Production Engineering https://doi.org/10.1007/s11740-018-0829-9

PRODUCTION PROCESS

Numerical and experimental investigation of manufacturing and performaance of metal inserts embedded in CFRP Markus Muth1   · Jan Schwennen2 · Alexander Bernath3 · Julian Seuffert3 · Kay André Weidenmann1 · Jürgen Fleischer2 · Frank Henning3,4 Received: 16 November 2017 / Accepted: 6 April 2018 © German Academic Society for Production Engineering (WGP) 2018

Abstract Due to their outstanding specific mechanical properties, carbon fibre reinforced plastics (CFRP) exhibit a high application potential for lightweight structures. With respect to multi-material design and to avoid drilling of structural CFRP parts to join them to other components, embedded metal elements, so called inserts, can be used. The inserts consist of a shaft and a baseplate which is embedded between the fibre layers. So far, only punctiform inserts have been subject to research. One feasible geometry are linear inserts which have not been studied yet. In this work, the performance of two different types of linear inserts will be investigated. The shapes are based on a punctiform insert which is made out of a threaded shaft welded onto a baseplate whose performance under different types of loading has been investigated before. The first type of linear inserts has the same cross-section as the reference punctiform insert but is of a linear form. The second type is a quasi-linear insert which consists of a baseplate with the same dimensions as the first linear inserts and three threaded shafts welded onto it. All samples are manufactured by resin transfer moulding (RTM). Depending on the geometry of the insert and the preforming concept it is potentially possible to maintain the fibre continuity. For the inserts with a continuous shaft and in the proximity of the insert, it is necessary to cut fibres of the top layers which are aligned perpendicular to the shaft. For the quasi-linear insert, it is possible to maintain the fibre continuity as the fibres are guided around the circular shafts. Additional to mechanical tests that are carried out, mould-filling and curing simulations are performed for different inserts to analyse the influence of the process parameters onto the part quality. In the main series of tests, the specimens are characterized regarding their failure behaviour and load bearing capacity under quasi-static loads. The results of the experiments show that, compared to the punctiform reference insert, the linear load introduction elements exhibit higher load bearing capacity. However, the linear load introduction elements are inferior regarding specific load bearing capacity and furthermore increase process complexity during preforming and production. Keywords  Hybrid · Composite · Joining · Linear inserts · Tensile tests

* Markus Muth [email protected] Jan Schwennen [email protected] Alexander Bernath [email protected] Julian Seuffert [email protected] Kay André Weidenmann [email protected] Jürgen Fleischer [email protected]

1



Karlsruhe Institute of Technology (KIT), Institute for Applied Materials IAM–WK, Kaiserstr. 12, 76131 Karlsruhe, Germany

2



Karlsruhe Institute of Technology (KIT), wbk Institute of Production Science, Kaiserstr. 12, 76131 Karlsruhe, Germany

3



Karlsruhe Institute of Technology (KIT), Institute of Vehicle System Technology, Chair for Lightweight Technology, Rintheimer Querallee 2, 76131 Karlsruhe, Germany

4



Fraunhofer Institute for Chemical Technology (ICT), Joseph‑von‑Fraunhofer Str. 7, 76327 Pfinztal, Germany

Frank Henning [email protected]

13

Vol.:(0123456789)



1 Introduction When low component weight is required but high structural loads have to be carried, continuous CFRP components are increasingly used. In order to offer an inexpensive and detachable load-bearing structure, mechanical fasteners are selected with regard to multi-material design [1]. If bolts or rivets are used as a connecting element, the CFRP component has to be drilled beforehand. Due to the drilling, the load bearing capacity of the component is significantly reduced by the applied stresses and the local load caused by the connection [2–4]. To avoid this disadvantage, metallic inserts can be used as load introduction elements. Depending on the type of insert, cutting of the fibres in the preform process can be avoided, which is achieved by preforming the fibres around the shaft, resulting in an optimum load path [5]. In previous investigations, as in this contribution, only the maximum forces causing component failure were evaluated. Ferret et al. [6] have tested so-called bighead® inserts under tensile (pull-out), bending and compressive (push-in) loads. Inserts with a large baseplate fractured around the inserts stud, which in comparison to inserts with a smaller baseplate led to a reduction of load-bearing capacity. Hopmann et al. [7] have carried out pull-out tests on inserts and bonded fasteners (so-called onserts). It was shown that the performance of inserts is up to 37% higher compared to onserts. The influence on the performance of various surface treatments [8], geometrical and manufacturing parameters [9] is described by Fleischer et al. [8] and Schwarz et al. [9]. Gebhardt et al. [10] have applied quasi-static tensile, bending, torsion, shear and compressive loads to embedded inserts and were able to identify the tensile load as the critical load case. Debonding of the inserts from the lower laminate layers is identified as the damage mechanism which leads to the high decrease in load bearing capacity during pull-out test. In this context, Soliman et al. [11] and Ferguson et al. [12] examined the most important failure mechanisms under loading in through-thickness direction. They showed that delamination, debonding and fibre/matrix cracks are the main causes for failure. The investigations of Gebhardt et al. [10] have shown, that the maximum pull-pout force of a punctual load introduction can’t be increased with a bigger diameter of the insert baseplate significantly. Depending on the application, sometimes higher pull-out forces are required. One application example is the belt attachment to the car body. To increase the pull-out forces further a new load introduction concept will be investigated in this paper. The punctual load introduction gets stretched to a linear load introduction. It is expected that inserts with a linear load introduction will show a higher load bearing capacity. Therefore, the advantages, disadvantages and failure mechanisms regarding their

13

Production Engineering

quasi-static load bearing capacity have to be compared with the ones of punctiform inserts. In this contribution, the performance of a linear insert and a quasi-linear insert for quasi-static tensile loads will be investigated. The shape of the linear insert is based on a punctiform insert made of a threaded shaft welded onto a baseplate. Their performance in quasi-static tensile loads investigated in [5, 8, 10, 13–15] will be used as a reference in this contribution. The shape of the quasi-linear inserts baseplate is the same as for the linear inserts used. In contrast to the linear inserts, three threaded shafts, used for the punctiform reference insert, are welded on to the baseplate with 15 mm distance between the shafts. The performance of the two tested types of inserts is compared to the performance of the punctiform reference in regard of the inserts weight. The introduction of inserts and the mould temperature can have substantial influence on the process of mould filling [16–22]. Therefore, numerical simulations regarding both mould filling and curing of the reactive polymer are carried out. Two main goals are addressed by the simulations. Firstly, we evaluate the used process regarding filling time and curing time. Secondly, we derive optimization potential of the manufacturing process regarding filling time and time required to sufficiently cure the resin. Mould filling simulations are performed using a finite volume method (FVM) based approach, which was successfully applied to this kind of problems in the past [16, 17]. This simulation enables a modelling of the air phase as compressible fluid to better predict the formation, movement and closing of air entrapments in the mould filling process. The chemo-rheological behaviour of the resin during mould filling has already been addressed by other authors [18–22]. Most of these studies utilize finite element method (FEM) based approaches. In this study, the FVM based method is extended to take into account the evolution of cure degree and viscosity. They are both driven by temperature and the advancing cross-linking of the polymer, and can have a huge impact on mould filling. A comprehensive review on the chemo-rheology of resins is given by Halley and Mackay [23].

Fig. 1  Exemplary picture and dimensions of the punctiform insert (cf. [15]), all dimensions given in mm

Production Engineering

2 Materials and specimen geometry 2.1 Specimen geometry To allow a comparison of the results to previous work presented in literature, a specimen geometry based on the geometry as in [10, 14, 15] (Fig. 1) was chosen. This geometry represents a punctual load introduction. The base plate on the bottom with a height of 1 mm and a diameter of 30 mm introduces the load into the CFRP structure. For the load introduction only the base plate and the outer diameter of the bolt is relevant. The attachment to other parts can be realized by bolting or welding for example. In the next step the punctual load introduction gets stretched about 30 mm. This results in a linear load introduction (Fig. 2). The attachment to other parts can be varied as well. In this case two chamfers are used to connect the insert to the testing machine. For this type of specimen, a single insert is positioned in the centre of a flat CFRP plate. In Fig. 3 the appropriate dimensions of the CFRP specimens can be seen. The reference insert is manufactured by stud welding a threaded bolt onto a flat baseplate (Fig. 1). The linear inserts were produced by milling. In the case of a larger number of parts, production by a metal powder injection moulding process would be possible, which results in a strong decrease of the production effort, respectively. All types of inserts are made of stainless steel (ANSI 304) to prevent corrosion. The CFRP specimens are built from eight plies of biaxial non-crimp carbon fibre fabric (Hexcel NLT00 series, 0°/90°, 200 g/m2) and an epoxy matrix resin by Sika® (Biresin® CR170/CH150-3) with a resulting fibre volume fraction of 44.4%. During the preforming, the insert is placed between the upper and lower four layers of the carbon fibre fabric.

Fig. 3  Exemplary picture and dimensions of the specimen (c.f. [10]), all dimensions given in mm

Here, the layer structure is symmetrical to the inserts position. For the specimens with the punctiform reference and the insert with three bushings the fibres are guided around the shaft of the insert to maintain the fibre continuity. For the linear inserts a 38 mm long cut in the upper four fabric layers is necessary to put the upper four fabric layers over the inserts linear shaft. Finally, the specimens are manufactured by RTM. The infiltration is carried out on a hydraulic press (Lauffer type RP 400) by using a flow metering and mixing machine (Tartler Nodopur VS-2K). The infiltration pressure is 9 bar and the specimens are cured for 60 min at 70 °C. Further details of the preforming, the RTM process and the qualitycontrol can be seen in [5, 13].

2.2 Concepts of linear inserts The concepts of linear inserts have been derived from already investigated punctiform load introduction elements [15]. This common basis of geometries and dimensions also forms the basis for the comparison of the results of linear and punctiform inserts. 2.2.1 Linear reference insert Figure 4 shows the punctiform reference (left) and the linear reference (centre and right). The furrow, which is milled into the linear shaft, serves to clamp the insert into the tensile

Fig. 2  Dimensions of the linear reference insert, all dimensions given in mm

Fig. 4  Exemplary picture of the linear reference insert with the dimensions shown in Fig. 2

13



Production Engineering

Fig. 5  Exemplary picture of the quasi-linear insert

Fig. 6  Experimental fixture (c.f. [10]), all dimensions given in mm Fig. 7  Experimental setup of the quasi-static tensile tests

testing machine (see chapter III). The width of the groove is ­bN = 2 mm and the height of the groove ­hN = 3 mm. 2.2.2 Quasi‑linear inserts with three welded shafts Figure 5 shows the insert with three threaded shafts. This type of insert is called quasi-linear because of the elongated baseplate of the punctiform reference but in contrast to the linear reference, the load is transferred in to the baseplate by three threaded shafts instead of a linear continuous shaft. This allows, like for the punctiform reference, to maintain the fibre continuity. Care must be taken to ensure a clean and continuous weld around the threaded shaft. Otherwise, premature failure may result from the threaded shafts being pulled out during the tensile test. Fig. 8  Attachment tools for tensile tests

3 Experimental setup To compare the load bearing capacities for quasi-static tensile loads of components using the two different types of linear inserts the fixtures shown in Fig. 6 were used. A crosshead speed of 3 mm / min was selected in the quasistatic tests on a universal testing machine from Zwick. The load is introduced by the two different attachment tools shown in Fig. 7. For the linear inserts a linear force introduction must be ensured. In order to realize this requirement, a furrow for pulling must be inserted in the linear shaft of the inserts. Along this furrow the component is pushed into the attachment for pulling (Fig. 8, right). For the components with the three threaded shafts three screws are passed through the

13

attachment tool holes and screwed into the threads (Fig. 8, left).

4 Numerical methods and simulation model 4.1 Mould‑filling simulation The mould filling simulations are based on the Navier-StokesEquations for continuity (assuming incompressibility)

∇ ⋅ v= 0 and the momentum equation ( ) 𝛿v 𝜌 + ∇(𝜌vv) = −∇p + 𝜇∇2 v + S, 𝛿t

(1) (2)

Production Engineering

with the fluid velocity v, the fluid density ρ , the pressure gradient ∇p , the viscous term 𝜇∇2 v with the dynamic viscosity𝜇 and the sink term S (𝜇 ) 1 S=− + 𝜌|v|F v (3) K 2 S contains the Darcy term K𝜇  , as well as the Forchheimer term 12 𝜌|v|F , which is normally neglected for small fluid velocities as existant in RTM processes. Those governing equations are solved using a finite volume method (FVM) and a volume of fluid (VoF) scheme for interface tracking between the resin and air phase. The simulations are done using the open-source library OpenFOAM®. The Darcy term contains the two influencing material parameters of RTM mould filling simulations. The first is the resin viscosity 𝜇 , which changes during the filling process and can be characterised by chemo-rheological models as described in the following section. The second important process parameter is the permeability K, which has the form of a second order tensor. The permeability depends strongly on fibre orientation and fibre volume fraction. Because of its complicated multi-scale behaviour, the permeability is determined experimentally by using the measurement setup explained in [24]. The results are shown in Fig. 9. As in the experiments, the resulting fibre volume fraction in the part is 44.4%.

4.2 Chemo‑rheological model The viscosity of the resin strongly depends on the ongoing cross-linking process. This relationship is given by a chemorheological model. In this work, this model is composed of firstly a reaction kinetics model, describing the evolution of

Table 1  Parameters of the Kamal-Malkin kinetic model

Table 2  Parameters of the Grindling kinetic model

3.8621e+06 1.0592e+11 6.2878e+04 3.2192e+05 1.5713 1.6296

A1 A2 E1 E2 n1 n2 m c1 c2 K2,diff ,Tg ΔTg

1.6117e+07 6.8176e+04 6.7632e+04 4.8852e+04 3.9227 1.5940 0.8518 2.1388e + 03 7.4994e+03 8.3407e−02 1.0916e+02

the cure degree, and secondly, a model for the cure dependent viscosity. Since the process is carried out under isothermal conditions, a suitable kinetic model will need to take into account vitrification if the whole curing process needs to be considered [25]. However, due to the low degree of cure usually achieved during mould filling, a simple model is sufficient in this stage of the process. Therefore, the Kamal-Malkin model [26] is applied during mould filling and the Grindling kinetic model is used for calculating the remaining part of the process. This switch to a more sophisticated model is vital since during the curing process, vitrification occurs, especially when using low isothermal temperatures. Both kinetic models as well as the applied fitting process and the measurements used have been extensively described in [25]. The used model parameters for both kinetic models are given in Tables 1 and 2. The Castro-Macosko rheology model [27] is used in this study for the description of the cure and temperature dependency of the viscosity of the resin:

𝜂 = 𝜂0 ⋅

[

𝛼g 𝛼g − 𝛼

𝜂0 = B ⋅ exp

Fig. 9  Isotropic permeability for different fibre volume fractions of the biaxial non-crimp fabric

A1 A2 E1 E2 m n

(

]C1 +C2 ⋅𝛼

) Tb . R⋅T

In this equation, 𝛼g is the cure degree at which the material gels (point of gelation). C1 , C2 , B and Tb are model parameters which were identified using a similar approach as was used for parametrization of the kinetic model. For this purpose, rheology measurements at different isothermal

13

Table 3  Parameters of the Castro-Macosko rheology model

Production Engineering

C1 C2 B Tb αg

3.9066 2.1167e−13 1.4142e−12 8.4891e+03 0.72

Fig. 11  Simulation domain and position of inlet and outlet boundary

Table 4  mould filling simulation parameters Inlet pressure Outlet pressure Resin inlet temperature Mould temperature

Fig. 10  Viscosity of the resin at different isothermal temperatures

temperatures were conducted using an Anton Paar MCR 501 parallel plate rheometer. The resulting values are given in Table 3. The point of gelation 𝛼g is gathered from rheology measurements by analysing the temporal evolution of the axial force. As soon as the material gels, stress is build up due to chemical shrinkage leading to a measurable axial force. The corresponding degree of cure is calculated by using the kinetic model. Figure 10 shows the prediction of the chemo-rheological model together with experimental measurements for three different isothermal temperatures.

9 bar 1 bar 50 °C 60, 70, 80, 90, 100 °C

numerical studies, the presence of the insert does not influence the mould filling time [17]. To analyse the influence of the mould temperature on the filling time, different temperatures are simulated for each of the insert geometries: 60, 70, 80, 90 and 100 °C. Those temperatures are set at all wall boundary conditions in the simulation domain. This leads to a total of ten simulations performed and analysed in this work. The viscosity of the resin is modelled with the chemo-rheological model explained in Sect. 4.2. The process parameters are summarized in the following Table 4.

5 Results 5.1 Quasi‑static tensile tests

4.3 Simulation model

5.1.1 Types of failure behaviour

Two simulation models are used in this work, first the mould filling containing the linear reference insert and second the quasi-linear insert (cf. Figs. 4, 5). Analoguos to the RTM tool, inlet and outlet boundaries are located in opposite corners of the cavity. The threedimensional simulation domain of the linear reference insert is shown in Fig. 11. The resin flows above and under the embedded insert baseplate and around the insert shafts while impregnating the fibre preform. As was shown in prior

In general, three different failure behaviours are observed, which are shown in Fig. 12. Failure behaviour A results in failure due to fibre fracture in combination with inter-fibre fracture. Frequently the fibre fracture occurs crosswise at one end of the linear shaft of the insert. One possible reason for this is the stress concentration at the ends of the inserts. Due to the alternating 0°/90° lay-up of the biaxial laminate, fibre fracture starts in one layer and in the transverse layer inter-fibre fracture occurs.

13

Production Engineering

to delamination, inter-fibre breakage always occurs. A clear damage to the matrix material is observed. Due to the frequently occurring and very strong delamination, it is assumed that the inter-fibre fracture is caused by the high stress caused by buckling during the curving of the upper laminate layers. Failure behaviour C is a combination of A and B. In this case, the fibre breakage occurs in combination with delamination and inter-fibre fracture. In previous experiments it could not be observed, which type of failure occurs first. Dry areas in the laminates outer layers at the position of the insert, which have been the only fabrication induced defect, had no influence to the load bearing capacity of the tested specimens. 5.1.2 Load bearing capacity and failure behaviour of specimens with linear reference inserts The average load bearing capacity for samples with the linear reference inserts is 5.805 kN for the maximum load capacity at a standard deviation of 0.431 kN (Table 5). In the graphs, a smaller or larger decrease of force is observed due to first damages. Furthermore, it can be seen in Fig. 13 that there is a strong variation in the Table 5  Comparison of the punctiform reference inserts with a linear insert and a quasi-linear one

Punctif. ref. Linear ref. Quasi-linear

Fmax in kN

St. dev. In kN

Area related ­Fmax in N/ (mm2)

4.884 5.805 7.317

0.151 0.431 0.366

6909 3613 4554

Fig. 12  Different types of failure. a Fibre fracture (red ellipses), b delamination (orange rectangle, more obvious in the side view), and inter-fibre fracture (yellow ellipses) c fibre fracture, inter-fibre fracture and delamination. (Color figure online)

Failure B is characterised as a delamination of all upper layers. This frequently results in a strong lifting off of the upper laminate layers. In addition, the adhesive bond between insert and laminate dissolves. The load introduction element is then only held by the form closure with the upper laminate layers in the component. The delamination usually extends over the complete free area of the samples. It is noticeable in the linear samples that the delamination occurs more frequently and occurs in a stronger form than in the case of the punctiform inserts (Fig. 12b). In addition

Fig. 13  Force–displacement curves for the linear reference insert

13



force–displacement curves of individual samples. Due to these differences, stiffness is difficult to determine. Especially in the case of samples with a failure behaviour of type B and C, very flat curves are recorded. The stiffness is presumably affected by the delamination and the associated lifting of the upper laminate layers. Table  5 shows the average maximum load bearing capacity of the various reference inserts. The linear reference insert achieves an increase in the load bearing capacity by 19% compared to the punctiform reference (5.8–4.88 kN) (Table 5). However, the linear reference inserts have two decisive disadvantages. Firstly, it weighs about five times the punctiform reference, 46 g to 9.2 g, which corresponds to a large increase in mass.In the case of the failure behaviour of the samples with linear reference inserts, fibre fracture and delamination can be observed. Often, both failure mechanisms occur in combination (failure behaviour C). More rarely, failure due to pure fibre fracture or pure delamination can be observed. However, all three failures occur in the samples tested. 5.1.3 Load bearing capacity and failure behaviour of specimens with quasi‑linear inserts Figure 14 shows the tensile curves of the quasi-linear inserts. The curves are deviating less compared to the other linear insert variants. The average load bearing capacity is 7.317 kN with a standard deviation of 0.366 kN. In the plots, smaller or larger decrease of force is recognizable by means of first damages. Compared to the other insert variants, the quasi-linear insert is the one that achieves the highest load capacity (Table  5). Compared to the linear reference inserts, the quasi-linear insert achieves an increase of the load capacity

Fig. 14  Force–displacement curve for the quasi-linear insert

13

Production Engineering

by 26% from 5.805 to 7.317  kN. At the same time, the weight is reduced by approximately half (23.5–46 g). The production effort is also significantly lower. Due to the simple method of stud welding, a high degree of automation is achievable. Compared to the punctiform reference inserts, the insert with three threaded shafts achieves an increase in the load capacity from 4.884 to 7.317 kN. This represents an increase of 49.8%. In the case of the failure behaviour of the inserts with three threaded shafts, delamination or fibre fracture with intermediate fibre fracture usually occurs. This corresponds to failure behaviour A or B. A combination of the two behaviours (failure behaviour C) could only be observed for one sample.

5.2 Mould‑filling and curing simulation During production of the specimens, the applied process parameters revealed to be suboptimal with respect to cycle time. Therefore, in order to investigate potential improvements of the manufacturing process, mould filling and curing simulations are performed using four isothermal temperature levels. Since the insert geometry has little effect on the mould filling process, especially on the filling time, results of the simulations are very similar for both investigated insert types. Thus, only results of simulations of the quasi-linear insert are discussed in the following. Figure 15 shows characteristic quantities of the simulation results. Since a constant pressure is applied at the inlet of the cavity, the mould filling time strongly depends on

Fig. 15  Mould filling times, achieved cure degrees at the end of mould filling and cure time required in order to reach a minimum Tg of 70 °C

Production Engineering

the resin viscosity and thus also on cure temperature. Furthermore, similar degrees of cure are achieved at the end of the mould filling process for all investigated temperatures. However, considering the small cavity size and thus short flow paths, this may not apply to larger composite parts. Figure 15 also shows the cure time required to reach a sufficiently high cure degree. This is crucial since the cure state significantly affects the glass transition temperature, which is an important property of the final part. In this study, a minimum Tg,min of 70 °C was targeted. As expected, the required cure time is longest in case of low curing temperatures and can be significantly reduced by using higher temperatures. Figures 16 and 17 show the temporal evolution of the cure degree and the corresponding Tg . Additionally, Tg,min as well as the cure degree 𝛼min , at which this specific Tg is reached, are represented by black straight lines. Dotted lines refer to the cure degree of the material that entered the cavity last and therefore shows the lowest cure degree at the end of the mould filling process. Contrary to this, solid lines relate to the material that shows the highest degree of cure, usually located at the flow front. Since the cure rate slows down at high cure degrees, this spatial gradient in cure decreases with time. Figure 18 shows the spatial distribution of cure degree for both insert geometries at the end of the mould filling process. While the flow front progress is not much affected by the chosen insert geometry, the cure degree field is influenced. Whether this gradient and distribution remains in the manufactured part depends on the cure time.

Fig. 16  Evolution of cure degree for different isothermal cure temperatures after mould filling

Fig. 17  Development of glass transition temperature during cure after mould filling

6 Discussion The goal was to develop, manufacture, test and evaluate the manufacturing process of linear load introduction elements. By means of tensile tests of the manufactured specimens it

Fig. 18  Spatial distribution ( of cure degree ) at the end of mould filling for both insert geometries Tiso = 100◦ C

13



Fig. 19  Comparison of the load bearing capacity of all insert types tested. Dark grey, Fmax in kN. Light grey, the weight related values

could be shown that with linear inserts an increase in the load bearing capacity can be achieved in comparison to the punctiform reference inserts. The highest load bearing capacity is achieved by the quasi-linear inserts. It can be shown that the insert with the three threaded shafts represents an alternative to the punctiform reference insert. It achieves a significant increase in the load bearing capacity compared to the punctiform reference by 49.8% (Fig.  19). This load bearing capacity is achieved by an increase of the mass by two and a half times the punctiform reference. The average load bearing capacity of the other linear inserts are well below those of the insert with three threaded shafts. In addition, their mass is about five times that of the punctiform inserts and the double compared to the insert with three threaded shafts. Five punctiform inserts could be used instead of a linear reference insert, with the same weight of the CFK samples. Whether the significant increase in the mass is worthwhile compared to the small increase in load bearing capacity must be decided on the respective situation. In addition, the integration of several punctiform inserts into one component could result in a similar or better result for the load bearing capacity. In general, a slightly altered failure behaviour between the linear inserts and the punctiform inserts can be observed. The fibre fracture and inter-fibre fractures of both insert types do not differ in their appearance [5, 10, 14]. But in case of the linear inserts the amount of delamination increased. One of the possible reasons is the larger area of the baseplate. The linear inserts have a surface area of 1607 mm2 and the punctiform inserts have an area of only 707 mm2. Due to the clearly enlarged surface area, the linear insert acts as a larger defect in the fibre composite. This results in an increased delamination and a greater degree of delamination between the laminates. The influence of the altered clamping geometry of the tensile testing device on the failure

13

Production Engineering

behaviour of the load introduction elements is not known. In the samples with heavy delamination, a significantly lower stiffness could be observed. Due to the detachment of the upper laminate layers, these are slowly pulled upwards without a sudden drop in the tensile force. In the case of these components, the stiffness is adversely affected by the delamination and the associated lifting of the upper laminate layers. This behaviour has been investigated in the past for shape an optimized punctiform insert [13] for quasi-static and tensile loads [28]. It is noticeable when looking at the force–displacement curves of all linear inserts that the scattering between the different samples is large. Only with the quasi-linear inserts, approximately the same behaviour can be observed. The variance is also reflected in the large standard deviation of the average load-bearing capacity of all linear samples. The standard deviations can be compared in Fig. 19 and Table 5. Simulations of mould filling and curing revealed a strong dependency of necessary cure time on cure temperature. Figures 16 and 17 show that by using a cure temperature of 60 °C, the maximum achievable cure degree and more importantly Tg is not much higher than Tg,min . It is therefore not advisable to use temperatures lower than 60 °C. The currently applied cure cycle with a temperature of 70 °C and a cure time of 60 min is sufficient in terms of achieved cure degree and Tg . Nevertheless, due to the long cure time, specimen fabrication is time-consuming and should be reduced. This can be achieved by using higher temperatures. However, increased temperatures are accompanied by higher residual stresses due to thermal mismatch, which is especially critical in case of hybrid structures. In the present study, this effect has neither been investigated experimentally nor by structural mechanics. It is therefore impossible to derive an optimum cure cycle taking into account both cycle time and structural performance.

7 Summary and conclusions Lightweight construction in multi-material design made of metals and fibre-reinforced plastics sets high demands on the required joining technologies. The usual thermal or mechanical joining methods are not suitable for the bonding of metal to FRP. A good alternative for thin-walled continuous-fibre-reinforced plastics are embedded force introduction elements, so-called inserts. These consist of a flat baseplate and a connection piece to introduce the loads. The so far insisted inserts consisted of a round baseplate and a welded threaded shaft. The geometry leads to a punctual load introduction into the FRP. At the present time, no comparisons of punctiform inserts to linear force introduction elements exist.Firstly, various concepts for linear load introduction elements have been developed. An insert from

Production Engineering

previous studies [15] was used as the basis for the linear inserts. One linear insert and one quasi-linear insert were designed. The samples were tested for load bearing capacity in the tensile test. As can be seen in Figs. 13 and 14 for the samples with the linear reference insert and the quasi-linear insert, despite the obviously higher forces. The failure behaviour of the punctiform reference has already been investigated by Pottmeyer et al. [29] where also in-situ computer tomographic images can be seen. The highest load bearing capacity is achieved for the inserts with the three threaded shafts. It achieves a significant increase of the load bearing capacity compared to the punctiform reference inserts by 49.8% and the linear reference by 26%. The load bearing capacity of the quasi-linear inserts is achieved by an increase of the mass by two and a half times the punctiform inserts. The goal of this contribution was the comparison of the punctiform reference insert with the (quasi-) linear inserts with regard to their load bearing capacity. It is shown that the linear reference insert achieves a slightly higher load bearing capacity than the punctiform reference inserts. However, comparing the area of the baseplate related figures, shown in Table 5, the punctiform reference is still in front. In regard to their weight the load bearing capacity is much lower compared with the punctiform reference inserts (Fig. 19). Whether the increase in the mass is worthwhile against the small increase in resilience must be assessed according to the respective situation. When comparing the quasi-linear inserts with the punctiform reference the load bearing capacity increased by nearly 50%. When comparing the weight related load bearing capacity also the quasi-linear inserts show a significant decrease (Fig. 19). In conclusion for all (quasi-) linear inserts tested it was possible to increase the load bearing capacity of a single insert. In regard to light weight design the weight related load bearing capacity decreased for all the inserts tested compared to the punctiform reference. In addition, the production effort for the linear insert is much higher than the one of the punctiform reference. The production effort of the quasi-linear is only slightly higher, hence three instead of only one bushing have to be welded onto the baseplate. Therefore, it must be decided whether the significant increase in weight and the higher production efforts are justifying the use of linear inserts or whether it is possible to integrate several punctiform inserts. In order to ensure a high quality infiltration of the specimens as well as to evaluate potential improvements to the currently applied cure cycle, mould filling and curing simulations were carried out. As was previously shown, it is possible to infiltrate the components with embedded linear inserts mostly without dry spots [17]. In this work, the

simulations were extended to additionally consider the variation of resin viscosity during the mould filling. A chemorheological model was therefore implemented into the mould filling simulation method. The results of the simulation show a strong influence of the process temperature on the mould filling time. Higher temperatures enable shorter mould filling times. The same is true for the period of time that is required to reach a sufficiently high degree of cure. Since the latter has a great impact on the cycle time, it is favourable to reduce it as much as possible. However, the results of the present study do not allow the derivation of an optimal process temperature since structural performance of the part is neither experimentally nor numerically investigated in this aspect. In future studies, the effect of mould temperature on residual stresses and on the resulting achievable tensile strength should be investigated, which is one major aspect when manufacturing hybrid metal-CFRP components. Acknowledgements  This paper is based on investigations of the subproject 3—“Fundamental research of intrinsically produced FRP-/ metal-composites—from embedded insert to load bearing hybrid structure”—of the priority program 1712 “Intrinsic hybrid composites for lightweight load-bearings”, which is kindly supported by the German Research Foundation (DFG).

References 1. Eriksson I (1990) On the bearing strength of bolted graphite/ epoxy laminates. J Compos Mater 24(12):1246–1269 2. Camanho PP, Lambert M (2006) A design methodology for mechanically fastened joints in laminated composite materials. Compos Sci Technol 66(15):3004–3020 3. Xiao Y, Ishikawa T (2005) Bearing strength and failure behavior of bolted composite joints (part I). Compos Sci Technol 65(7):1022–1031 4. Kolesnikov B, Herbeck L, Fink A (2008) CFRP/titanium hybrid material for improving composite bolted joints. Compos Struct 83(4):368–380 5. Gebhardt J, Fleischer J (2014) Experimental investigation and performance enhancement of inserts in composite parts. Procedia CIRP 23:7–12 6. Ferret B, Anduze M, Nardari C (1998) Metal inserts in structural composite materials manufactured by RTM. Compos Part A: Appl Sci Manufac 29(5–6):693–700 7. Hopmann C, Fecher ML, Lineman L, Bastian R, Gries T, Schnabel A, Greb C (2013) Comparison of the properties of onserts and inserts for a high volume production of structural composite parts. J Plast Technol 9(4):179–206 8. Fleischer J, Gebhardt J (2013)Experimental investigation of metal inserts embedded in composite parts manufactured by the RTM process. In: 13th Japan International SAMPE Symposium and Exhibition, Nagoya, Japan 9. Schwarz M, Magin M, Peil C, Schurmann H (2004) Thin-walled FRP-laminates and local bending moments-incompatible or solvable by a skillful design. In: Internationale Tagung für Verstärkte Kunststoffe und Duroplastische Formmasse, pp 28–29 10. Gebhardt J, Pottmeyer F, Fleischer J, Weidenmann K (2015) Characterization of metal inserts embedded in carbon fiber reinforced plastics. Mater Sci Forum 825–826:506–513

13

11. Soliman E, Al-Haik M, Taha MR (2012) On and off-axis tension behavior of fiber reinforced polymer composites incorporating multi-walled carbon nanotubes. J Compos Mater 46(14):1661–1675 12. Ferguson RF, Hinton MJ, Hiley MJ (1998) Determining the through-thickness properties of FRP materials. Compos Sci Technol 58(9):1411–1420 13. Gebhardt J (2016) Strukturoptimierung von in FVK eingebetteten metallischen Lasteinleitungselementen, Shaker Verlag, Herzogenrath 14. Pottmeyer F, Weispfenning M, Weidenmann KA (2016) Research of the load bearing capacity of inserts embedded in CFRP under different loading conditions. World Acad Sc Eng Technol Intl J Chem Mol Nucl Mater Metall Eng 11(1):1–8 15. Wilkening J, Pottmeyer F, Weidenmann KA (2016) Research on the interfering effect of metal inserts in carbon fiber reinforced plastics manufactured by the RTM process. In: 17th European Conference on Composite Materials, Munich, Germany 16. Magagnato D, Frey M, Bernath A, Steibler P, Henning F (2013) Experimentelle und numerische Untersuchung der Infiltration bei der RTM-Fertigung. In: A. Wanner, K. Weidenmann (Eds.) Verbundwerkstoffe und Werkstoffverbunde. Karlsruhe Institute of Technology, Karlsruhe 17. Seuffert J, Kärger L, Henning F (2017) Simulation of the influence of embedded inserts on the RTM filling behavior considering local fiber structure. KEM, Destin, 742 pp 681–688. https​://doi. org/10.4028/www.scien​tific​.net/KEM.742.681 18. Abbassi A, Shahnazari MR (2004) Numerical modeling of mold filling and curing in non-isothermal RTM process. Appl Therm Eng 24(16):2453–2465. https​://doi.org/10.1016/j.applt ​herma​ leng.2004.03.005 19. Lim ST, Lee WI (2000) An analysis of the three-dimensional resin-transfer mold filling process. Compos Sci Technol 60(7):961–975. https​://doi.org/10.1016/S0266​-3538(99)00160​-8

13

Production Engineering 20. Lin RJ, Lee LJ, Liou MJ (1993) Mold filling and curing analysis in liquid composite molding. Polym Compos 14(1):71–81. https​ ://doi.org/10.1002/pc.75014​0111 21. Shojaei A (2003) Simulation of the three-dimensional non-isothermal mold filling process in resin transfer molding. Compos Sci Technol 63(13):1931–1948. https​://doi.org/10.1016/S0266​ -3538(03)00161​-1 22. Young W-B (1994) Three-dimensional nonisothermal mold filling simulations in resin transfer molding. Polym Compos 15(2):118– 127. https​://doi.org/10.1002/pc.75015​0205 23. Halley PJ, Mackay ME (1996) Chemorheology of thermosets? Polym Eng Sci 36(5):593–609. https:​ //doi.org/10.1002/pen.10447​ 24. Magagnato D, Henning F (2015) Process-oriented determination of preform permeability and matrix viscosity during mold filling in resin transfer molding. Mater Sci Forum 825–829:822–829. https​://doi.org/10.4028/www.scien​tific​.net/MSF.825-826.822 25. Bernath A, Kärger L, Henning F (2016) Accurate cure modeling for isothermal processing of fast curing epoxy resins. Polymers 8(11):390. https​://doi.org/10.3390/polym​81103​90 26. Kamal MR, Sourour S, Ryan ME (1973) Integrated thermo-rheological analysis of the cure of thermosets. SPE Tech Pap(19):187 27. Castro JM, Macosko CW (1982) Studies of mold filling and curing in the reaction injection molding process. AIChE J 28(2):250– 260. https​://doi.org/10.1002/aic.69028​0213 28. Muth M, Pottmeyer F, Weidenmann KA (2016) Investigation of the failure behavior of metal inserts embedded in CFRP under cyclic loading. Int J Eng Res Sci 2(10):71–86 29. Pottmeyer F, Bittner J, Pinter P, Weidenmann KA (2017) In-Situ CT damage analysis of metal inserts embedded in carbon fiberreinforced plastics. Exp Mech 57(9):1411–1422