a study comparing existing pre-pregging technologies, such as scattering or film stacking, with the new impregnation technology. This work was based on an unbalanced carbon woven fabric (88% warp 3k carbon fibre, 12% weft glass fibre) impregnated with neat PA11, PPS and PEEK.
In the case of PA11, the new impregnation technology was compared to the double scattering. The products obtained were compression moulded into laminates and were characterised by 3 point bending tests. The influence of the pressure and time for the compression moulding process on mechanical properties was evaluated.
Irrespective of the moulding pressures or times, the semi prepregs produced using the new technology showed better mechanical properties for the same fibre content. At very short moulding times at very low pressures (1 or 3 bars), the improvement was linked to use of the new impregnation technology which resulted in an improvement of 50% for strength and 13% for modulus.
In the case of PPS and PEEK, the new impregnation technology was compared to the film stacking method. The same procedure used previously was followed. In the case of PPS, the new technology generated an improvement of 44% in terms of the maximum flexural strength and 14% in terms of flexural modulus for a short compression cycle of 5 min at 5 bar, the moulding temperature being 310°C.
In conclusion, the new impregnation technology has the following advantages:
Dry technology without the use of any solvents
Fast, one step processing
Better fibre opening and matrix distribution in composites compared to traditional technologies
Better control of resin flow due to the improved powder distribution between filaments
Higher mechanical properties at the same moulding conditions
The use of CNT reinforced materials for the manufacture of semi-pregs/UD tapes was not possible due to the problems related to grinding the compounded materials. Initial work was carried out using uncompounded materials but there was a significant health hazard due to the release of CNTs due to the vibrational nature of the process.
Novel Laminate Manufacturing Methods
Two techniques were investigated, these being internal and external bagging.
Internal Bagging
The internal bagging process has been evaluated using bags made from PEEK film, the bags being produced from flat film using a laser based welding process. In order to weld the bags a black film has been sandwiched between two ‘natural’ films, the black film acting as a radiation absorber.
An inflation adaptor was manufactured for the bags, this fitting into the inflation hole. The aim of the design was to have a system where bags could be supplied in a lay flat form and inflated at the point of use. Following inflation, the necks of the bags were welded and the inflation assembly cut off so that the adaptor could be re-used.
The mould was composed of two box sections with simple clamps to close and lock it. The individual bags were designed to fill one half of the tool in order to ease assembly of the two halves of the mould. The interior of the mould was coated with Frekote 55-NC to prevent the polymer sticking to the mould. The layers of 280 g m-2 5-Harness satin carbon fabric semi-preg were interlocked at the corners.
The layers of pre-preg were generally well bonded in the finished component although the external definition of the cube needed to be improved. The wetting of the fibres was very good when considering that the composite material was a semi-preg where there was virtually no impregnation of the fibre bundles prior to the bagging process. The wetting of the fibre, whilst not being perfect, is very good.
The limitation with the process appears to be the ability to pressurise the bags to a sufficiently high pressure without them failing. Currently levels above 1 bar internal pressure are difficult to achieve and pressures of 5 bar would be the ideal.
External Bagging
The external bagging technology uses a 200mm x 200mm hot plate tool from Surface Generation The hotplate has sixteen active heat/cool zones and uses an air heating system.
Laminate consolidation is good with UD tape but poor with fabric materials. This is due to the low forming pressure which is not sufficient to cause the impregnation of the fibre bundles with a semi-preg material.
Within the project there is a need to manufacture honeycomb panels which have skins. Experiments have been carried out to examine the possibility of in-situ bonding of PEEK film honeycombs to laminates as the laminates are being manufactured. The top surface of the composite component is the PEEK bagging film and when this melts it is possible to apply the PEEK honeycomb in a manner where the honeycomb melts locally at the film surface so forming a bond. Figure 107 shows the method being employed and the results have been very successful. The process was one where the laminate was formed as per the cycle described earlier with the exception that the honeycomb was mounted onto the panel when the PEEK was molten, a load of 0.001N/mm2 applied with the system held at temperature for 30s and then the system was cooled to room temperature.
The Evaluation of the Properties of the Laminates
The laminates evaluated were produced using a variety of techniques, these including:
Direct compression moulding of pre-pregs as discussed earlier in this report.
Compression moulding of the combination of dry carbon-fabric and polymer films, referred to as a film stacking technique.
Manufacture of Laminates by Film Stacking
Initial laminates were produced using 75μm PEEK film, the approach. The samples were produced by compression moulding at a pressure of 7 bar with a heating rate of 2°C/min and a 30 minute hold at 400°C.
The laminates produced using 8 plies with a +45/-45lay-up were of good quality as evidenced by ultrasound examination.
Laminates produced using 10 plies with a 0/90 lay-up and with a PTFE film between plies 5 and 6 were not as good, as measured by ultrasound. A PTFE film was inserted into the lay-up in order to prevent bonding through the thickness of the laminate so that samples could be cut to determine G1c.
Attempts were made to produce laminates using PEEK/CNT films pressed at Victrex but the initial batch of compounded material was degraded. This resulted in films which contained voids and the material viscosity that was too high, degradation involving cross-linking, for impregnation via the film stacking method.
Due to the lack of suitable PEEK/CNT film, suitable materials were sourced from outside the consortium. The films purchased were a reference PEEK film with no CNT which was around 110µm thick, a CNT containing film, NS10295, with 1.5% CNT type 1 which was around 75µm thick and a further CNT containing film, NS10309 film, with 3% CNT type 2 and a thickness of around 91µm.
Using Differential Scanning Calorimetry, the CNTs were found to have no effect on the melting point or the crystallisation of the PEEK. The inclusion of CNTs resulted in an increase in viscosity as shown in Figure 111.
Examination of the films using microscopy indicated that some CNT agglomerates existed.
Laminates
In addition to the materials mentioned previously, two further materials were purchased, these being:
UD tape based on IM7/PEEK, 220g/m² prepreg area weight, 34% resin content by weight, 145g/m² dry fibre area weight
UD tape IM7/PEEK with CNT, 216g/m² prepreg area weight, 34% resin content by weight, 143g/m² dry fibre area weight
A 5H fabric/PEEK without CNT, 280 g/m² dry fabric, 38% resin content was supplied from within the consortium.
Fracture Mechanics
G986/PEEK and IM7/PEEK Laminates
Testing was performed in accordance with Airbus specification AITM 1.0005.
The introduction of CNT into the G986/PEEK film composite slightly reduces the G1c values whatever the testing temperature. No increase was observed in the values of G1c following the addition of CNT as was originally expected. The G1c value for G986/PEEK film laminate is higher than that for the UD tape IM7/PEEK as expected when laminates made from UD tape and fabric are compared. The result obtained with semi-preg G986/PEEK is lower than would be expected and this may have been due to the presence of porosity in the composite.
PPS Matrix Samples and G986/PEEK – Samples 5381-5384
The interlaminar behaviour of laminates based on three thermoplastic matrices was investigated using double cantilever beam (DCB) , Mode I, test and the end notch flexure (ENF) , Mode II, tests. The laminate lay-up was a 12-ply [(0,90)3,(90,0)3]S construction and the matrices were:
PPS: Fortron 0214 [3];
PEEK: Poly Ether Ether Ketone, Victrex 150P [4];
PEEK+CNTs: PEEK Victrex150P reinforced with 0.5 wt% of non-functionalised (“naked”) Nanocyl NC7000 CNTs
G1c values were determined from the peak values from the force-displacement tests.
Compression After Impact Tests
The drop weight hemispherical impactor used was 16 mm diameter and had a mass of 5.5 kg. The specimen is clamped on a base plate with a cut-out of 75mm by 125mm. The specified ratio of impact energy to specimen thickness defined in the relevant standard is 6.7 J/mm, the drop height being calculated based on the average thickness of the samples.
The extent of the damage resulting from the impact was determined using ultrasonic examination and the results are given in Table 22.
The compressive strength data was normalised to remove any effects related to minor changes in the volume fraction of the matrix. The results indicate that, in general, a smaller damaged area results in improved compression performance. However, the data for the matrix containing the CNTs does not fit this pattern and it is thought that this is due to local delamination in the outer layers of the laminate.
Tensile Properties
In the warp direction failure happens perpendicular to the loading direction. PPS 0205 is a higher flow [hf] resin that PPS 0214. The CF/PPS(0205)/CNT delaminates extensively on the way to failure.
Acoustic Emission (AE) measurements and digital strain mapping were also made during the tensile tests.
Shear Properties
Shear properties of the laminates was evaluated using a specimen with a tensile +/-45°, tests being carried out according to the Airbus standard document AITM 1.0002.
There is no difference in the shear modulus generated with different pre-preg materials. However, the shear strength does show a strong dependency on the type of pre-preg used. The results are ranked with deceasing properties from UD tape, to semi-preg and G986/PEEK film.
The addition of 1.5% of CNT in the G986/PEEK composite significantly increases the shear strength of the laminates but the addition of 3% of CNT in the G986/PEEK results in no improvement in the shear strength when compared to the results without CNT. There is however an improvement in shear modulus following the addition of 3% CNT but the increase is less than that found following the addition of 1.5% CNT.
In the case of the IM7/PEEK laminates, the addition of CNT significantly reduces the shear strength but does not affect the shear modulus.
Peeling Tests
The influence of the temperature and time of consolidation for the test specimens was investigated in order to optimise the processing parameters.The addition of CNT to the IM7/PEEK prepreg has only a limited effect on the peeling load.
The consolidation time does not influence the peeling load level. The heating and cooling rate in the autoclave is 2°C/min, the complete heating and cooling cycle taking approximately fifty minutes. The zero minute cycle has undergone the heating and cooling process, the zero referring to the time the autoclave was held at 400°C prior to the cooling stage being initiated. The data clearly indicates that the heating/cooling cycle in itself is sufficient to generate a good bond between the plies. As with the consolidation temperature, the addition of CNT does not result in any significant change in the peeling loads measured.
The Development of Sensor Technologies for Monitoring the Health of the Multi-Scale Materials
A structural health monitoring system based on the swept-wavelength coherent interferometry technique has been successfully demonstrated. Several methods for the surface application on composite materials and also techniques for the integration of distributed fibre optic sensors (DFOS) were tested and evaluated. For surface applied DFOS the investigation of suitable materials mainly focused on the adhesive and the fibre coating material, but also extended to appropriate surface preparation of the composite structures in order to ensure a reliable strain transfer. In order to evaluate the performance of the sensors, DFOS with polyimide coating were applied on the surface of VICTREX® PEEKTM 450G test samples. For integrated DFOS, the adhesive is no longer necessary, but the optical fibre has to withstand the process conditions during the manufacturing of the composite structures. In case of Poly-Ether-Ether-Ketone (PEEK) this means the DFOS have to withstand a temperature of 385 °C and a pressure of 10 bars which makes high demands especially on the fibre coating due to the fact that it is the sole protection of the sensitive optical fibre. Therefore, silica optical fibres with copper-nickel coating CU1300 and high-temperature polyimide coating were chosen for the integration into TenCate Cetex® TC1200 PEEK AS-4 test samples.
The DFOS and the measurement technique have been shown to provide excellent results during fatigue tests containing 10 million load cycles for DFOS applied on the surface and integrated into composite structures. The DFOS were able to detect crack formation as well as determining local strains along the axis of the fibre used. As a result, it has been demonstrated that DFOS are reliable structural health monitoring sensors during their estimated service lifetime for the application on the surface of composite structures and the integration into composite structures under process conditions.
Furthermore, the DFOS were applied to composite engine stiffeners that were developed and manufactured within this project. Based on the FEM-simulation data from CIMNE, the maximum load for the mechanical tests was determined as 200 N. In order to monitor the engine stiffeners, DFOS and displacement sensors were used in the test setup.
The position of the DFOS was determined at the position where the highest strain was expected according to the FEM-simulation.The derived displacement and strain values differ from the expected values from the FEM-simulation which is mainly caused by a significant difference in thickness of the real composite stiffeners compared to the simulated stiffeners.
Carbon fibre reinforced polymers (CFRP) are multifunctional materials and they can be also used as sensors themselves for strain and damage for structural health monitoring. This self-sensing concept has the advantages of being low-cost and applicable to a large volume of structural material with absence of any mechanical property loss. The concept is based on monitoring the changes in electrical conductivity in order to detect the onset, nature and evolution of damage in advanced polymer-based composites. Carbon fibres are highly conductive making the composite also a conductor. Electrical conduction in CFRP can occur by two means:
By current flow in the longitudinal direction along the carbon fibre
By conduction in the transverse and through thickness direction due to contacts between the fibres.
In the longitudinal direction, the conductivity is influenced by the number of conducting fibres while in the transverse direction conducting properties are related to the density of fibre to fibre contacts.
The composites were prepared using UD carbon fibres (Hexcel IM7) at a volume content of 38% in a matrix of PEEK. Measurements were performed on samples with length 60mm, width of 10mm and thickness of 0.15mm.
Three different configurations were used for the measurement of resistance and voltage change. For the measurement of longitudinal resistance a four-wire method was used by applying a fixed current through the two outer electrical contacts and measuring the voltage between the two inner electrodes. The configuration for through thickness resistance measurement involves a current applied in the two opposite contacts and the voltage drop is measured from the same wires. The configuration for oblique resistance measurement has two contacts in the oblique direction that are used for applying a fixed current and the voltage drop is measured from the same wires. In this configuration a second voltmeter is used in order to measure the voltage drop between two opposite contacts in the through thickness direction in the middle of the sample.
Resistance increases monotonically with strain from the minimum strain. The increase of ΔR/Ro is more than 10% at fracture. In the curve corresponding to the second sample a steep increase of resistance at 150s could be observed, this being an indication of fibre breakage.
A slight increase of ΔV/Vo in the oblique direction is observed up to 0.6 % of strain which becomes more profound for higher strain up to fracture. As concerns ΔV/Vo of through thickness in the middle of the specimen a negligible change is observed up to 0.8% strain followed by a decrease up to the fracture.
In all the experiments using different configurations an increase of resistance change upon loading was observed when voltage drop was measured by the same contacts which served for current injection. In general, the increase in the resistance during tensile loading is ascribed to:
The load-dependence of the surface resistivity
Geometrical change
Destruction of the conductive paths such as fibre breakage and the reduction in the contact area of each fibre due to deformation and/or damage.
At low strains the effect of (i) and (ii) are dominant while (iii) is more pronounced at higher strains. The results indicate that measurement in the longitudinal direction is more sensitive to fibre breakage while measurement in the through thickness direction leads to higher ΔV/Vo upon tensile loading.
For the oblique measurement an increase of the ΔV/Vo up to about 5% is observed. After the first load cycle the viscoelastic deformation results in an increase of resistance change at the unloaded state. The above finding implies that there is an irreversible damage during the first cycle that could be attributed to the permanent damage of the matrix. However, the ΔR/Ro is reversible in the consecutive cycles. The resistance after the second cycle is recovered back to the level after the first cycle and a slight decrease of resistance level at the minimum strain is observed. The above behaviour is also observed when the maximum displacement of every cycle is set to higher values. Contrary, the ΔV/Vo measured at the through-thickness direction shows a decrease upon increase of stress. Furthermore, a slight decrease of ΔV/Vo at the minimum loading is observed that may be attributed to relaxation processes.
Summarizing, the results reported above suggest that longitudinal resistance is more sensitive to fibre breakage while through thickness resistance gives higher sensitivity for strain sensing. In addition, oblique resistance changed irreversibly during tensile cyclic loading at various stress levels, which could be attributed to matrix damage.
Development of the DIRIS Honeycomb Structures
The DIRIS (Directionally Reinforced Integrated Stretched Single yarn) honeycomb (European Patent 1950034) is a composite, isogrid sandwich panel reinforced (up to this project) with continuous CFs, along all of its principal directions. The main function of the sandwich core is to provide resistance against in-plane and out-of-plane shear loading.
The preform, which provides the panel with resistance to in or out-of-plane shearing forces and also tensile resistance to forces acting in the plane of the preform, is formed by winding a single UD reinforced tape (typically 2mm x 0.5 mm thick) in successive layers in the directions of 0°,+60°, -60°, thus forming equilateral triangles as shown in Figures 144 and 145. Triangular prismatic cells, made of the same reinforced material, are placed into the triangular gaps created within the preform this being done such that the fibres run in a direction inclined at 45° to the edges of the prism. The triangular prismatic cells are fabricated by winding and simultaneous bonding of the UD tape, using temperature and pressure, with the fibres running at an angle of ±45°. Afterwards the prismatic cells are cut at their precise length and folded at one end so that they form one of the bases of the triangular prism. The bonding of the adjacent layers which are in contact is achieved through the simultaneous heating and pressing of the contact area, causing local melting and bonding of the thermoplastic polymer. Ultrasonic melting-bonding of the tape and of the cells is also used.
The CNT/PEEK tape used was reinforced with 0.5% wt CNTs and 40% wt AS-4 carbon fibre. The CNTs were ‘bonded’ on the PEEK grains as described previously and the long CF and the CNTs/PEEK grains were mixed and placed in the channels 750 mm long and compressed into tapes using the tape thermo-compression machine as described previously.
SEM images indicated that the fibre impregnation was almost complete in some places but in other parts voids were detected. The high pressure and the limited melt shearing along the fibre direction were clearly not sufficient to fully impregnate the fibres.
The multi-reinforced tapes were used to prepare three preforms and then three honeycomb cores. The component parts were joined by ultrasonic welding. The cores were 160.5 mm long, 69.8 mm wide and 12.6 mm thick. They weighed approximately 33 g (equivalent to a density 0.22 g cm-3), the dimension and weight being the same as those used in the previous projects in order to provide a comparison with the earlier results.
Shear testing of the specimens followed the ASTM C 273-00 standard test method. The loading rate was 0.2 mm/min. All experiments exhibited a linear elastic behaviour up to about 3.7 MPa and then yield started. After reaching a well-defined peak, the stress decreases with clear evidence of internal honeycomb destruction.
The observations suggest that the multi-reinforced core shows a higher resistance to in in-plane shear than those reinforced only with carbon fibre. Since the carbon fibre impregnation was partially satisfactory, firm conclusions about the extent of the improvement of the core performance, arising from the additional CNT reinforcement, cannot be reached.
The use of the dry impregnation technology developed within the project was investigated as a route for the manufacture of DIRIS honeycombs by preparing a preform in a suitable plastic mould and impregnating it with PEEK powder. The electrostatic repulsion of the charged fibres results in the fibre network opening, thus making the penetration of the powder particles relatively easy. The distribution of the powder on the fibrous structure is rather homogeneous and the impregnation process rapid (60 m/min). In practice, it was found that the powder penetrated into the preform structure, preferentially at the cross links of the fibre network, which is advantageous for the core strength.
In general the powder covered all fibres but several drawbacks were observed:
The powder diameter should be much smaller than the filament diameter (~ 10 μm) in order to ensure that the filament surface is covered with enough powder and the coverage is homogeneous. However, reducing the powder size to a few microns is an expensive procedure and time-consuming procedure.
The powder was loosely bonded on the fibres and therefore a consolidation process should follow.
The technology could only be applied to glass fibre and was unsuitable for carbon fibre.
This procedure could only be applied to the preform in several steps, this leading to an increased cost. The difficulties becomes more severe when the semi-closed cells are incorporated.
All these observations led to the conclusion that dry impregnation is not yet a suitable method for application in the manufacturing the DIRIS core/panel.
Following the limited results using the dry-powder impregnation technology the concept of toothed rolls was examined. The idea was the production, by two counter-rotating heated rolls, of a sheet with an anaglyph with a periodic geometry like the DIRIS core. The concept was one where the rollers acted as two toothed wheels that never come in contact with each other but instead left a small gap between them such that molten matrix could penetrate around a spreading sheet of fibre filaments placed between the wheels. Obviously the thickness of the core produced could not be constant at the tolerance of the flat thermoplastic sheet but the aim was to attempt to vary within a small acceptable range. The rotation of the rolls was initially numerically simulated from 0° to 9° in order to examine whether a core production similar to the one of the DIRIS architecture would succeed.
Two toothed geometries were initially examined. The first core geometry investigated was ellipsoidal and the calculations indicated that 52 teeth were required on each roller.
Melt impregnation technologies have been developed over recent years and Ticona offer a commercial material, Celstran, impregnated in this way. Molten thermoplastics have a high viscosity which tends to make impregnation difficult and normally the fibre tow is spread in order to ease the difficulty of impregnation. Tow spreading is often brought about using a series of bars in a chamber filled with melt, the fibre running over the bars and so spreading.
The problems associated with the various techniques include:
They are restricted in their ability to spread the fibres.
The process involves the use of relatively high tensions and this, coupled with the mechanical interaction of the fibres with the bars, can result in fibre damage.
There are restrictions in the width of the sheet of fibres.
The analysis indicated that the toothed-rolls could sufficiently deform a thin sheet of fibres and can generate an anaglyph resembling the DIRIS core architecture. The next stage was to examine whether the structure of a sheet of fibres sheet can follow the steepness of the proposed anaglyph and how the filaments can be spread to facilitate impregnation by melt. This requires:
Each filament to bend at a slightly different angle (upwards or downwards), according to the anaglyph.
To fully impregnate each filament with the molten matrix (PEEK or CNT/PEEK).
The first requirement can be met if:
Each filament can slide independently to the adjacent one.
The total contraction (in order to follow the anaglyph) of each filament over a specific length (a period) is equal for all filaments. This means that although the adjacent filaments may bend locally to a different extent, at the end of a specific length they should be equally contracted.
In general, the adjacent un-coated filaments of an un-twisted fibre can each slide upon one another.
A mathematical model was developed which maximised the permissible differential sliding along the axis of filaments in order to enhance the vertical gap between adjacent filaments. The main conclusions from the mathematical model are:
It is feasible to treat a filament in a toothed-rolling process without inducing normal differential strains in both spatial directions of an orthogonal fabric. Therefore the filaments can slide independently in the X or Y directions and the compressed sheet will be in full contact with the teeth anaglyph.
High contraction values can be achieved through this procedure, depending on the period and the amplitude. The filament contraction ratio is a metric of the relative sliding of the filament-resin system with respect to the inertial frame of reference of the die.
High relative shearing can result in both spatial dimensions, depending on the geometrical form of the 2D periodic function. Local shear strains can attain very high values (up to 1200%) in many areas depending on the steepness of the slope. The mean absolute shear strain along the surface can also attain high values (up to about 800%).
The design and production of the tooth-rollers, their incorporation in the pultrusion chamber and the calibration process would require prolonged and extended work. In order to investigate the concept and its numerical interpretation the generation of the anaglyph on two flat surfaces of a mould was evaluated during the present project.
The anaglyph was produced on two prismatic pieces of stainless steel by the use of a CNC OKUMA cutting machine. After polishing the surfaces, a thin orthogonal sheet of glass yarns was placed on the one part and compressed by the other. Although the yarns slid differentially along their axis (X or Y) and were also separated, the filaments within each yarn were slightly affected. Since the aim is to separate the filaments, the steepness (the gradients) of the anaglyph needs to become greater and the process would also be helped by using orthogonal sheets with a smaller number of filaments per yarn.
An alternative is to use two thin UD sheets, placed in an orthogonal manner, one pretensioned along the axis of the (pultrusion) chamber.
This work is continuing and involves a German University.
Recycling of Multiscale-Reinforced Thermoplastics
Most recycling is ‘open-loop’ in that the recycled material does not go back directly into the same end-use. However, in plastics factories it has been common practice for many years to practice closed-loop recycling where such things as sprues and edge trims are ground and fed directly back into the manufacturing process. Recycling within a factory can result in significant cost savings both in terms of raw materials but also in respect of disposal.
The materials being used within the M-RECT project have added value over and above the value of the raw materials. It therefore makes sound environmental, technical and commercial sense to reprocess and reuse these advanced materials.
The polymeric component of the composite is as valuable, if not more valuable, than the carbon fibre and so it is logical that the materials would be recycled into injection mouldable compounds. Because thermosets cannot be reprocessed, the normal route with these materials is to degrade and pyrolyse the thermoset resin to leave the carbon fibre.
Separation of the CNTs and polymers is excluded as a method of recovery on the bases:
PEEK and CNTs are approximately equivalent in terms of cost/kg. Thus there is no financial advantage in separation to recover one component which involves the destruction of the other. PPS has a significantly lower value than PEEK so there may be some commercial advantage to removing the PPS.
Cleaning PEEK and PPS contaminated machinery is best done by thermally degradation. In the case of PEEK, temperatures of the order of 600-650°C are used for this process. If thermal degradation were to be used to recover the CNTs then the temperatures being used would almost certainly result in damage to the CNTs. Thus, the value of the residual CNTs would be very small and recycling the CNTs would result in low value materials which potentially had very poor properties.
The PEEK and PPS could be removed by chemical means but both materials exhibit good chemical resistance and so this approach would not be easy. In the case of PEEK the route would probably be to dissolve the PEEK in concentrated sulphuric acid. This would involve safety issues plus the residue of sulphuric acid would need to be recovered. In principle this would be possible but the costs would be very high.
Filtration to remove the CNTs from the polymers would be very difficult given the dimensions of the CNTs. Extremely fine filters would be required and these would be expected to block quite quickly so making the process very difficult and expensive.
The materials produced in the project are suitable for recycling into injection mouldable compounds. Previous work has shown that APC-2 (unidirectional carbon fibre tapes impregnated with PEEK) can be readily re-processed to produce good quality short-fibre reinforced materials4.
Separation of the individual components would be very difficult to achieve and so it seems logical to reprocess the CNT containing materials directly into mouldable materials. The basis of this document is that the materials produced in the project are suitable for recycling into injection mouldable compounds.
Principles of Waste Collection
Recycling of post-consumer waste introduces two significant problems, those of separation and cleaning. Ideally, any in-factory process should be clean and should also offer the benefit of not having to separate as waste can readily be segregated at source. However, there will potentially be cases where cross-contamination of waste streams does occur and so measures must be put in place to prevent any problems resulting.
Cleanliness
Normally, lay-up of composite pre-pregs occurs in a clean-room facility and so contamination of the materials up to the point of pre-preg lay-up should be minimised. Collection of any waste needs to be in appropriate sized containers fitted with lids to stop subsequent contamination.
Segregation
In the case of pre-pregs there are a number of variants possible in respect of the resin, for example PEEK 90 and 150 grades, and fibres, for example AS4 and T700S, used, and where possible, the waste from each variant should be stored separately.
The elevated processing temperatures for engineering thermoplastics such as PPS and PEEK mean that it is vital not to cross-contaminate them with other thermoplastics and thermosets. Cross-contamination will result in the lower performance polymers degrading and black residues being formed. Good quality recyclate can only be produced from good quality waste.
The basic principle should be to ensure that measures are in place to keep waste stream separate in such a manner as to limit any potential cross-contamination. Effective segregation at source produces a high quality waste stream which in turn produces high quality recyclate.
Transportation
Transportation is an important cost in the recycling process. It is unlikely that any one factory will use sufficient quantities of the thermoplastic composite to warrant the installation and running costs for the conversion of the waste into injection moulding compounds.
Any recyclate supply chain must be sustainable in that potential customers will require stability of supply. It therefore would seem to make commercial sense to have a limited number of sites for compound manufacture, these sites also forming the sales and marketing organisation for the materials.
Size Reduction
Size reduction in the plastics industry is usually carried out using granulators/grinders which essentially consist of sets of rotating and stationary blades between which the material is trapped. Size reduction of the material is usually achieved not by cutting the material as the two blades pass each other but rather by fracture arising from the impact as the material is trapped between the blades.
The styles of machine used for films and for thicker solid materials, such as composite laminates, do differ. Generally film type materials are often converted into compressed pellets following grinding in order to improve their ease of use in subsequent stages of the recycling operations.
In the case of large composite components, the primary size reduction would probably be via water jet or laser cutting followed by size reduction in a standard granulator system. Smaller components could be fed directly into the granulation stage.
Economics
Whilst the quantities of material will probably be relatively small, certainly in the early stages the adoption of the technology, the value of the materials and, more specifically, the value of the product derived from them makes it worthwhile recycling any waste as long as the waste streams are free of contamination. In the case of the PPS composites the base resin cost is slightly less than the cost of the carbon fibre and so recycling may be more marginal than with PEEK where the resin cost exceeds that of the carbon fibre.
Reprocessing of CNT-filled Polymers
Material
Materials containing functionalised and unfunctionalised CNTs were evaluated.
The CNTs were compounded into the polymer and the product ground, the result being PEEK powder particles, d50 of 1000μm, with the CNTs being bound into the polymer.
Rheology
The rheological properties were determined using a Rosand RH 10 capillary rheometer.
It had been intended to measure the rheological properties of the material containing the functionalised CNTs before and after reprocessing. Unfortunately, it proved impossible to determine the rheological characteristics of the as-supplied material as there was evolution of significant quantities of volatiles as the material was loaded into the Rosand RH10 rheometer. This caused the plug of material to extrude out of the filling aperture, so leading to an extended loading time, this in turn resulting in the degradation of the material.
The material containing unfunctionalised CNTs was tested successfully and the results were compared to unfilled Victrex 150G and 450G samples. The CNT compound was based on a Victrex 150 material. It can be seen that whilst the viscosity of the CNT containing material is higher than that of the base resin it is still significantly lower than that of a 450-base material.
Recycling Procedure
The recyclability of the material was evaluated via injection moulding, the material being passed through a Boy 12A injection moulding machine.
The sample moulded was an ISO 1BA test bar. The limited amount of material available necessitated the use of such a small test bar. In the gauge length of the bars, the dimensions were 2mm thick by 5mm wide.
It was noted that there was a slight chemical smell within the machine hopper during processing, the intensity of the smell reducing as the number of recycling passes increased. The chemical could not be identified from its smell but it was not typical of those found when normally processing PEEK.
Mechanical Properties
The injection moulded 1BA bars were tested using an Instron 5969 with a cross-head speed of 2 mm/min. A 25mm extensometer was used for strain measurement and the tests were carried out at room temperature. All the samples failed without yielding and necking.
The results suggest that multiple passes through the moulding machine tends to improve the properties of the material which could suggest an improvement in CNT dispersion.
Incorporation of the CNTs has resulted in an increase in modulus although the tensile strength and the elongation at break has reduced.
Electrical Properties
The electrical characteristics of the samples were measured using equipment supplied by Electro-tech Systems Inc. A model 844 two-pin probe was attached to a model 880 meter. Surface measurements were taken at three points on each specimen, near the gate, in the narrow waisted central region and at the end furthest away from the gate. In all cases the samples were shown to be totally insulating.
Validation of Recycling Procedure in Commercial Components made from Recycled Compounds
It was originally anticipated that sufficient waste CNT containing product would be available for the manufacture of commercial components, the properties of which could be compared against existing components made from existing materials. The components intended for this work were gears and the testing was to have taken place on two state-of-the-art gear test rigs at Victrex.
Unfortunately, the amount of PEEK-like functionalised CNTs manufactured during the project was low and delivery of the materials was late in the project. The consequence of this was that all the functionalised CNTs available had to be used for the evaluation components being manufactured and tested in the project. Limited amounts of PEEK-like-functionalised-CNT reinforced PEEK were made available for the recycling activities but this amounted to less than 500g of material.
A gear weights around 50g and the minimum quantity of material required to mould a set of gears is around 5-10kg. Thus, there was insufficient material to manufacture the test gears and the work was not carried out.
However, previous work has indicated that recycled PEEK based composite materials do perform well when recycled into injection moulding compounds. Thus, it is certain that the materials produced within M-RECT could be recycled successfully into finished products.
Costs of Recycling
The main cost associated with the conversion of scrap material into a moulding compound will be the resin costs associated with reducing the fibre content of the scrap down to ca 30% by weight in order to produce the moulding compound.
As mentioned earlier 150kg of scrap would produce something of the order of 300kg of moulding compound. To produce 1T of recyclate would require an addition of approximately 500kg of virgin resin which, in the case of PEEK would result in a cost of the order of £50k. The final price would depend on the cost of collection, transport, size reduction and compounding but this is likely to be relatively small when compared to the costs of the virgin material. Costs could be reduced by, given that the composite material is scrap, utilising reworked polymer although this may complicate issues around product quality.
Potential Applications for the Recycled Material
The potential applications for these materials are limited by a number of factors:
In order to develop a market, sufficient scrap material has to be available to ensure a continuous supply in the quantities required by the end user. In the initial stages of manufacture this may well not be the case and the only apparent option, that of supplying premium grade non-recycled materials as an alternative at the same cost to cover the shortfall, would be very expensive and incur significant losses.
The material is recycled and initially the on-going quality of the recyclate may be difficult to guarantee with time and as the operation grows, more data will be available on quality and larger supplies of scrap material will be obtainable and so the quality questions should be overcome.
There will probably be a reluctance, certainly initially to use the materials in situations where the mechanical performance is at a premium.
It may be difficult to penetrate markets such as aerospace with these recycled materials due to the extensive qualification processes that are required.
These problems will probably create issues initially with the reuse of the materials and it is likely that they will be ‘downcycled’, that is used in less demanding applications. There will be properties of the materials that will be unaffected, for example the chemical resistance of the PEEK and PPS. Thus there may be applications where the properties of the materials are actually superior to the incumbent materials and, due to the lower cost, a share shift will be possible from the incumbent to the recycled PEEK and PPS materials.
Suitable applications may therefore be as replacements for acetal and nylon in gears, under-hood clips, brackets and cowlings in the automotive business and in cooking utensils.
Numerical Modelling of the Multi-Scale Materials
The software development was at two levels. The first dealt with the numerical modelling of the damping behaviour of CNT reinforced thermoplastics and the second the numerical modelling and optimization of multiscale reinforced thermoplastics.
Numerical Modelling of the Damping Behaviour of CNT Reinforced Thermoplastics
The stages of this development were:
Derivation of the mechanical properties of SWCNT.
Parametric studies of the influence of tube wall thickness, diameter and chirality.
Modelling the Representative Volume Element (RVE) for CNT reinforced PEEK.
Modelling viscoelasticity from frequency dependent test data derived from Dynamic Mechanical Analysis (DMA).
Implementation of a frictional-type bond slip model simulating the interface characteristic behavior between the matrix and CNT.
Simulation of the damping behavior of an RVE subjected to cyclic load.
A SWCNT was been modeled as a space-frame structure using 3D Bernouli beam elements to simulate the C-C bonds, as shown in Figure 159. The mechanical properties of the beams were derived through the Modified Molecular Structural Mechanics (MMSM) approach using values of Young’s Modulus = 5.59TPa and Shear Modulus = 0.871TPa.
The exact size of the wall thickness of SWCNTs was ambiguous and is under investigation as it seems to affect the mechanical properties of carbon nanotubes, as shown in Figure 160. There is a wide scatter of values for it in the literature, ranging from 0.066 nm to 0.69 nm. In the majority of studies, it has been assumed that the wall thickness of CNTs is equal to the interlayer spacing of graphite which is 0.34 nm.
In order to estimate the effect of nanotube diameter and chirality on the elastic moduli of SWCNTs, a series of simulations were conducted in tension and torsion loading. The mean diameter of the carbon nanotubes ranges from 1 nm to 20 nm and two types of chirality were examined, armchair (n,n) and zigzag (n,0). All models had an approximate aspect ratio about 10 (length/diameter) and the value of the interlayer spacing of graphite for the wall thickness simulations was used.
The simulations revealed that the effect of the diameter on Young’s and shear modulus is very weak, especially for large diameter and the chirality as far as armchair and zigzag is concerned has no significant effect.
The C-C bond in the atomistic scale is represented by a 3D Bernoulli beam which is the structural element for the space-frame CNT. The elastic properties of this beam were derived through an MMSM approach. After a series of simulations, where the space-frame CNT was subjected to tension and torsion loadings, the surrogated 3D beam was derived with elastic properties calculated by FE analysis. Using this constitutional element any geometry of the longer (wavy or straight) micro-scale CNT can be modeled. This CNT is embedded into the matrix, so building the Representative Volume Element. The matrix of the RVE is a rectangular solid which is modeled with 20-nodes quadratic brick elements. The material used is the viscoelastic thermoplastic PolyEtherEtherKetone (PEEK) by VICTREX.
The viscoelasticity of the PEEK was done using a Generalised Maxwell model, as shown in Figure 163, which is a spring connected in parallel with Ν Maxwell elements. A Prony series expansion is used to define the relaxation modulus of the viscoelastic material.
In order to investigate the damping behavior of CNT-reinforced thermoplastics, a series of simulations were conducted in RVEs with CNT volume fractions of 1.5 3.3 and 5.8 % when subjected to a cyclic load at frequency 1 Hz. Two geometrical shapes of CNTs were examined, a straight and a wavy-spiral CNT, the latter being the norm for the CNTs as shown in Figure 166. The geometry of the matrix was a rectangular parallelepiped with a constant length of 1µm for all RVEs, and dimension of 30, 20 and 15 nm.
A spectral technique was developed to take into account the waviness of the CNTs and to determine the mechanical performance of the CNT/matrix system. The model indicates that the effect of waviness is to reduce the damping capacity by around 20%, straight CNTs performing better.
An interfacial bond-slip model was developed where the interfacial load transfer is taken into account through a friction-type bond-slip model at the interface between the polymer and the CNT. A critical interfacial shear strength (ISS) is used to define initialization of slippage of the CNT. A procedure was incorporated into a Finite Element code to compute shear stresses, as shown in Figure 168, from normal stresses acting at the end nodes of each CNT beam element. Assuming constant shear stress along the element, the equilibrium imposes:
The use of large RVEs that contained many CNTs imposes computational requirements that are outside the scope of the current project. Relatively simple models would have something of the order of 100000 degrees of freedom (DoF) and more accurate models could have up to five million DoF. The generation of large RVEs would involve random spatial distribution of the CNTs, random CNT axis orientation and random Gaussian length distributions. The use of large RVEs was being explored within the complementary Master project (Mastering the Computational Challenges in Numerical Modelling and Optimum Design of CNT Reinforced Composites, project number 291239, start date 01/03/2012).
The overall damping capacity of RVEs was found to be due to the viscoelastic nature of the PEEK matrix and slippage of CNT inside the matrix, this occurring when a critical interfacial shear stress is exceeded (τc=40MPa). With an RVE with vf≈1.5% the damping was almost entirely due to viscoelastic effects, the slippage having a minimal contribution. The hysteresis effect in the stress-strain loop is increased with increasing CNT volume fraction due to a greater contribution arising due to slippage. The stiffness of the reinforced PEEK is increased with increased concentration of CNTs in the matrix.
The modelling indicated that the stiffness of RVEs with an embedded wavy-spiral CNT was less than that found with RVEs with a straight CNT.
The process of solution involves:
Compute the incremental displacements
Loop over all beam elements and check each element for slippage. When slippage occurs, the axial stiffness is reduced to zero resulting in a local modified tangent stiffness matrix.
Correction of internal forces of the elements and update global force vector.
The microscopic behaviour of the heterogeneous medium is projected to an homogenous equivalent continuum model at an upper scale. The heterogeneous medium is assumed to be statistically homogeneous so that an RVE can be recognised. Micro-macro relations have to be defined and in this work, macroscopic quantities are formulated as volume averages of the corresponding microscopic state. The average of a quantity at the microstructure level is defined as its integral over the corresponding RVE volume V. The averaging theorem requires the strain energy of the micro-structured RVE to be equal to that of the homogenised macro-continuum.
In defining a representative constitutive model for the aligned and oriented CNTs-RC material, which will be assigned to the homogenised model, the following observations have to be taken into account:
Unidirectionality of CNT-reinforcement induces anisotropic stiffness to the CNT-RC RVE material.
Slippage is observed at the CNT-polymer interface region leading to anisotropic energy dissipation mechanism. Hill’s orthotropic plasticity model combined with Prager’s nonlinear kinematic hardening law is capable to accurately predict damping
Stress-strain behaviour of the CNT-RC is subjected to cycling loads.
The six initial yield stress values can be calculated by imposing the corresponding macro-strain to the RVE model with increasing magnitude up to the point where the ISS is violated. Equivalent yield stress value is calculated from the multiaxial stress state. The nonlinear kinematic hardening data are derived by calculating the equivalent yield stresses as a function of the accumulated plastic strains. Tabular data of equivalent yield stresses and accumulated plastic strains are graphically defined from the nonlinear stress-strain curve derived after tension or shear of the RVE model. Note that the nonlinearity is attributed solely to the CNTs slippage.
The macro-strain is applied in the form of imposed displacements on the boundary nodes of the RVE micro model. The macro-stress is computed from the calculated boundary nodal reaction forces in a volume average manner.
The models were employed in the design of a shear testing jig manufactured during the project, an inlet fan blade and the DIRIS honeycomb.
Numerical Modelling and Optimization of Multiscale Reinforced Thermoplastics
This activity aimed to produce numerical tools for predicting the mechanical behaviour of sheets as well as honeycomb core-panel and complex-shape parts produced with long CF reinforced CNT/PEEK and CNT/PPS matrices.
The numerical model was developed and implemented into an in-house finite element code and integrated with the pre and post processor GiD making a single package called COMPACK . The code was applied to the analysis of the mechanical behaviour of straight beams and DIRIS honeycomb panels made of multi-scale reinforced thermoplastics (PEEK and PPS).
The code developed by was validated against experimental results and then applied for the analysis of complex-shape components proposed by the industrial partners in the project.
The approach to the modelling used an “Advanced Serial-Parallel Model” where each phase (MWCNTs, matrix, long fibres) of the composite material has its own constitutive law (elastic, elasto-plastic, visco-elastic, damage…). The final material behaviour depends on these constitutive laws, volumetric participation and morphological distribution of each component and, eventually, on the inter-phase performance. The model is based on the mixing theory. It can be considered as a constitutive equation manager capable of combining two different performances:
Parallel behaviour Components have the same strain.
Serial behaviour Components have the same stress.
The composite is divided in several layers; each one with a different orientation of reinforcement.
The Parallel/Serial Theory assumes parallel behaviour for the fibre axis and serial behaviour for the other axes.
The new formulation divides the composite into the matrix and a new element, shown in Figure 179, resulting from the coupling of the CNTs with the interface material:
Matrix and the CNTs-interface are related with the parallel mixing theory
CNTs and the interface are bonded together with a combination of parallel and serial mixing theories
The stresses in the interface and the CNTs, at both ends, are assumed to be equal (serial behaviour). The load is transferred from the interface to the nanotube in these regions. The central section has a parallel performance: the strains are equal and the CNTs are capable of developing their full strength.
The model was validated for the thermoplastic materials, PPS and PEEK, used in the project. In the case of PPS, the constitutive model was validated with experimental results taken from the literature.
In the case of PEEK, the constitutive model was validated with experimental results taken from the literature.
The effect of temperature and viscosity on the matrix was included in the code.
The finite element software was written in FORTRAN and compiled into the framework of MS Visual Studio 2010. The Code, made available to M-RECT Consortium, was distributed in the form of executable files for 32 and 64 bit computer operating systems. Graphical Pre and Post Processor and the corresponding interface to the Code have also been offered to the M-RECT Consortium.
The Code, PLCD , is an implicit finite element code, which allows quasi-static and dynamic analysis of solid mechanical problems, bi and tri-dimensional, linear and non-linear, in small or large deformations. It also possible to analyse thermo-mechanical and coupled problems.
Pre-processing of the problem to be analysed is done in GiD by definition of the geometry, meshing, materials and lay-up, assignation to structural parts, boundary conditions, imposed displacements, loading history and control parameters. As soon as the model is complete (GiD), a calculation file is generated for PLCD; input data for PLCD are introduced as a text file. Once the analysis is finished, the results are imported into GiD post-processing. Different options for results output are implemented into PLCD: after iterations, converged load steps, displacements, stresses, internal variables (damage, plasticity), etc.
The library of elements available in PLCD was enlarged to include a three-node triangular shell element with nine degrees of freedom (three per node). This new element joined an optimal membrane element with drilling freedoms and a discrete Kirchhoff (DKT) plate-bending element, counting the membrane-bending coupling effect.
Different solvers (Frontal, LU, GMRES, Cholesky) have been programmed into PLCD. Despite that and in order to accelerate the solution process for large three-dimensional systems, the solver PARDISO was included. The package PARDISO is high-performance, robust, memory efficient and easy to use software for solving large sparse symmetric and asymmetric linear systems of equations on shared-memory and distributed-memory multiprocessors.
Concerning non-linear large deformations, there are different approaches that can be used for analysis. In the Total Lagrangian (TL) approach, equations are formulated with respect to a fixed reference configuration that is usually the initial configuration. In the Updated Lagrangian (UL) approach, the reference configuration is the last converged solution. Both approaches are offered in PLCD when analysis is performed using solid (3D) elements. Details on the different methodologies can be obtained consulting classical references Bathe K-J. (1996), Finite Element Procedures. Prentice Hall and Crisfield M.A. (1991), Non-linear Finite Element Analysis of Solid and Structures, Wiley.
The co-rotational (CR) approach is the most recent formulation developed for geometrically non-linear structural analysis. In this method, the finite element equations are referred to two systems: a fixed configuration and a co-rotated configuration. The main advantage of CR formulation is its effectiveness for problems with small strains (1-4%) and large rotations. The co-rotational (CR) approach has been implemented into PLCD for shell elements.
The Code was applied for the analysis of the mechanical behaviour of straight beams and the DIRIS honeycomb panels made of multi-scale (CNTs and LCF) reinforced thermoplastics (PEEK and PPS).
Complex-shape components proposed for industrial partners have been analysed.
Manufacture and Testing of Components
Five components were manufactured to assess performance of the multi-scale materials and to validate the numerical models. These were:
An engine stiffener
An engine oil pan
An air-supported floating beam
An inlet fan blade
Honeycomb
A satellite structure
Engine Stiffener
The objective of this work was to design and manufacture a medium duty engine stiffener for a 8l EU6 engine using multi-scaled thermoplastic composite material in order to improve the engine’s overall NVH characteristics (by improving the damping characteristics compared to the current aluminium design) while guaranteeing functionality of the part (increasing the engine first bending mode frequency).
Several different stiffener designs were developed and analysed using Finite Element Analysis (FEA) methods, based on a PEEK-based composite material, for:
Static stiffener analysis
Road load analysis
Powertrain modal analysis
The final stiffener concept is an 8,5mm thick compound of different layers of both materials with optimised composite layer orientations.
Fabric semi-preg was used in the core of the stiffener in order to ensure drapeability and the main orientation (0°) is in the principal stiffener direction.
A tool for press-forming the stiffener was developed using ProEngineer CAD software.
The stiffeners were fitted to a test engine located at AVL in Graz and the NVH performance evaluated. The engine was connected to an external dynamometer capable of developing a static or dynamic brake torque on the engine in order to simulate engine loads. It was also capable of driving the engine without combustion (motor mode), this being important to analyse the noise emissions of moving parts decoupled from the combustion process.
The engine tests have been performed in the following order for all three configurations:
- Motored engine run-up (500rpm-2500rpm) without combustion
- Engine run up with 400Nm constant torque (500rpm-2500rpm)
- Full load engine run up (500rpm-2500rpm)
The composite stiffeners resulted in a general increase of sound pressure of around 2dNB over the whole engine speed range. Tests involving the insulation of the composite stiffeners clearly showed that they were the cause of the increase in noise when compared to the performance of the existing aluminium design of stiffener.
To investigate the damping behavior of the stiffeners and to identify the local modes of the stiffeners, hammer tests were performed with the stiffeners installed on the engine and for the stiffeners alone, the stiffeners being attached to elastic bands in order to approach free boundary conditions.
The results showed that in the case of the installed stiffener, the damping of the composite is small in comparison to the aluminium whereas in the free test, the damping is much higher. The reasons for this could be the smaller flexural stiffness of the composite stiffener compared to the aluminum stiffener and/or the decreased screw contact forces due to reduced static friction values at the composite /steel interface. The combination of decreased damping and the resonance of a local stiffener mode with a local powertrain mode has been identified as main reason for the increased noise emission using the composite stiffeners.
Numerical analysis indicated that if the material properties (Young’s modulus, shear modulus) were decreased by 10% and the weight of the stiffener was by 10% the modeled properties are comparable to those measured in the engine test. Thus, it seems evident that small changes in material properties do lead to significant changes in the dynamic behaviour of the stiffeners.
The stiffeners were manufactured by the stamping of a pre-consolidated laminate. This process involves re-melting the polymer matrix and it seems likely that the orientation of the fabric plies was changed during the stamping and the thickness may also be affected in specific regions. This would result in a non-optimise orientation the plies and the possible reduction of the stiffness of the stiffener in specific directions. As noted earlier, there were differences in the results of tests carried out on the stiffeners and those determined by FE analysis.
Whilst the noise levels of the composite stiffeners were higher than those found with the existing aluminium stiffeners the mass of the stiffeners was reduced by more than 50% and the functional requirements were fulfilled. The modelling work suggests that increasing the thickness of the stiffeners to 10mm (ca. 25% increase in weight) will improve the performance so that it exceeds that of the aluminium components.
Engine Oil Pan
The oil pan design developed was that from a Volvo MD8 engine, this being the same engine model to that used for the stiffener evaluation.
A number of alternative materials for the oil pan were modeled and compared with the existing steel oil pan which is 2mm thick. The steel oil pan has a weight of 10.5kg.
The oil level in the pan influences the vibrational performance of the pan. Two oil levels were considered, the levels relate to the maximum when the engine is not running and the normal running level. The oil level had a significant impact on the performance of the oil pan.
The use of a standard Victrex injection moulding compound 90HMF40 was investigated. The moulding of the oil pan was simulated using Moldex 3D and it was found that the pan needed to be 3.5mm thick in order for the cavity to fill. The effect of 24l and 17l of oil in the pan was to reduce the mode one frequency from 390Hz to 137Hz and 147Hz respectively.
The use of stiffeners in the pan structure was investigated and found to reduce the mode one and two frequencies.
Two composite constructions were analysed, these being based on Ten Cate R5690/1 PEEK-matrix semi-preg. The two lay-ups were:
[0/+-45/90]s. A total of eight plies and a thickness of 2.48mm
[0/90/+-45/0/90]s. A total of twelve plies and a thickness of 3.72mm.
The production of a composite oil pan used an existing steel oil pan as a mould. Following manufacture, the oil pan was fitted to a test engine at AVL and its function compared to the existing steel oil pan.
Measurements of the airborne sound from the engine showed that there was an increase in noise with the composite oil pan when compared to the steel version.
Insertion loss analysis of the characteristics of the steel and composite oil pans using forced vibration to excite the pans showed an increase of around 10dB in the airborne noise generated by the composite pan. A speaker was mounted onto the oil pans to generate the noise and the oil pan was attached to a steel plate, the noise being measured one meter away from the oil pan surface.
Comparison of the results from the steel and composite oil pans indicates:
The 1m Airborne sound level increase by about 2 dBA with the composite oil pan at almost all engine speeds and load conditions.
The differences can be observed over a broadband frequency range from 800 Hz to 4kHz, this range being the dominating contributor to the overall noise.
The decoupling of the oil pan showed a positive effect at the engine speed range above 1800rpm.
Insertion loss testing was carried out and the measured noise levels were up to 10 dB higher at some frequencies for the composite oil pan with the same noise speaker excitation.
The difference of the insertion loss was due to the differences of the area weight with a factor of approximately four. (sheet steel 2mm = 15,7 kg/m2 / composite with 8 layers = 3,8 kg/m2)
The structural analysis showed that the material damping loss factor is about twice as large for the composite material but this did not compensate the lower insertion loss of the material.
Air-Supported Floating Beam
The beam developed weighed 35kg and could support a load of 300kg with an air pressure of 100mbar.
The beam was analysed computationally and the results compared to experimental loadings. The two give results which are in close agreement. When inflated with a pressure of 220mbar with a point load applied in the center of the air beam, the vertical deflection in this point is 116 mm, this being in good agreement with the model.
The compression element made of PPS/carbon fibre works very well. However, to increase the size of the air-bridge to 15 m long, a new design of composite construction would be required as this length of span would need a very thick plate. A honeycomb construction would also be a possible candidate for increasing stiffness without significantly increasing weight.
Inlet Fan Blade
The primary objective of this work was to design, build and test a multi-scale reinforced fan blade that reduces blade weight by 15%. The blade should have similar quasi-static and dynamic behaviour to real fan blades.
The two main generic styles of blade are:
Civil airliner : >1m length, 0.4 m width and10 kg mass
Military: » 0.4 m length, 0.3 m width and 4 kg mass
The key dimensions of these blades are the aspect ratio, the camber, the twist and the fan diameter. Maximum blade tip velocities are approximately Mach 1.3 and the blade thickness varies depending on the local loading.
Initial test work was carried out using a PEEK-matrix quasi-isotropic laminate and, as the fan blades can reach 150°C in use, testing was carried out at 20°C and 150°C. Testing was be focused on tensile, shear and CTE performance.
Several designs of tab were investigated for the manufacture of tensile test specimens. However, none of the designs yielded test data where the samples yielded and so the design was altered to one which featured a v-notch, where the notch had a 1mm radius. In the v-notch specimens, the failure occurs where it is expected but it is not possible to measure strain or elastic modulus. It is however, possible to determine the mechanical data using FE analysis to model the tests.
Interlaminar shear tests were carried out by introducing notches in the specimens to make them similar to a lap shear joint. Tabs were used to reduce the loading on end of composite and minimise any crushing effects and a guide fixture was used to prevent any tendency for buckling. As with the tensile tests, the strain cannot be measured directly with this type of specimen and so FE analysis was used to estimate the shear modulus
The CTE was estimated from test data to be 28.8µm m-1 ºC-1.
Work was also carried out to evaluate free vibration, fatigue, creep and high speed impact. The latter evaluation was carried out using a piston driven by compressed air to launch a projectile that contained a force sensor. A high-speed camera was used to record the impact and could be used later to determine the displacement time history and velocity of the projectile. The results indicated that the composite based on PEEK with fCNTs had an increase in flexibility just before rupturing.
Using the data measured, a blade was modeled and designed.
Comparing the natural frequencies of the composite blade to a solid titanium blade, analysis using FE indicates that the natural frequencies of the composite blade are relatively close to those of the equivalent titanium blade. However, the lower damping ratio and lower frequency for mode 3 for the titanium blade would result in higher stresses from the harmonic loading.
The total mass of the existing configuration using solid titanium would be ~1.637kg whereas with the composite configuration the mass is ~0.525kg. It should be noted however, that the titanium blade would normally be hollow so the mass would be less than that calculated.
Three blades were manufactured as listed below:
Plain PEEK film with woven carbon fibre (film stacking technique)
PEEK/fCNTs film with woven carbon fibre(film stacking technique)
Plain PEEK/woven carbon fibre pre-preg
The pre-preg blade was found to be stiffer than the others and the damping levels were slightly lower. This could be because the pre-preg results in fewer voids and gives better wetting of the carbon fibres. The expected mass of the blades was 547 grams based on nominal fibre fraction of 50% and accounting for the PEEK core at the root. This is slightly higher than the actual mass of the blades and suggests that some voids may be present.
The two blades made using film stacking had very similar properties although the one with nanotubes was marginally stiffer and in most cases had slightly higher levels of damping.
From the work carried out it is possible to make the following observations:
The manufacturing process produced blades with reasonably repeatable properties. There was some uncertainty regarding the level of wetting of fibres when following the film stacking process.
The mode shapes obtained were exactly those that were predicted from finite element models while the natural frequencies were somewhat lower than expected.
A model with isotropic properties came surprisingly close to predicting the dynamic properties accurately.
Adding nanotubes to the material appeared to increase the stiffness by a small amount.
Damping levels were higher in the composite blades than would be seen in equivalent metal ones. Some additional damping may be due to incomplete wetting of the fibres.
Properties of the blades were very similar around 150°C to those measured at room temperature.
The blade behaviour was relatively linear with respect to vibration amplitude and exposure to extended high level vibration did not cause damage.
In general therefore, it could be concluded that the blade design and manufacturing process was successful although some improvements could be made to the method used for consolidation.
Honeycomb
The objective of this work was to develop SILICOMB cellular materials to allow for the structures to morph for use in space antennas.
The work necessitated the development of mechanical, thermal and dielectric models for the SILICOMB cellular configuration. The mechanical models were formulated through a rigorous Finite Element homogenisation technique based on the theory of periodic structures. The dielectric and thermal models for the SILICOMB honeycomb were developed using variational principles. Direct application of the use of PEEK Aptiv films in the models made it possible to predict the multiphysics properties of PEEK-based SILICOMB configurations.
PEEK SILICOMB cellular plates have been manufactured and tested for in-plane loading conditions. The plates were manufactured through laser-jet and water-jet cutting of thermo-formed PEEK Aptiv films with two different thicknesses (0.25 mm and 0.6 mm). The SILICOMB PEEK plates were tested under uniaxial tensile loading along the two orthogonal directions, and their elastic and failure modes measured. Good comparison has been observed between the experimental and numerical models developed.
The method of manufacture of the honeycombs, based on Kirigami techniques.
Thermal ageing was been carried out on APTIV PEEK films and they were shown to have a stable
performance.
The compressive behaviour of the Silicomb geometries was found to be nearly twice as good as that of the equivalent commercially available Nomex honeycombs. It was found that the normalised shear modulus was comparable with HRH-10 and Flexcore and that the normalised shear strength was 3-4 times higher than the competing honeycombs.
Panels were made by bonding the honeycomb onto PEEK composite skins and the properties of these panels was evaluated.
Three of the SILICOMB cores were produced with PEEK inserts and the other three with PEEK+CNTs inserts. The inserts are all 10x5 mm and the average thickness of the PEEK and PEEK+CNTs films used for the inserts were 0.22 mm. The nominal size of the SILICOMB cores in plane was 76x76 mmx15 mm thick with the skins being larger than this. The flatwise compression tests were performed according to the ASTM standard [3]. The loading was in displacement controlled mode, at a rate 0.5 mm/min.
The results show that the maximum compressive stress for the sandwiches with plain PEEK inserts is a slightly higher than for the sandwiches with PEEK+CNTs inserts: 1.18±0.015 MPa against 1.11±0.04 MPa. However, the residual strain characteristic the sandwiches with PEEK+CNTs inserts is better with a lower residual strain than for the sandwiches with plain PEEK inserts: 0.031±0.0075 for sandwiches with PEEK+CNTs inserts against 0.042±0.020 for the reference group.
The honeycomb system was further developed with SILICOMB corrugations for multistable applications. Finite Element nonlinear elastic models to simulate the snap-through and bistable behaviour of the plates for different curvature radiuses were developed and plates with SILICOMB corrugation and curvature being manufactured the corrugation and curvature being generated during a thermoforming process. A rig was built to investigate snap through on curved panels and a rig was developed for the manufacture of such curved panels, typical panel dimensions being 200mm x 168mm x 15mm thick with a radius of curvature of 90mm.
Curved structures were tested and the flattened structures showed a recovery to within 0.2% of the original vertical dimensions on release of the compressive loading. The behaviour of the curved panels was modelled and the results were in close agreement with the experimental measurements although the experimental samples showed a higher linear response and a lower stiffness hysteretic return.
Numerical analysis showed that:
The equivalent effective height was inversely proportional to non-dimensional radius and was independent of the rib lengths.
Radial stiffness followed a similar dependency over internal cell angles as per flat SILICOMB configurations.
Radial stiffness was approximately inversely proportional to density, variations over radius and internal cell angles.
Work was also carried out investigating curved corrugated plate structures.
Numerical analysis, performed through parametric analysis of these types of structures had identified the fundamental characteristics in nonlinear deformation of such corrugated PEEK plates.
A prototype curved SILICOMB PEEK honeycomb with an elastomeric zero-ν skin for morphing applications was manufactured. The skins were attached with a 3D knitting technique.
The panel showed large hysteresis during quasi-static cyclic loading and it was found that the energy dissipated per unit volume was strain dependent.
Copper skinned sub-reflectors were constructed for testing and evaluation in the Electronics Department at the University of Bristol.
The snap-through resulting from the use of bistable honeycomb systems in antenna meant that the antenna was easily reconfigured. A number of antenna designs were modeled using the performance of the bistable honeycombs developed and it was found that configurations of antenna with R=2500mm and γ = 13° and 15° had displacements at snap-through which was compatible with Ka bands
The significant results arising from this work are:
Production and testing of cellular structures in PEEK. To the best of knowledge, no other cellular structures in the thermoplastic material have been developed/disclosed in open literature.
Development of multiphysics models for SILICOMB PEEK structures which predict, in a robust manner, the overall performance of the structures
Development of the first case of PEEK-based plates with multistability
Development of a manufacturing technique to produce SILICOMB honeycombs at industrial prototype scale in PEEK.
The overall flatwise properties of the SILICOMB PEEK honeycomb, when normalised against Young’s modulus and compressive buckling stress, are significantly higher compared than other advanced honeycombs for high-end performance.
In the final stages of the project, the first SILICOMB honeycomb was made entirely out of CNT reinforced films although it did not prove possible to evaluate this honeycomb.
Satellite Structure
The support structures that were designed and manufactured were modular structures where a module, a panel-resembling the DIRIS panel, is used repeatedly. For each structure, the size and the distribution of the cells were optimised numerically, attempting to address the requested structural characteristics.
Only structures of one and two modules were experimentally investigated, since it was assumed (and numerically confirmed for compressive loading) that key parameters of their performance are representative for structures with several modules
It was found from the experiments that the reinforcement of PEEK by CNTs is addressing only the requirement of increasing the buckling load by increasing the E-modulus and the bonding strength as a result of improving the thermal conductivity of the material.
Another observation associated with the compressive testing of the module, is that its yield in compression occurs at a much lower stress level than the anticipated strength derived by adding the contribution of all the cells of a module. It should be noted that the bonded preform laterally supports the cells from all sides and therefore the buckling load for each cell should be higher than the one without support from the preform. An explanation of this observation could be related with the cell geometrical imperfections which may cause premature buckling to a few cells, which then leads the surrounding cells to buckle and finally the module to yield.
The structure with two modules was subjected to a dynamic, multiaxial, compressive–tensile fatigue loading up to 4.2 x 106 cycles. The minor changes to the actuator motion recorded during the testing support the argument that the mechanical degradation of the structure was limited. The structure was also subjected to cyclic compressive tests up to the yield, before and after the fatigue testing. The stiffness was not reduced during fatigue testing. Also, the measured stress-strain relations confirmed visco-elastic behavior.
The logarithmic decrement and the hysteresis loop methods were used to measure the damping parameters of the structures. Single-layer modules and two-layer modules were tested in a cantilever beam configuration by applying an impulse force (always in the X direction). The structure response at the free end was measured using two accelerometers.
The loss factor measured using the hysteresis loop method was very similar to that determined via the logarithmic decrement method. There was a small difference between the results and this could be attributed to the differences in the generated stress-strain amplitudes at micro-structural level. Damping increases with increasing frequency due to the viscoelastic behavior of the PEEK/CNTs material.
The measured natural frequency (in the X direction) of the DIRIS panel and the core are high enough and outside the usual range of vibration of transport systems. When the cell size increases and also the cell distribution in the structure is changed such that there are more gaps, the natural frequency decreases. This offers the possibility to tailor the structure design in order its natural frequencies to coincide with the vibration of the system and therefore maximising the energy absorption.
Life-Cycle Assessment of a Multi-Scale Reinforced Engine Stiffener
Life Cycle Assessment (LCA) has been used to compare the current design of engine stiffener used today with an alternative produced in composite materials to see if the environmental impact could be reduced. The function is to stiffen the engine-gearbox assembly, i.e. increasing the first bending modes to a level where prop shaft excitation is avoided. The current version is made from die-cast aluminium and the composite version was developed in polyetheretherketone (PEEK) with woven carbon-fibre fabric with and without CNTs.
In this study, the LCA results were characterized by Global Warming Potential (GWP), Acidification Potential (AP), Ozone Depletion Potential (ODP) as well as environmental indices expressed in the
EPS 2000 method (Environmental Priority Strategies).
The LCA method adopted was based on ISO 14040:2006 and based on a medium duty truck with an 8 litre engine which is expected to cover 750000km in a lifetime of 15 years.
The production phase consists of raw material production and parts manufacturing. For the raw material production, the inventory data (in- and outflows of various resources, emissions and energy flows) as well as the transportation of the raw materials were included.
For the manufacturing phase, the available dataset from the GaBi software was used.
The use phase included production of diesel and urea which are consumed during the truck’s lifetime, the exhaust emissions and maintenance. Maintenance covers the materials consumed such as transmission oil, engine oil, coolant etc. and replacements of spare parts over the truck’s lifetime. However, no maintenance is expected for the engine stiffeners themselves. Hence, the maintenance phase was excluded from the study.
Generally, as regards the EoL of the truck, 100% of the steel, aluminum, copper and 97% of the scarce metals (e.g. platinum and palladium) are considered to be sent for recycling; 80% of gold, silver and stainless steel are considered to be sent for recycling due to lack of data for recycling process. Others are set as landfill. For the engine stiffeners only the recycling of the parts themselves will be considered.
The following activities were excluded (cut-off criteria) in this study due to the activities being assumed to be insignificant or no data was available.
Surface treatment of the engine stiffeners
Production, use and EoL of packaging materials;
Manufacturing data of the metals and polymeric materials, which are not included in GaBi and where no other data was available.
No maintenance of the engine stiffeners are expected.
Waste water treatment and waste treatment are not included in the materials’ inventory data.
Transports of raw materials and between suppliers and sub-suppliers as the part is still in a pre-industrialisation phase and such parameters are unknown at this stage.
The results show that the weight of the engine stiffeners has a significant effect on the use phase because of the reduction in fuel consumption. This means that the lower weight of the composite engine stiffeners significantly reduce the overall environmental impact. The use of carbon fibre in the composite has a significant impact comparable to the use phase for most environmental impact categories as they significantly reduce the weight of the composites in comparison to systems based on such materials as glass fibres.
Thus the engine stiffeners with a reduced weight would be advantageous as long as the other property requirements of the part(s) are fulfilled, for example those concerning mechanical, physical, chemical, noise, vibrations, fire resistance, crash resistance performance.
The following recommendations may be applicable for vehicle applications where the fuel consumption is very important
Reduce the weight of the parts as much as possible without compromising the requirements with regard to other properties (mechanical, physical, chemical, noise, vibrations, etc)
Is it possible to find alternative raw material sources if the impact of the current raw material is high? For example, the production of carbon fibre.
Is it possible to redesign part(s) so that more parts can be integrated in a larger part that may be considered for recycling?
Risk Analysis
Risk assessment is usually based on a structure where the ranking of the level of overall risk being based on the product of the likelihood and the impact of an event.
The different risks which occur in this type of project include:
Technological risks
Partnership risks
Market risks
Legal risks
Management risks
Environmental/regulation/safety risks
The risks were assessed by the participants and the major risks were determined and are presented in the Table below.
Major Risks Solutions and Actions Comments
Significant dependency on other technologies. The materials developed within the project depend on the successful functionalisation and incorporation of the CNT into the laminates.
To improve the incorporation of the CNTs within the matrix, laminates were manufactured with a novel impregnation technology (patent application was proposed). Problems were faced in the manufacture of the PPS laminates when CNT were incorporated due to the formation of cracks.
The functionalisation of CNTs do not reach the expected target Technologies for the successful functionalisation of CNTs were developed. Patent application for a route to the chemical functionalisation of the CNTs
Industrialisation at risk: an industrial partner leaves the market This risk was considered to be a high risk during the project. But the correct decisions were made by the participants which minimised the risk. The project started using Baytubes supplied by Bayer as these were in a “granular” form and hence they were advantageous from the health and safety perspective. At 6M of the project one partner started to use the Nanocyl CNTs. After agreement of the consortium the Nanocyl materials were chosen even if more precautions were required for the handling. 6M before finishing the project Bayer announced that they are no longer to produce Baytubes, so the coordination actions to switch to Nanocyl material was, retrospectively, be considered to have been a good decision.
Cultural differences and mode of operating of partners cause delays Coordinating actions were successfully developed to improve the communication between the partners.
Worthless result: performance lower than market needs The benchmark performance of current materials in the market is known.
This risks show that the tensile properties of the CNT containing laminates exhibit inferior properties to those without CNTs; but the fracture toughness of the PEEK/CNT laminates show promising results. For the new technologies to be acceptable in the market, their cost-benefit ratio must be good enough but the market must feel the need for them or must be able to see new regulations being prepared
Nobody buys the product. Too expensive. Development of a Life Cycle Cost Benefit Analysis to determine the benefits of the implementation at industrial level. It was shown by the life cycle assessment carried out that the engine stiffeners made out PEEK 37% Carbon fibre and 0.5% CNT had 56% less weight than the engine stiffeners made out of die-cast aluminium. The reduction of weight by using composite materials will imply a decrease of the fuel consumption during the life time of the vehicle showing a positive environmental and social impact.
Inadequate or ineffectual communication among partners. Coordinating actions were successfully developed to improve the communication between the partners
Lack of awareness in the use of personal protection when handling nanomaterial The handling of nanomaterial in a safety lab should follow the recommended practices for laboratories.
Lack of awareness in regards to the stability of the nanomaterial contained in a solid matrix, during use and mobility. Engineering controls should be applied when handling the final product in the laboratories and further studies should be carried out for analysing the lifecycle of the nanomaterial in the solid matrix. However it is known that once the nanomaterial are embed in the solid matrix the health risks are low.
The environmental hazards of the final product are unknown Preparation of the SDSs according to the format provided by REACH and CLP regulations. The preparation of SDSs will imply the full characterisation of the product as well as the identification of the health and physical hazards.
Main Identified Risks
The information developed during the project included information concerning:
Practical approach for managing safety of nanomaterials in a lab
Material Safety Data Sheets (MSDS)
Possible Hazards for the life cycle of Nano-Materials
Scientific research about Nanomaterials
Guidelines for Carbon Nanotubes, Nanofibers
Recommendations
Exposure Assessment
Recommended Exposure Limit (REL) for CNT
Engineering Controls
Respiratory protection for CNT
The different risks posed in M-RECT project were evaluated by the consortium and the partners evaluated 11 out of 52 (21%) risks as high or very high. The high risks faced during the project were successfully managed by the coordination activities, different solutions and actions were identified.
One of the highest concerns raised by the partners was the lack of awareness in regards to the safety aspects, for example in the health and physical hazards of CNT, in the use of personal protection devices and safety controls and in the hazards of nanomaterials embed in a solid matrix. As a consequence, guidelines were developed for managing safety with nanomaterials, especially for CNT
It is important to note that several studies have investigated the toxicity of carbon nanotubes (CNTs) in experimental animal studies. The results from these studies indicate potential respiratory health risks from exposure to CNTs, including granulomatous pneumonia and fibrosis. Evidence also indicates that when multi-walled carbon nanotubes (MWCNTs) are administered intraperitoneally to mice, the MWCNTs have asbestos-like pathogenicity. Although a causal link has not been established, there is concern about possible cancer hazards in addition to potential for fibrosis/non-malignant respiratory disease.
Assessment of Uncertainties
All measurements are affected by errors coming from different sources e.g. the measurand, the measurement instrument, the measurement procedure, the environmental conditions, etc. Therefore, it can be assumed that the result of any measurement is affected by an uncertainty, which is, an interval of values inside of which will be a true value of a quantity.
The work involved the calculation of uncertainties in material characterization and a sensitivity analysis dealing with the possible aberrations from nominal product characteristics and included a web based resource dealing with the test methods for the determination of such things as the tensile properties of plastics.
Data from partners was collected and initial uncertainty calculations developed by following the steps:
Identify the measurand, and the quantities to be reporting
Identify the influence factors: sources of uncertainty i.e. test piece, testing system, environment, test and measurement procedure.
Classify the uncertainty types as below:
Type A:
Evaluation is by calculation from a series of repeated observations, using statistical methods.
Type B:
Evaluation is by means other than those used for Type A evaluation. For example, by using data from: calibration certificates, manufacturer’s specifications, previous measurement data, experience with the behaviour of the instruments, and all other relevant information.
Estimate the standard uncertainty and sensitivity coefficient for each source of uncertainty
Estimate the standard uncertainty and sensitivity coefficient for each source of uncertainty
Calculate the combined uncertainty
Calculate the expanded uncertainty, U, by multiplying the combined uncertainty uc calculated in Step 5, by a coverage factor, k. For a normal probability distribution k=2; this corresponds to a confidence interval of 95%. The expanded uncertainty U is defined in the GUM as “the interval about the result of a measurement that may be expected to encompass a large fraction of the distribution of values that could reasonably be attributed to the measurand”.
Report the value
Data from a shear test (+/- 45° tensile test) on composite laminates underwent preliminary analysis, and the uncertainties calculated provided an overview of the calculation for 6 samples of the Type A uncertainty. The calculation of the uncertainty Type A is done by the following steps:
Determination of the mean value and the standard deviation of the measurand
Determination of the confidence region of the mean value by using t factor of student’s distribution at 95% of confidence and n-1 degrees of freedom
For the combined uncertainty of the maximum stress (σ= F_max/(2*w*t) ) a sensitivity analysis of the parameters was performed by the use of tornado diagrams. The calculations indicate that for the Type B uncertainties of the σmax, the parameter which has the highest influence is the maximum force (Fmax) in comparison to the width (wo) and the thickness (t0). This is also confirmed by consideration of the influence of each variable in percentiles to the σmax, each percentile indicating the value below which a given percentage of observations fall. For example, the random data from Fmax disturbs the final uncertainty for σmax, in a way that, 95% of the uncertainty falls below 2.16 MPa and 5% of the uncertainty falls below 2.12 MPa.
Estimation of uncertainty allows the comparison of equivalent results with other laboratories, and with values given in specifications or standards. A standard methodology for determination of uncertainties in materials characterisation was applied within the project.
It is important to note that when developing new materials if the mechanical properties are worse than the standard materials and if the certainty of the results is lower, a low benefit is expected. In the case, when properties are better compared to the standard materials, a high benefit is expected only when the certainty of the results is also high.
Surface Coatings
Most polymers are susceptible to degradation as a result of exposure to UV radiation. The chemical structures of PEEK and PPS result in significant absorption within the UV spectrum and so measures are needed to prevent/inhibit/reduce any degradation mechanisms that may result from this absorption.
Whilst additives can be used in the polymers, this involves compounding with the subsequent problems related to powder manufacture.
Thus, the chosen route for UV protection of structures was by applying coatings, the coatings needing to have high levels of adhesion to PEEK and PPS. This was difficult due to the nature of the polymers, the difficulties being compounded by highly fluorinated systems having very limited adhesion capability. The research carried out was directed at introducing a suitable primer layer or using a suitable polymer matrix to support the aluminum platelets used to generate the desired coating properties. In the case of the aluminum platelets they are automatically orientated parallel to the surface during solvent evaporation and this builds up a uniform reflective mirror.
It was envisaged that the total coating would be thin (< 50 μm) and the bonding to the substrate would be characterized as chemical, rather than physical, bonding.
The test specimens were cut out of selected standard PEEK films and composite laminate supplied by partners in the consortium.
Various developments and tests were carried out to assess the coatings developed for the protection of the PPS laminates and these included:
Optimized polymer / hardener ratio
Optimized additive content – new formulation
Qualified layer thickness
Qualified drying / storing and curing data
Qualified vibration test results
Qualified industrial application procedure (spraying)
Bonding qualified (grid cut and pull off)
Chemical resistance qualified
Protective properties of coating qualified
The optimised formulations described as numbers RK 022 and RK 023 passed all tests, especially chemical resistance and bonding. Most important in this application is resistance to gasoline and benzene, as these were considered to be challenging reference chemicals for motor/truck applications.
Various developments and tests were carried out to assess the coatings developed for the protection of the PEEK films and laminates and these included:
Optimized polymer / hardener ratio
Optimized additive content aluminum pigment – new formulation
Qualified layer thickness
Qualified drying / storing and curing data
Qualified bending test results
Qualified industrial application procedure (spraying)
Bonding qualified (grid cut and pull off)
Chemical resistance qualified
Protective properties of coating qualified after special 3000 hrs UV impact and 100 KGray gamma radiation impact
0% transmission of light UV and 70 % of Reflection after application of intended load including UVC load
The complete PEEK –coating testing “Phase II” resulted with formulation V 12 showing the best technological and application performance. The aluminium-pigmentation is a mixture of a so called conventional “non-leafing Alupigment” and a high reflective leafing “bronze” pigment, which achieved a reflection of ~70 %. The transmission measured between 250 nm and 700 nm was 0%. Formulation V 12 / V 12.2 showed the best behavior in bonding, flexibility, UV resistance and gamma resistance.
Durable UV and heat resistant coatings were successfully produced and evaluated for both types of composite matrix.
Potential Impact:
The main outcomes to the project are related to potential economic and environmental impacts.
The life-cycle-analysis carried out in the project clearly indicates that the major factor in respect of environmental impact is the ‘use’ phase. Reducing the weight of structures through the use of composite materials generates significant reduction in the impact of the ‘use’ phase on CO2 production and resource usage.
Fuel economy is directly related to the weight reduction and the reduction in fuel usage will have both environmental and financial impacts. Current designs of composite bodied aircraft are prone to increased levels of NVH due to their construction. The use of additional materials for damping purposes within these applications leads to increase in weight and so the development of smart multi-functional materials is important.
Information from the aircraft industry on the effect weight savings suggests the following average yearly fuel burn per kg weight carried in the table below. These savings have a considerable effect on the running costs of an aircraft and the environmental impact that the aircraft makes.
Aircraft Type Annual Fuel Savings per kg
Airbus A310 200kg
Airbus A330-200/300 145kg
Boeing 777 160kg
CNT Functionalisation
The development of the functionalisation processes for the CNTs has resulted in:
• Large increases in the efficiency of functionalisation both in respect of the times required and the quantities of chemicals used and their recovery.
• The development of novel reactor and mixing equipment that is likely to also have potential applications in pharmaceutical manufacture.
• Developed technologies where CNTs, which are initially loose and fluffy and so present a health hazard, are converted into functionalised CNTs which aggregate, so reducing the potential health hazards. This aggregation does not hinder the dispersability of the CNTs in the polymer matrix as would normally be expected.
• The chemical route developed for the PEEK-like functionalisation of CNTs is the subject of a patent application WO 2013/190162 filed on the 20th June 2013.
Once combined with the matrix material, the CNTs are fully encapsulated and so pose no safety hazards.
The outcome of the project in relation to the use of functionalised CNTs was not as expected, the materials containing functionalised CNTs tending to exhibit poorer performance when compared to those containing unfunctionalised and no CNTs. However, the results do indicate a clear impact of functionalisation and has indicated the route forward in order to improve their performance.
Pre-Preg Manufacture
The novel impregnation technology developed has significant impacts on the economics and safety of the impregnation process. The benefits include:
• Higher manufacturing outputs for UD tapes. Typical line speeds with dispersion based processes are <10m/min. The new impregnation technology has successfully run at 50m/min, the line speed being limited by ancillary equipment, not the fundamental nature of the process. This increase in speed was not to the detriment of the quality of the tapes produced.
• Fabric pre-pregs are usually produced by powder scattering technologies and so the level of impregnation of the matrix into the fabric is relatively poor. This necessitates the use of high pressure processes to convert the pre-preg into finished products. The development work in the project has generated fabric pre-pregs which are very much better impregnated than the normal materials produced by powder scattering. Work is still required to produce a fabric which is fully impregnated but this technology would offer:
o The ability to use low-pressure processes such as oven consolidation for the manufacture of parts.
o This in turn would reduce any capital expenditure required to establish production lines for the conversion of these materials.
o The outcome would be faster production at lower costs, both revenue and capital.
A patent application, FR1361523, was filed on the 22nd Nov 2013 for the impregnation process and the beneficiary is currently having discussions with seven companies for validation trials (three are French and four are European/worldwide) concerning the exploitation of the technologies developed.
Component Manufacture
Novel methods of component manufacture were investigated during the project which could lead to:
• The manufacture of hollow components without the need for extractable bladders using oven heating. This would result in:
o The reduction in capital expenditure required to initiate such processes
o Faster uptake of such technologies and the materials being used in the project
o Stronger, stiffer component structures that will result in the ability to use lighter structures so improving efficiency of aircraft and automobiles.
• The possibility of bonding elements together during the primary manufacturing process, for example bonding honeycombs to skin panels. This will reduce the production steps required and so speed up the manufacturing process and reduce the cost of manufacture.
The technologies developed will result in the faster manufacturing of composite panels this resulting in higher output rates from lower cost equipment so reducing the overall cost of manufacture.
Recycling
Minimising waste and maximising the use of resources is important both economically and environmentally. The materials can be re-melted and reprocessed as many times as it is required without any deterioration of their physical and chemical properties. The project has clearly indicated the recyclability of the materials developed, identified potential markets for the recycled material and the methodology for waste collection. In most cases, it appears that the waste material could be ‘up-cycled’ into high-end applications yielding significant economic, as well as environmental, advantages.
Recycling was not considered during the life-cycle-analysis and so this would further reduce the environmental impact of the materials.
The recycling process did not involve the degradation of one or more components of the waste material and hence is far more environmentally friendly than the processes used for thermoset composites. In the case of the latter materials, the waste is chopped up and pyrolysed to degrade and drive off the resin component so as to leave the carbon fibre. This process involves significant energy usage in order to elevate the temperature of the waste to the point where the matrix degrades and also results in the evolution of significant quantities of waste gaseous by-products from the reaction which may be toxic. The proposed recycling methodology did not result in any similar issues.
A similar situation arises at the end of the service life of the components in that they can be readily broken up and recycled, possibly ‘up-cycled, and so they should not be treated as waste but as a valuable resource.
As with the recycling situation, the CNTs contained within the multi-scale composites pose no health hazard as they are bound into the matrix material. The matrices used are ductile in behaviour so the likelihood of the release of CNTs on failure of the components is highly unlikely.
Development of Numerical Models
The numerical models developed to predict the behaviour of the multi-scale composite materials being developed were shown to be very accurate.
The ability to model the materials and structures accurately will result in:
• Shorter design cycles so allowing industry to respond more quickly to any required design changes
• Reduce the development time for new components and also, by coupling systems, finished structures such as automobiles and aircraft.
• Reduced cost of development and manufacture by being able to design on a ‘right-first-time’ basis. Many problems are designed into structures and good design tools are essential in avoiding these sorts of problems.
The model development work has resulted in a further EU project, Mastering the Computational Challenges in Numerical Modelling and Optimum Design of CNT Reinforced Composites, project number 291239, which started on 01/03/2012.
Software for the determination of the performance of multi-scale reinforced composite materials is now commercially available.
Validation Components
Engine Stiffener and Oil Pan
It has been observed that most of the engine noise in a modern automobile comes from internal engine covers such as oil pans and valve covers. These covers are excited in the acoustic relevant frequency range between 500 Hz and 5 kHz. Passive and active damping control solutions have been already tried on these components but the results have been poor so far since their cost-effectiveness, despite their technical merit, is usually low.
The engine stiffener and oil pan manufactured and tested during the project did not perform as well as expected in terms of noise, vibration and harshness (NVH). Whilst the components developed during the project did not lead to the anticipated reductions in NVH, the results obtained provide a good engineering basis and understanding for the further development of components which will result in such reductions. This should result in a direct social impact resulting from noise control and pollution, particularly in automotive applications, especially in urban areas and highways.
It should be noted though that the reduction in weight of these components as compared to the aluminium and steel counterparts is clearly advantageous in terms of the environmental impact of the solutions.
Air Supported Floating Beams
Market demand for new light-weight and highly resistant bridge structures based on flexible light-weight materials is constantly increasing and related manufacturing processes and technologies are rapidly changing. In 2002, the South African Parliament dealt with a budget of over 3 million € for the acquisition of a temporary stock of emergency bridges of various sizes, in order to cover 14 of more than 262 path destroyed by floods.
The new adaptive and functional inflatable bridge structures developed in the project are applicable to a wide array of sectors in civil and construction industries, ranging from fast deployable road bridges for emergency situations, to temporary bridges for auxiliary tasks in the construction of highways and railways.
Manufacturing of the air bridge structures considered in the project is possible but an integrated and automated process chain has not yet been developed. Work in the project has resulted in the definition of the production equipment and the related manufacturing system for the massive industrial production of such highly resistant and adaptive air-beam/bridge structures. A fully automated and adaptable manufacturing system is currently being developed which will address the vertical integration of the process production chain including the definition of cutting pattern, assembly of patterns by sewing or gluing, transport, deployment, layout, dismantling and storage, among other specifications.
The air bridge will be applicable for facilitating emergency evacuation tasks, communication, provision of resources and rebuilding of devastated areas due to natural disasters. Indeed, many other applications of the air-bridges developed in this project exist in surface transport engineering and in the building construction sector, among others.
In summary, the growing need for emergency civilian bridging systems clearly establishes a need for a new generation of unique light weight and easily deployable bridge systems as the air-sandwich bridge.
Fan Blade Aero-Engine
It has been estimated that a reduction in the mass of a fan blade by 15%, a project objective, would be translated to a reduction of the weight of a large twin engine aircraft by 350kg. General Electric claimed a weight reduction of nearly 350kg by switching to a composite fan system for their new GEnx engine for the Boeing 787. These weight savings are multiplied when one considers the reduced mechanical performance requirements in terms of bearings and the fan containment casing due to the lower mass of the fan blade.
The main factors to consider in the design of a inlet fan blade are the strength, the weight and the damping behaviour.
Analysis using FE indicates that the natural frequencies of the composite blade are relatively close to those of the equivalent titanium blade. However, the lower damping ratio and lower frequency for mode 3 for the titanium blade would result in higher stresses from the harmonic loading and so the composite blade would have an improved performance in this respect..
The total mass of the blade configuration developed during the project when using solid titanium would be approximately 1.64kg whereas with the mass of the composite blade is approximately 0.53kg this being a reduction of approximately 70%. It should be noted however, that the titanium blade would normally be hollow so the mass would be less than that calculated but the projected weight saving of 15% seems very feasible.
From the work carried out it is possible to make the following observations:
• The mode shapes obtained were exactly those that were predicted from finite element models while the natural frequencies were somewhat lower than expected.
• A model with isotropic properties came surprisingly close to predicting the dynamic properties accurately, this taking some complexity out of the analysis.
• Adding nanotubes to the material appeared to increase the stiffness by a small amount.
• Damping levels were higher in the composite blades than would be seen in equivalent metal ones and so the performance in respect of vibration should be improved over the metal equivalents.
• Properties of the blades were very similar around 150°C to those measured at room temperature and so the blades should perform in a satisfactory manner in an engine under load.
• The blade behaviour was relatively linear with respect to vibration amplitude and exposure to extended high level vibration did not cause damage.
The blade design developed shows great promise in respect of reducing the mass of the engines on an aircraft and so improving fuel performance and reducing environmental impact from emissions and also giving better damping and so improved performance of the blade.
Morphing Honeycombs
The use of honeycomb structures offers advantages in respect of the noise and thermal insulation as well as the ability to manufacture light stiff structural components.
SILICOMB honeycombs, in the form of honeycombs and laminate structures, have been made and evaluated. The properties of the honeycombs were in many cases superior to those of the current commercial materials both mechanically and in terms of thermal performance.
The morphing structures developed proved to be very suitable for deployment as transformable space antenna, numerical modelling approaches being developed for the assessment and design of such structures.
The work has resulted in a further project being started, FP7-AAT-2012-RTD-LOMORPHELLE, which should result in exploitation of the technologies developed.
Space Antenna Assembly
The support structures designed and manufactured using PEEK film reinforced with CNTs, were modular structures where a module, a panel-resembling the DIRIS panel, is used repeatedly. For each structure, the size and the distribution of the cells were optimised numerically, attempting to address the requested structural characteristics. Only structures of one and two modules were experimentally investigated, since it was assumed, and numerically confirmed for compressive loading, that key parameters of their performance are representative for structures with several modules.
The performance of the modules was assessed using dynamic, multiaxial, compressive–tensile fatigue loading, cyclic compressive tests up to the yield and the logarithmic decrement and the hysteresis loop methods were used to determine the loss factor.
The measured natural frequency (in the X direction) of the DIRIS panel and the core were high enough and outside the usual range of vibration of transport systems. Changes in the cell size and the cell distribution in the structure can be used to tune the natural frequency system. This offers the possibility of tailoring the design of the structure such that its natural frequencies coincide with the vibration of the system, therefore maximising the energy absorption and so the design offers great potential for satellite applications and similar applications where stable platforms are required.
Surface Coatings
Most polymers are susceptible to degradation as a result of exposure to UV radiation. The chemical structures of PEEK and PPS result in significant absorption within the UV spectrum and so measures are needed to prevent/inhibit/reduce any degradation mechanisms that may result from this absorption. Whilst additives can be used in the polymers, this involves compounding with the subsequent problems related to powder manufacture and so the chosen route for UV protection of structures was by applying coatings.
The coating development work and associated testing for the PPS laminates resulted in optimised formulations described as numbers RK 022 and RK 023 which passed all the tests including vibration tests, bonding tests, and chemical resistance and could be applied by spraying, a qualified industrial application procedure. Chemical resistance included resistance to gasoline and benzene, as these were considered to be challenging reference chemicals for motor/truck applications.
Similar activities resulted in Formulation V12 for the PEEK laminates, the adhesion of the coatings to PEEK being more difficult than to PPS. Testing for the PEEK laminates also included subjecting the coated laminates to 3000 hours of UV radiation and 100 KGray of gamma radiation impact, the coatings performing well.
The coatings contained aluminium pigmentation, this being a mixture of a so called conventional “non-leafing Alupigment” and a high reflective leafing “bronze” pigment, which achieved a surface reflection of approximately 70 %. The transmission measured between 250 nm and 700 nm was 0%. Durable UV and heat resistant coatings were successfully produced for both types of composite matrix and are now commercially available.
Structural Health Monitoring
One major problem with composite structures is damage detection. The ideal approach is to be able to mount a sensor system into a composite structure such that the behaviour of the structure can be measured and analysed in real time.
A structural health monitoring system based on the swept-wavelength coherent interferometry technique has been successfully demonstrated in this project. Initial work involved the successful surface mounting the fibre on the composite materials and also the development of techniques for the integration of distributed fibre optic sensors (DFOS). Later work involved moulding the fibre into PEEK composite which showed that the DFOS was capable of withstanding temperatures of 385 °C and pressures of 10 bar.
The DFOS and the measurement technique have been shown to provide excellent results during fatigue tests containing 10 million load cycles and also the ability to detect crack formation as well as determining local strains along the axis of the fibre.
As a result, it has been demonstrated that DFOS are reliable structural health monitoring sensors during their estimated service lifetime for the application on the surface of composite structures and the integration into composite structures under process conditions.
Thus, sensors can be embedded into composite structures and used to measure the performance of the structure in real time over the full lifetime of the composite structures. This technology can be used to ensure the safety of composite structures, for example in aircraft, and provide early warning of potential problems associated with the structures whilst in use. The results can also be used as the basis for the future development of guidelines or standards which, in safety conscious industries such as aerospace, are very important.
In conclusion, the developments within the project should stimulate the introduction of such composite materials into applications within the aerospace and automotive markets with the components produced forming good case studies. There will be economic benefits of adopting the new materials in manufacturing, usage and end-of-life as well as significant environmental benefits particularly in usage. Whilst the NVH benefits were not fully met during the project, clearly defined routes to achieving this have been developed.
List of Websites:
http://mrect.risk-technologies.com/