Sei sulla pagina 1di 11

HOME | SEARCH | PACS & MSC | JOURNALS | ABOUT | CONTACT US

The efficiency of fibrous reinforcement of brittle matrices

This article has been downloaded from IOPscience. Please scroll down to see the full text article. 1971 J. Phys. D: Appl. Phys. 4 1737 (http://iopscience.iop.org/0022-3727/4/11/318) The Table of Contents and more related content is available

Download details: IP Address: 213.175.169.250 The article was downloaded on 30/05/2008 at 13:04

Please note that terms and conditions apply.

J. Phys. D: Appl. Phys., 1971, Vol. 4. Printed in Great Britain

The efficiency of fibrous reinforcement of brittle matrices


V. LAWS
Department of the Environment, Building Research Station, Garston, Watford, Herts
MS. received 10th May 1971

Abstract. Efficiency factors are proposed describing the effectiveness of utilization of fibre stiffness and strength when a random distribution of extensible fibres is incorporated in a brittle matrix. A length efficiency factor for strength calculation is derived, which includes the effect of sliding friction during fibre pull-out; this reduces to the two previously applied efficiency factors in the limits. The interaction of fibre length and orientation is considered, and general efficiency factors describing the effect of both length and orientation on the strength of randomly reinforced short fibre brittle matrices are derived. 1. Introduction In the field of composite materials, especially those used in building, increasing attention is being given to the possibility of improving the properties of a brittle matrix by the incorporation of fibrous reinforcement, notably metals, polymers and glass. Fibre reinforced gypsum plaster and cement differ from most other fibre reinforced materials in that the matrix failure strain is very much lower than that of the fibre; the matrix is highly porous and the bond between fibre and cement is discontinuous; practical volume fibre fractions are low; and the orientation of the fibres in composite now being produced is approximately random in either two or three dimensions (Ryder 1969). Theories of the elastic response and strength of composites based on the assumption of perfect bonding between fibre and matrix, and alignment of the fibres with the stress axis, do not apply. The problem then arises of determining the efficiency factors of length and orientation that should be applied to describe the effectiveness of the reinforcement when the fibre length is finite, the bond imperfect and the orientation random.

2. Notation The subscripts c, f and m refer to composite, fibre and matrix; o and I refer to orientation and length. The basic symbols are as follows: E is the elastic modulus; G the shear modulus; v Poissons ratio; nz (=.&/Em) the modular ratio; U and u u the stress and ultimate, or breaking, (tensile) stress; L the load; E and EU the strain and ultimate, or breaking, strain; U the volume fraction; I the fibre length; a the fibre cross-sectional area; p the fibre perimeter; 7 the interfacial bond strength; 9 the efficiency factor. Other symbols are defined in the text.
3. The theoretical model Krenchel (1964) has proposed a theory of reinforcement in the elastic range that is based on the assumption that the fibres in the composite support a load only along their longitudinal axes. He assumed that matrix and fibre extend together. If there is no
1737

1738

V. Laws

lateral contraction of the composite, the modulus of elasticity of the composite in the direction of the applied stress is given by
Ec VoEfuf + Emvm (1) where yo describes the effect of fibre orientation and has a value between zero (for orientation perpendicular to the stress axis) and unity (for alignment with the stress axis). The stress supported by the composite at the failure strain of the matrix is then

(2) After the matrix failure strain has been reached, it is assumed that the fibres alone support the stress, so that the ultimate failure stress of the composite is simply
ucu = 70Uf"Vf. (3) Equation (3) implies that the fibres break when the extension in the direction of application of the stress is equal to the breaking extension of the fibre, irrespective of the fibre orientation, and applies only if this condition holds. These relations (1) to ( 3 ) are modified for short fibre composites by the inclusion of a length efficiency factor 71 in the term describing the fibre orientation. In this paper, the efficiency factors that should be applied to fibre reinforced brittle building materials are considered. Combined efficiency factors describing the effect of orientation and length on the composite strength are derived and lead to conditions for reinforcement.

u c = VoEfOfE m '

+ Om'Um.

4. The length efficiency factor Previously two different efficiency factors have been used to describe the effect of the fibre length on the efficiency of reinforcement. Krenchel (1964) has used a factor 771 = 1 - Ic/Z, where lC/2is the minimum length required to allowl the fibre to reach its breaking stress without slipping, and is given by

Bortz and Blum (1968) have used a factor of 1 -lc/21 in their determination of the strength of metal fibre reinforced ceramics. A consideration of the simple model-that is, an elastic response to stress of both components up to the failure strain of the matrix, followed by a region such that the stress at failure is determined by the fibres alone-leads to an efficiency factor that varies with extension. The two regions, elastic deformation and final failure, will be considered separately. 4.1. Elastic region For small extensions of the composite, both matrix and reinforcement can be considered to deform elastically. Provided the shear strength at the interface is sufficient, stress will be transferred from matrix to fibre by shear. Cox (1952), Dow (1963) and others have shown that the tensile stress in the fibre builds up from zero at the ends to a maximum at the midpoint. The average stress in a fibre of length 1 subjected to a strain E$ is

where /3= (Gm/Ef)'/' F(vf), F(vf) being a function of volume fibre fraction. The length efficiency factor is given by the bracketed term in (5) and applies for small extensions where the interfacial bond remains intact. At matrix strains such that the maximum shear stress developed at the interface becomes comparable to the interfacial or

The eficiency of jibrous reinforcement of brittle matrices

1739

matrix shear strength, the two materials will debond progressively from the fibre ends, and stress transfer can then occur only if there is an effective frictional bond at the interface. This could arise for instance if there is a difference in Poissons ratio of fibre and matrix. Let the frictional bond be described by a (static) frictional bond strength rs. The stress in the fibre will then build up linearly from zero at the ends to EfE, over a distance 1,/2= EfExa/prs and the average stress in the fibre is
(5 = (1 - ls/21)

Efe,.

The length efficiency factor to be applied in the elastic region if this mechanism occurs is ~2 = 1 - &/21 and is a function of the extension cX. At matrix failure

It follows from (6) that a factor 1-lc/21 could apply only if the fibre and matrix failure strains were equal, a condition not met in the case of glass reinforced gypsum plaster or cement, since E f U C r : 100EmU. Using the data given by Majumdar and Ryder (1970), the length efficiency factors that apply at composite strains up to the matrix failure strain can be calculated according to (5) or (6) for typical glass reinforced gypsum plaster and cement composites. Calculation shows that, over this region, the length efficiency factor is unlikely to be less than 0.98. 4.2. Composite strength After the matrix has failed, the situation is very different. Consider fibres of length I aligned along the stress axis and arranged so that the fibre ends are uniformly distributed along the length of the composite. A crack perpendicular to the stress axis is assumed to release the fibres over a crack width (separation of the faces of the crack) small compared with the fibre length. The fibres are intersected by the crack so that the shorter lengths of fibre embedded in the matrix are uniformly distributed between 0 and 112. At a crack extension ex, those fibres will slip that have a shorter embedded length less than Ix/2, where Ix/2= EfExa/prS,and the probability that a fibre will slip is

Id.

The average stress supported by all the fibres at an extension ex is

where IC is defined by (4) above. From (7) it follows that the average stress increases with strain to reach a maximum value (6z)max at an extension (Eg)max, where

and

If this maximum (t;s)max is reached at an extension (Ez)max which is less than the fibre breaking strain E f U , the breaking stress of the mat is defined by this maximum (equation (8)). In this case the composite fails by fibre pull-out. If, however, the fibre breaking strain is reached before this maximum stress is developed, the fibres that have not slipped will break, and this maximum is not achieved. The strength of the composite is then given by (7), where e x = qu.

1740 Thus there are two possibilities : (i) If 1< 21,,

V , Laws

($1.
If, after the (static) interfacial bond T~ fails, there is a sliding or ploughing frictional force between fibre and matrix described by a bond strength T d , the average stress supported by the fibres that have slipped at a crack extension eZ is
where i is the average length of the fibre ends that have slipped; that is, i= 12/4. The total average stress supported by the fibres at an extension ex is then

where Zc=$lc(2-

T ~ / T ~ )

and the efficiency factors (9) and (10) are modified accordingly:

The length efficiency factor for fibres satisfying the condition 1 2 2Ic varies between 1 -lc/21 and 1 -lc/l depending on the ratio T d / T s . The bond strength measurements of de Vekey and Majumdar (1968) suggest that 5-4/78 for cements could be as large as # leading to an efficiency factor of 1 -2lC/3l. Bortz (1969) measured the maximum bond strength by pull-out tests and reported that, for uncoated stainless steel fibres in a glass-ceramic matrix, the pull-out stress dropped to zero after the bond was broken. Putting T d = 0 in (1 1) leads to a length efficiency factor of 1 - lc/Z and overcomes the large discrepancy between his theoretical and experimental strength results for uniaxially aligned composites. Using this factor, it is unnecessary to include the matrix strength contribution in the calculation, an assumption that it is difficult to accept at the high strains involved. 5. The orientation factor Cox (1952) has derived expressions for the elastic moduli of mats of fibres, assunung that the fibres are stiff and that they support a load only in tension. The two cases of interest here are the planar mat and the solid mat, in which the fibres are arranged randomly in two and three dimensions respectively. Cox has shown that the ratio of the specific Youngs modulus of the mat to that of the fibre is Q for the planar random mat and for the solid random mat; the Poissons ratios of the mats are Q and 4respectively. Krenchel (1964) has likewise considered the deformation of the fibrous component, and has added a matrix contribution, assuming that both matrix and fibrous mat suffer equal extensions and that Hookes law applies. He showed that the Youngs modulus of the composite is

The eficiency of fibrous reinforcement of brittle matrices

1741

The efficiency factor defined by (1) and derived from (12) varies with volume fibre fraction and with the modular ratio m (=Ef/E,). The departure from Q is, however, insignificant in most practical cases. If the Poissons ratio of the matrix is zero and the fibrous mat is constrained so that it is subjected to deformation only in the direction of the applied stress, the Youngs modulus of the composite is simply
Ec=$Ef~f+ EmJ

and the efficiency factor is obviously 9. At matrix failure, it is assumed that a crack is formed perpendicular to the stress direction and that the composite strength is determined by the strength of the released fibrous mat. The fibres will fail in sequence as they reach their breaking strain, beginning with those aligned with the stress axis. It can be shown that the maximum stress that the fibrous mat can support is reached when the fibres aligned with the stress axis reach their breaking strain. The efficiency factors that apply at composite failure are therefore also and 3 for the unconstrained and the constrained cases respectively. Krenchel derived an efficiency factor of Q for a composite containing fibres randomly distributed in three dimensions, and constrained to prevent lateral contraction, but did not consider the unconstrained case. This follows simply froin Coxs analysis, when a matrix contribution is added in a manner similar to that used by Krenchel in the two-dimensional case, and is outlined below. Consider a composite consisting of an (unconstrained) fibrous mat of volume fraction uf. Let a stress uz be applied to the composite in the x direction. The strains eC, cy and eZ of the matrix are assumed to be equal to those of the fibrous mat. Assuming that the matrix is isotropic and the strains are small, the stresses in the matrix can be calculated from the generalized Hookes Law. The stresses induced by the strains e x , eZ/ and eZ are

and similarly for a m y and umz. Writing k=vm/(l vm) (1 -2vm) and using the elastic constants derived by Cox, the total stress in the x direction is ux and in the y and z directions is zero, so that

uz = kEm{ ex( 1 - vm)

+ vm( + et)} + +Eftif( + Q EU ++


~y
2

EZ)

O=kEm{~y(l-

v m ) + vm(ez+

~ z ) ) + + E f v f ( + ~ z~+ + + e z ) . y
2( 15kvm mvf) 15(15k 4)

Putting ey = eZ and solving for Ec (= u Z / e p ) and vc (= - ey/cZ), we have

k( 1 - vm) + +mvf and


vc =

+ +

15kvm mvf 15k 4mvf *

The efficiency factor of orientation over the elastic range of the composite, derived from (13), is a function of volume fibre fraction and of modular ratio m. The departure from Q is, however, insignificant in most practical cases. Since the composite breaks when the extension of the mat is equal to the fibre breaking strain EfU, the efficiency factor of Q also applies in the calculation of the strength of the composite. The factor is obviously if the mat is constrained to prevent lateral contraction.

6. The interaction of orientation and length efficiency factors We consider now the efficiency factor for strength of composites containing short fibres arranged randomly in two or three dimensions.

1742

V. Laws

6.1. Fibres random in two dimensions Consider a fibre in a (two-dimensional) composite at an angle 8 to the stress axis. The extension of the fibre along its axis is related to the extensions e x and by
Ef = ex

cos2 8 +

sin2 8.

It follows that, since the length efficiency factors all depend on the extension q, they therefore depend on the orientation 8. We consider here the combined orientation and length efficiency factor for the strength of a two-dimensional random fibre composite when lateral contraction is prevented (eY= 0) and the fibre length is finite. After the matrix has failed by the opening of a crack perpendicular to the applied stress, it is assumed that the extension of the fibrous mat determines the strength of the composite. For an extension of E$ of the fibrous mat held across the crack, the load supported by a fibre at an angle 8 is Lo= E fa= Efa ex cos2 8. f The fibre will pull out from the matrix if its shorter embedded length is less than Z0/2, where 1eP=LeipT8. The average stress supported by all the fibres at 8 is

If there are n fibres per unit length in any direction uniformly separated, the number of fibres at 8 crossing unit length of crack face is n cos 8. The stress contribution of the fibres at 8 to the total stress in the x direction is therefore nGo cos2 8. The average stress of all fibres over all angles is then

On integration, this becomes

. where IC = 2 ufua/p r g The behaviour then follows that of the aligned short fibre composite outlined in 54 above. The composite fails when ex reaches its maximum, at an extension ( E & , % ~ < q u . There are two possibilities : (i) If l<%,
and (ii) If 1>$I,,

Comparison of (14) and (15) with (9) and (10) respectively (length efficiency factors for aligned fibre composites) shows that the combined factor differs from that obtained from the two separate factors ( q = q0ql). The difference is 20 % when I < 51,/3; when 12 5IC/3 the difference depends on the ratio Ic/l. The theory can be modified to take into account a sliding or ploughing frictional bond stress 7 6 after the fibre starts to slip. Reasoning analogous to that outlined in $4 above leads to the two possibilities:

The eficiency of fibrous reinforcement of brittle matrices

1743

2D random

ill, Figure 1. The total efficiency factor 17 as a function of the ratio of fibre length to critical fibre length, when 7 d = T~ (broken curves) and 76 = 0 (full curves). The total efficiency factors as a function of the ratio of fibre length to critical fibre length defined by (4) are plotted in figure 1 for Td = 0 (full curves) and Td = T~ (broken curves). The difference between the two estimates is large at low values of l/lc,decreasing as l/lcincreases.
6 . 2 . Fibres random in three dimensions An analysis similar to that above, for a composite containing short fibres randomly distributed in three dimensions, leads to the following expressions for the efficiency factor for the breaking stress :

(i) If Z < a&',

These total efficiency factors are also shown in figure 1. 7. Conditions for reinforcement The conditions for reinforcement of the elastic modulus and the strength of the composite follow simply from equations (1) to (3) above, modified to take into account both fibre

1744 length and fibre orientation.

V. Laws

These can be written


Ec = E m
(sC=

+Vf(7Ef-E m )
UmU)

(16)

o m + Uf(7EfEmU-

acu = 7 UfUZf

(1 7)

where 7 and 7 are the combined efficiency factors at the elastic limit and at failure of the fibrous mat released after the matrix has cracked. Obviously the elastic modulus Ec is greater than E m , and the stress supported by the E> composite at the matrix failure strain, oc,is greater than a m u provided that ~ E E m , a condition that is independent of volume fibre fraction cf. Since the composite breaks at a stress equal to or greater than ac, this is the only condition for reinforcement. However, if ( q E f - E m ) is small, theimprovement given by (16) and (17) is small. Auseful improvement in breaking strength will result only if, on matrix failure, the fibres can support a stress suitably greater than a m u . The conditions for an improvement of x(with x > 1) then is
a c U = ~ ~ f U ~ f 2 ~ ~ m U .

This condition is obviously dependent on v f ; if we put x= 1, it also leads to a minimum volume fibre fraction necessary to maintain a breaking stress equal to that of the matrix, provided the condition qEf> Em does not apply. This situation could arise, for example, for a three-dimensionally random fibre composite of low modular ratio Ef/Em. The data quoted by Majumdar and Ryder (1970) have been used to calculate the tensile strength of glass reinforced gypsum plaster as a function of volume fibre fraction (figure 2).

0.05 0.1 Volume fibre fraction

0.15

Figure 2. The composite tensile strength as a function of volume fibre fraction when Ic/l = 2 . The broken lines refer to the case when T d = T~ ; the full lines when T d = 0. The ratio of fibre length to critical length l/lc is given as 1.95. Since no data is available on the values to be assigned to the ratio T d / T s , two limits have been chosen, namely Td = 0 (full lines) and T d = T s (broken lines). The difference between the two estimates is large. Apart from the need to have measurements of both T d and T~ in order to test the theory, other difficulties arise when glass reinforced gypsum plaster and cement composites are considered. The critical length almost certainly varies with volume fibre fraction, since difficulties in compaction at higher volume fibre fractions lead to increasing porosity

The eflciency of Jibrous reinforcement of brittle matrices

1745

(Majumdar et al. 1968), and with fibre length, since increased separation of the fibres in the bundle might be expected as the fibre length is reduced. Experimental results (Ryder 1969) strongly suggest that both factors are operating.

8. Conclusions Although the mechanism of fibrous reinforcement of brittle building materials is not yet fully understood, some estimate can be made of likely efficiency factors of length and orientation that should be applied. In the 'elastic' region, defined as the region before the matrix failure strain is reached, the lengthefficiency factor for an aligned short fibre composite is likely to be very nearly unity for practical composites, provided there is a frictional 'bond' at the interface. The orientation factor in this region for a composite containing fibres randomly arranged in a plane is 5 (unconstrained) and (constrained); for a threedimensionally random arrangement of fibres the corresponding factors are & and respectively. These values are approximate only, since the total efficiency factor at the small strains involved depends on fibre length and, if the composite is free to move laterally, on the modular ratio and volume fibre fraction. If the composite strength is determined by the maximum stress supported by the fibrous mat held across a continuous crack perpendicular to the stress direction, efficiency factors of strength can be derived. These are summarized in table 1 for the case where the fibrous

Table 1. Efficiency factors for strength restrained fibrous mat

Orientation Continuous fibres Aligned Random, two-dimensional Random, three-dimensional


1

Efficiency factor Short fibres


tlll'c/ 1- 1 c / l
&lllfc +(l - W l ' C / l )
i&#'C

%
J .

(1< 21c ') (la21c') ( I < +IC '>


(13 $1.') (1 < -'r"c ') (1k +Qlc ')

&(1 + l ' C / l ) -

where IC '= i l c ( 2 - T d / T s )

mat is constrained to prevent lateral movement. It is seen that the combined efficiency factor for strength is not simply the product of orientation and length efficiency factors as has previously been assumed. The formulae derived allow for the frictional force required to pull out a fibre from the matrix after it has begun to slide. In the case of the aligned short fibre composite, the efficiency factors reduce to 1/41, for 1 < 21, and 1 - lc/l for I > 21, when the sliding frictional bond strength Td is zero ; and 1/21, for 1< 1, and 1 - 1,/21 for 12 1, when the, frictional bond strength T d is equal to the bond strength T~ operating before the fibre begins to move. The theory thus reduces to the previously applied factors in the limits. The difference between the two estimates depends on the ratio Ic/l and can be large-for typical glass reinforced gypsum plaster composites with random planar fibre orientation and a fibre length approximately twice the critical length, the difference is about 35 %. Apparently no attention has hitherto been given to the role of the sliding frictional force during fibre pull-out on the strength efficiency factor of fibrous composites, and no measurements of the sliding frictional bond strength have been reported. This would appear to be an important point, warranting further attention.
Acknowledgments

The work described in this paper is carried out as part of the research programme of the Building Research Station of the Department of the Environment and is published by permission of the Director.

1746

V. Laws

References
BORTZ,S . A., 1969, Structural Ceramic Composite Systems (Chicago: I T Research Institute), I AD 692 149, pp. 1-62. BORTZ,S. A., and BLUM,S . L., 1968, Special Ceramics, 1967, ed. P. Popper (London: British Ceramic Research Association), pp. 29-36. Cox, H. L., 1952, Br. J. appl. Phys., 3, 72-9. DOW, F., 1963, General Electric Co. Report R63 SD 61, AD 414 673, pp. 1-42. N. KRENCHEL, 1964, Fibre Reinforcement (Copenhagen : Akademisk Forlag). H., A. J. MAJUMDAR, J., and RYDER, F., 1970, Sci. Ceram., 5, 539-64. MAJUMDAR, J., RYDER, F., and RAYMENT, L., 1968, J. mater. Sci., 3, 561-2. A. J. D. RYDER, F., 1969, Properties of Glass Reinforced Gypsum as a Material, Building Research Station J. Seminar, Notes No. 21 1/69. DE VEKEY, C., and MAJUMDAR, J., 1968, Mag. Concr. Res., 20, 229-34. R. A.

Potrebbero piacerti anche