Applied Composite Materials

, Volume 19, Issue 5, pp 799–812 | Cite as

Influence of Fibre Architecture on Impact Damage Tolerance in 3D Woven Composites

  • P. Potluri
  • P. Hogg
  • M. Arshad
  • D. Jetavat
  • P. Jamshidi
Article

Abstract

3D woven composites, due to the presence of through-thickness fibre-bridging, have the potential to improve damage tolerance and at the same time to reduce the manufacturing costs. However, ability to withstand damage depends on weave topology as well as geometry of individual tows. There is an extensive literature on damage tolerance of 2D prepreg laminates but limited work is reported on the damage tolerance of 3D weaves. In view of the recent interest in 3D woven composites from aerospace as well as non-aerospace sectors, this paper aims to provide an understanding of the impact damage resistance as well as damage tolerance of 3D woven composites. Four different 3D woven architectures, orthogonal, angle interlocked, layer-to-layer and modified layer-to-layer structures, have been produced under identical weaving conditions. Two additional structures, Unidirectional (UD) cross-ply and 2D plain weave, have been developed for comparison with 3D weaves. All the four 3D woven laminates have similar order of magnitude of damage area and damage width, but significantly lower than UD and 2D woven laminates. Damage Resistance, calculated as impact energy per unit damage area, has been shown to be significantly higher for 3D woven laminates. Rate of change of CAI strength with impact energy appears to be similar for all four 3D woven laminates as well as UD laminate; 2D woven laminate has higher rate of degradation with respect to impact energy. Undamaged compression strength has been shown to be a function of average tow waviness angle. Additionally, 3D weaves exhibit a critical damage size; below this size there is no appreciable reduction in compression strength. 3D woven laminates have also exhibited a degree of plasticity during compression whereas UD laminates fail instantly. The experimental work reported in this paper forms a foundation for systematic development of computational models for 3D woven architectures for damage tolerance.

Keywords

3D weaves Impact Damage tolerance Compression after impact 

1 Introduction

Advanced composites have seen a rapid growth in recent years driven by civil aircraft programs such as Boeing 787, Airbus A350XWB and Bombardier C series. Significant volumes of production are coming from wind turbine blades and automotive chassis for battery powered cars such as BMW Megacity. Traditional method of composites manufacturing based on manual prepreg lay-up and autoclave curing is a big bottleneck for the projected increase in volumes. In addition, prepregs are too expensive for high-volume applications. While automated tape laying and fibre placement processes are addressing throughput issues, composites industry is seriously looking at dry fibre preforms in conjunction with resin infusion techniques. 3D woven preforms are particularly attractive because of the reduced part count, their ability to create near-net shapes as well as the presence of through-thickness reinforcement. In addition to manufacturing costs and production rates, damage tolerance has become a major issue for the composites industry. Various resin toughening schemes developed over the years for prepreg systems are not suitable for resin infusion processes due to their high viscosity. As a result, there is an urgent need to develop alternate toughening schemes for dry preforms. Amongst various toughening schemes, inter-laminar fiber bridging is considered to be the most effective method in improving the damage tolerance. This paper investigates 3D woven structures for their ability to resist damage in comparison to 2D laminar structures.

The need for damage tolerant composites transcends all parts of the composites industry common to structures made from carbon or glass fibres for applications in aerospace to marine and wind turbine applications. Perhaps the most high profile and historically most critical applications have been carbon fiber composites in primary aerospace structures. Here the problem of damage is compounded by the fact that damage is commonly sub-surface, undetectable by eye, but can have a considerable detrimental effect on key laminate properties such as compression after impact [1]. The importance of damage tolerance in the aerospace sector is increasing given the trend to manufacture more primary civil aircraft structures from CFRP, including the fuselage which is susceptible to significant airfield abuse (baggage handling, vehicles etc) and from large hail stones. However damage tolerance in general is also of prime concern to the wind turbine industry as it is simply not economic to replace damaged blades in service, especially if the turbines are deployed off-shore [2].

The industry has made considerable progress for a 30 year time period in improving the damage tolerance of all classes of composites. This is exemplified by the improvements in carbon fiber composites intended for service as primary structures in aerospace where the damage tolerance of the composite laminate is usually quantified by measuring the compression after impact strength according to some recognized standard [3]. Low energy impact is a significant in-service problem which results in difficult to detect damage, primarily consisting of delaminations, which can seriously reduce the compression strength of a composite laminate.

The compression after impact properties have increased when measured using the Boeing standard test method from a level of about 170 MPa to almost 400 MPa in this period. The improvements have been largely the result of increases in resin toughness leading to inter-laminar shear strength. The developments have included the introduction of formulated epoxies, followed by rubber toughened systems, thermoplastic toughened systems, through to thermoplastic, phase inverted thermoplastic-epoxies, monolithic thermoplastics and interleaf toughened epoxies.

Mechanisms for damage tolerance for various material systems have been studied by number of authors over the years. Challenger [4] conducted a key workshop on damage tolerance in 1986 funded by Office of Naval Research. This workshop identified two important areas for further study: i) understanding of micro-mechanisms for damage formation and growth ii)development of analytical/computational models. Hull et al [5] conducted a detailed study of the damage morphology and concluded that BVID (Barely Visible Damage) in UD laminates is triggered by intraply cracking and subsequent delamination propagation. Shyr et al [6] reported the importance of number of layers on the mode of failure—fibre fracture dominated a thirteen layer laminate where as delamination dominated a seven layer laminate during an impact event. Edgren et al [7] studied damage tolerance analysis of NCF sandwich composite panels for both un-notched and notched specimens. Naik and Logan [8] studied damage resistance of 3D woven architectures. Damage resistance is investigated by impacting test panels with predetermined impact energy levels and measuring the resulting damage using non-destructive or destructive methods. Damage tolerance, which is the ability of a structure to retain its load carrying capacity, is evaluated using in-plane tensile, compressive and shear tests [9]. Of these tests, compression after impact (CAI) is the most important test method. The magnitude of this damage has been characterized by different groups according to either the area of the damage zone or the width of the damage zone [10]. Delaminations propagate typically perpendicular to the loading direction; this has prompted the use of damage width as being the key damage parameter [11].

2 3D Weaving

3D woven fabrics have the potential to reduce preforming cycle time and at the same time provide through-thickness reinforcement. In this work, we developed a range of 3D woven architectures on a rapier loom equipped with an electronic Jacquard shedding.

Four different 3D weave styles have been developed using 660 Tex, S2 glass yarns: layer-to-layer, orthogonal, modified layer-to layer and angle interlocked weaves (Fig. 1). An attempt has been made to keep the area density of various 3D woven fabrics approximately same in order to compare the results. In addition, (0/90) UD cross ply and 2D plain woven fabrics are also produced with similar density for comparison. Modified layer-to-layer weave has an extra plain weave on top and bottom layers, in comparison to layer-to-layer weave. Only orthogonal fabric has warp stuffer yarns which are followed by binder yarns. All other architectures have no warp stuffer yarns.
Fig. 1

a) orthogonal weave b) angle interlock weave c) layer-to-layer interlock weave d) modified layer-to-layer weave

Conventional rapier weaving machine is utilized to produce all the 3D structures and 2D plain fabric. Considering machine set up time, design for all the structures is done such that warp density remain more or less same and weft density can be varied according to the requirement. Warp yarns are coming through creel arrangements which maintain the individual warp tension, even though they follow different path. 2D plain fabric density is kept such that 3 layers of plain fabric approximately equal density of other 3D architectures. 0/90 Unidirectional fabric is produced on robotic tow placement machine at the University of Manchester [12]. A pin board has been employed in order to place the fibres around the pins and then tufting is done on edges in order to handle the fabric conveniently for infusion.

2.1 Sample Preparation

3D woven as well as 2D laminates were prepared using vacuum infusion process. Standard un-toughened epoxy resin system has been chosen, as the main objective of this study is to understand the influence of fibre bridging provided in 3D woven architectures. Additionally glass fibres rather carbon fibres were used in this study in order to improve the visibility of damage area as well as tow visibility in x-ray tomography images.

Table 1 presents the preform data for the four 3D weaves, 2D woven fabric as well as UD preform. 600 Tex S glass yarn has been used in producing all the samples—only the inter-layer binding has been modified in different samples. Area density, ends/cm, picks/cm and crimp % values for all the weave styles were measured using standard textile measurement procedures and the values presented in Table 1. Table 2 presents the composite laminate data in terms of weave angles, fibre volume fractions. Weave angle, off-axis angle of the undulating yarn, has been measured using image analysis of the tomography images.
Table 1

Preform specifications

Preform Structure

Layer to layer

Orthogonal

Modified layer to layer

Angle interlock

Plain 2D (3 layers)

(0/90) UD

Yarn Count (Tex)

660

660

660

660

660

660

Areal density

2101

1846

1841

1966

2051

1848

(g/m²)

ends/cm

12

16

15.3

14.66

15

16

picks/cm

12

9.66

8.66

11.66

13.5

12

No.of warp layers

6

6

5

6

3

4

No. of weft layers

5

5

4

5

3

3

Warp Crimp % Binder/stuffer

8.2

6

4.3

2.5

  
 

2.7

1.92

 

4.14

0

Weft Crimp %

1.4

0.81

1.67

4.6

1.3

0

Table 2

Laminate specifications

Laminate

Layer--to-layer

Orthogonal

Modified layer-to-layer

Angle interlock

Plain 2D (3 layers)

(0/90) UD

Thickness (mm)

1.83

1.81

1.72

1.86

1.87

1.71

binder weave angle (o)

22

35

15.9

11.2

14

0

stuffer weave angle (o)

 

8.5

   

0

Fibre Volume Fraction %

37.1

35.2

47.0

40.3

48.5

53.1

2.2 Geometry of Weaves

Figure 2 presents the cross-sectional images of 3D woven and 2D woven structures in the warp direction. In the present work, all the tests were conducted in the warp direction.
Fig. 2

X-ray tomography images of weaves

Following observations can be made from the cross-sectional views parallel to warp direction:
  • In layer-to-layer weaving, all the warp yarns are crimped and the individual warp yarns connect two adjacent layers (Fig. 2a).

  • Binder yarns in an orthogonal weave have large waviness and as a result weft yarns do not neatly stack-up (Fig. 2b). This arrangement leads to relatively low fibre volume fractions. Orthogonal woven laminate has the lowest fibre volume fraction (35%). Stuffer yarns are slightly crimped by the nesting of the weft yarns (Fig. 2c).

  • Modified layer-to-layer construction has better fiber volume fraction(47%) due to the presence of additional warp yarns closer to the surface (Fig. 2d)

  • Angle interlocked weave has relatively straight yarn segments inclined at an angle to the warp direction. The weft yarn packing is more orderly. The fiber volume fraction is 40.3%.

3 Impact Testing

All the test panels were less than 2 mm in thickness and hence the impact as well as CAI tests could not be conducted as per ASTM D 7136/7137. In this work, test methodology developed by Hogg [11] for thin samples has been adopted. Laminates were subjected to drop weight impact tests with energy levels ranging from 5 to 30 J, using a modified clamping rig developed for thin laminates (Fig. 3). A test coupon with dimensions of 89 mm x 54 mm was clamped between plates having 40 mm diameter hole in order to prevent excessive bending during the impact test.
Fig. 3

Clamp for drop weight impact test

Five samples per each type of 3D weave, multi-layer 2D weave and UD laminates were tested at each energy level ranging from 5–30 J, using a drop-weight impact tester with an impactor diameter of 20 mm. Damage area as well as damage width of each sample were measured with the aid of image processing software. Figure 4 shows the damage area of all the specimens impacted at 20 J energy as an example. In the UD cross-ply, damage propagated parallel to fibre direction in each ply and the damage appears to be intra-ply (it is difficult to see if there is any inter-ply delamination). In is interesting to see that in 2D woven laminate, in spite of interlacement, there appears to be some intra-ply damage along warp and weft directions in addition to a large area of inter-ply delamination. In all four 3D woven laminates, the damage is highly localized to the area under the impactor—damage does not seem to spread along the tows.
Fig. 4

Impact damage area at 20 J impact energy

It can be seen from Fig. 5 that the damage area increases approximately linearly with the impact energy. 3D weaves exhibit significantly lower damage area in comparison to 2D woven and UD cross-ply laminates; amongst 3D weaves, Orthogonal weave has the highest damage area and the modified layer-to-layer has the least damage area. UD laminate has the highest damage area, significantly larger than 3D weaves.
Fig. 5

damage area versus impact energy

Naik et al [8] defined damage resistance as the energy required for causing unit area of damage (J/cm2). Damage resistance for each weave style has been calculated as inverse of the gradient of the lines presented in Fig. 5. The damage resistance is lowest for the UD cross-ply, somewhat higher for 2D weaves and significantly higher for 3D weaves (Fig. 6). Amongst 3D weaves, modified layer-to-layer laminte has the highest damage resistance followed by layer-to-layer and angle interlocked laminates; orthogonal laminate has the least damage resitance amongst 3D weaves but significantly higher than 2D and UD.
Fig. 6

Damage resistance of various weave styles

4 Compression After Impact Testing

Figure 7 shows the compression after impact fixture developed by Hogg [11] for thin compact samples. In this fixture, test specimen are supported and held in place by two vertical clamps (instead of knife edge supports used in the Boeing fixture). This fixture attempts to minimize the sample size and hence the chances of buckling.
Fig. 7

CAI rig developed for thin samples

Compression tests were conducted on both the undamaged as well as the damaged samples using the fixture shown in Fig. 7. There is a gradual reduction in the compression strength with respect to the impact energy. Figure 8 presents compression strength versus impact energy for all laminates, normalized to 50% fibre volume fraction.
Fig. 8

CAI strength versus impact energy

UD cross-ply has the highest undamaged strength followed by angle interlock weave; layer-to-layer weaves have the lowest compression strength due to tow waviness; orthogonal weave has median compression strength due to the presence of combination of stuffer and binder tows. Compression strength of a 2D woven laminate is closer to orthogonal weave. Rate of change of compression strength with respect to impact energy is somewhat similar for all four 3D weaves, and comparable to the UD cross-ply. 2D woven laminate appears to have steeper reduction in compression strength with respect to impact energy.

4.1 Influence of Weave Angle on the Compression Strength

It may be observed from the Fig. 8 that compression after impact strength of composite laminates is primarily a function of undamaged compression strength. Undamaged compression strength is in turn a function of tow waviness in the loading directions. There have been several studies on the influence of off-axis angle on compression strength [13, 14, 15]. Emehel et al [16] presented a tow collapse model for 3D woven and braided composites, based on Budiansky’s model.

Tow waviness angles have been measured from the tomography images and the average values are presented in Table 2. Figure 9 presents undamaged compression strength versus average weave angle. Orthogonal woven has two different values for the tow waviness angles, for stuffers and binders; weighted average of the stuffer and binder angles has been used for plotting the graph. It can be observed from Fig. 9 that there is an approximate linear relationship between the tow waviness angle and the undamaged compression strength. It may be concluded that the undamaged compression strength depends primarily on the tow waviness angle and not so much on the weave architecture.
Fig. 9

Compression strength vs tow waviness

4.2 CAI as a Function of Damage Width

It is a common practice to plot CAI values, normalised to the undamaged compression strength (CAI/undamaged compression strength), against damage width [11]. Figure 10 gives a useful comparison between various material systems. Rate of change of CAI with damage width in 2D woven laminates is higher than UD laminates. In 3D woven laminates, there appears to be no loss in CAI strength up to a critical damage width (~0.5 cm). After this point, there is a sharper reduction in CAI strength in comparison to UD and 2D laminates. However, this aspects needs to be interpreted with care.
Fig. 10

Normalized CAI strength versus damage width

In 3D woven laminates, damage penetrates through the thickness and the damage area may be assumed as a circular hole. Soutis et al [17] predicted CAI strength based on the model developed for open-hole compression. In this work, a simple net-section analysis has been carried out and the results presented in Fig. 11. For each compression after impact test, peak compression load was divided by net section area in order to calculate net-section strength.
Fig. 11

Normalised net-section compressive strength

For damage width in the range of 8–10 mm, ratio of net-section strength to undamaged compression strength is close to unity (Fig. 11). At lower damage levels, this ratio is higher than 1, indicating that the damage area has some residual compression strength.

4.3 Nature of Compression Failure in 3D Laminates

There is a significant difference in the nature of compression failure in 3D woven laminates in comparison to UD laminates. UD laminates exhibit a sharp reduction in stress upon reaching the failure stress (Fig. 12), and there is complete breakage of the sample (Fig. 13b). However, 3D weaves exhibit a gradual reduction in the compression load. Figure 13a shows that in 3D weaves, the binder tows delaminate but not completely break resulting in the formation of a plastic hinge. 3D woven laminates exhibit a degree of plasticity.
Fig. 12

undamaged compression stress–strain curves

Fig. 13

Compression failure in a) 3D woven laminate b) UD laminate

5 Conclusions

The main objective of this study was to understand the influence of inter-laminar fibre bridging found in 3D weaves. The choice of material system (un-toughened epoxy) was to isolate the influence of weave architecture; glass fibre tows were used in order to improve damage visualization. Four different weave styles, orthogonal, angle interlocked, layer-to-layer and modified layer-to-layer were investigated. 2D woven laminates with plain weave as well as unidirectional cross-ply laminates were produced for comparison. Following key observations can be made from this study:
  1. 1.

    Damage area increases approximately linearly with the impact energy; all 3D woven laminates have similar order of magnitude of damage area. Individual differences may be accounted for by the fibre volume fractions (FVF). Amongst 3D weaves, Orthogonal laminate had the lowest FVF and the highest damage area; modified layer-to-layer laminate had the highest FVF and hence the lowest damage area. UD cross-ply laminate exhibited highest damage area with 2D woven laminate having damage area half way between UD and 3D laminates.

     
  2. 2.

    In UD laminates damage appears to spread along the fibre directions and the damage is predominantly intra-ply. In 3D woven laminates, damage appears to be highly localized and spread in the thickness direction (damage area smaller than the impactor area).

     
  3. 3.

    Damage resistance, energy required for damaging unit area, is lowest for UD cross-ply laminate. 3D woven laminates exhibit significantly higher damage resistance.

     
  4. 4.

    Undamaged compression strength is primarily a function of average tow waviness angle and less influenced by the topology of interlacement.

     
  5. 5.

    Rate of change of CAI with respect to impact energy is similar for all 3D weaves; UD laminate also exhibited similar gradient. CAI strength of 2D woven laminates appears to drop at steeper rate compared to other weaves.

     
  6. 6.

    Angle interlocked laminates exhibited highest compression strength values in both the undamaged and damaged samples. In this weave, the warp yarns appear to be un-crimped but inclined at an angle to the loading direction. Orthogonal and layer-to-layer weaves exhibit high crimp and hence have lower compression strengths.

     
  7. 7.

    3D woven laminates have a critical damage width below which there is no apparent degradation of compression strength. UD and 2D woven laminates do not exhibit similar behavior.

     
  8. 8.

    In 3D woven laminates, rate of change of CAI with damage width (after the critical damage width) appears to be steepest (Fig. 10). However, this reduction in strength is also a function of the sample width, as the net-section analysis shows that CAI strength is a function of undamaged width. For larger panels, rate of change of CAI will be less steep.

     
  9. 9.

    3D woven laminates exhibit a degree of plasticity during a compression test and form a ‘plastic hinge’ at the mid-section. UD laminates exhibit a sharp failure where as 2D woven laminates exhibit a degree of plasticity before a sharp failure.

     

It was demonstrated that 3D weaves can significantly reduce the damage area. By improving the fibre architecture if one can reduce tow waviness in 3D weaves then CAI values can be improved at the same time.

References

  1. 1.
    Bibo, G.A., Hogg, P.J., Backhouse, R., Mills, A.: Carbon fibre non-crimp fabric laminates for cost effective damage tolerant structures. Composites Science and Technology 58, 129–143 (1998)CrossRefGoogle Scholar
  2. 2.
    Future perspectives for design and testing of wind turbine blades, A white paper on a rational approach to defects and damage tolerance, Det Norske Veritas AS, Report no WTDK-6022,2008. (http://www.dnv.us/binaries/damage_tolerance-white_paper_2008_tcm153-295072.pdf)
  3. 3.
    D 7136 – “Test Method Measuring the Damage Resistance of a Fiber-Reinforced Polymer Matrix Composite to a Drop-Weight Impact Event,” American Society for Testing and Materials, West Conshohocken, PennsylvaniaGoogle Scholar
  4. 4.
    Challenger, K.D.: The damage tolerance of carbon fiber reinforced composites—a workshop summary. Composite Structures 6, 295–318 (1986)CrossRefGoogle Scholar
  5. 5.
    Hull, D., Shi, Y.B.: Damage mechanism characterization in composite damage tolerance investigations. Composite Structures 23, 99–120 (1993)CrossRefGoogle Scholar
  6. 6.
    Shyr, T.W., Pan, Y.H.: Impact resistance and damage characteristics of composite laminates. Composite Structures 62, 193–203 (2003)CrossRefGoogle Scholar
  7. 7.
    Edgren, F., Soutis, C., Asp, L.E.: Damage tolerance analysis of NCF composite sandwich panels. Composites Science and Technology 68, 2635–2645 (2008)CrossRefGoogle Scholar
  8. 8.
    Naik. R.A., Logan, C.P., Damage resistant materials for aeroengine applications, AIAA Paper 99–1370, 40th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials (SDM) Conference, St. Louis, MO, April 1999Google Scholar
  9. 9.
    Raju, K.S., Smith, B.L., Tomblin, J.S., Liew, K.H.: Impact damage resistance and tolerance of honeycomb core sandwich panels. Journal of Composite Materials 42(4), 385–411 (2008)CrossRefGoogle Scholar
  10. 10.
    Saito, H., Kimpara, I.: Evaluation of impact damage mechanism of multi-axial stitched CFRP laminate. Composites Part A 37, 2226–2235 (2006)CrossRefGoogle Scholar
  11. 11.
    Prichard, J.C., Hogg, P.J.: The role of impact damage in post-impact compression testing. Composites 21, 503–511 (1990)CrossRefGoogle Scholar
  12. 12.
    Sharif,T., Potluri,P.:Robotic preforming of thick near-net composites, Texcomp 9, October 13–15,2008, Newark, DEGoogle Scholar
  13. 13.
    Rosen, B.W: Mechanisms for composite strengthening, in Fiber Composite Materials (ASM,1965) chapter3Google Scholar
  14. 14.
    Budiansky, B: Micromechanics, Computers and Structures, 16(1),3-12(1983)Google Scholar
  15. 15.
    Wisnom. M.: The effect of fibre misalignment on the compressive strength of unidirectional carbon fibre/epoxy, Composites, 21(5),403-407 (1990)Google Scholar
  16. 16.
    Emehel, T.C., Shivakumar, K.N.: Tow collapse model for compression strength of Textile composites. Journal of reinforced Plastics and Composites 16, 86–101 (1997)Google Scholar
  17. 17.
    Soutis, C., Curtis, P.T.: Prediction of post-impact compressive strength of CFRP laminated composites. Composites Science and Technology 56, 677–684 (1996)CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media B.V. 2012

Authors and Affiliations

  • P. Potluri
    • 1
  • P. Hogg
    • 1
  • M. Arshad
    • 1
  • D. Jetavat
    • 1
  • P. Jamshidi
    • 1
  1. 1.North West Composites Centre, School of MaterialsUniversity of ManchesterManchesterUK

Personalised recommendations