Contents

In a unidirectional fiber composite, direction 1 runs along the fibers and directions 2–3 are transverse. The transverse modulus $E_2$ is matrix-dominated and far harder to predict than $E_1$. It governs matrix cracking and is the critical input for fatigue criteria. Several analytical homogenization models exist; they diverge significantly at high fiber volume fractions.

Voigt and Reuss bounds

The simplest bounds assume either uniform strain (Voigt) or uniform stress (Reuss) throughout the composite.

Voigt upper bound — iso-strain:

\[E_2^{\rm V} = v_f E_f + (1 - v_f) E_m \tag{1}\]

Reuss lower bound — iso-stress:

\[\frac{1}{E_2^{\rm R}} = \frac{v_f}{E_f} + \frac{1 - v_f}{E_m} \tag{2}\]

Both are exact bounds on $E_2$ but loose — in practice the transverse modulus lies far below the Voigt bound for typical fiber/matrix stiffness ratios.

See also: Voigt & Reuss — derivation, E₁ and E₂ forms, comparison chart

The Eshelby Tensor

The Eshelby tensor $\mathbb{S}$ is a 4th-order tensor introduced by Eshelby (1957) [3] to solve the inclusion problem: an ellipsoidal region that would undergo a stress-free eigenstrain $\boldsymbol{\varepsilon}^*$ if unconstrained instead develops a uniform constrained strain inside the inclusion:

\[\varepsilon^c_{ij} = S_{ijkl}\, \varepsilon^*_{kl} \tag{3}\]

The key result is that $\mathbb{S}$ depends only on the shape of the inclusion and the Poisson ratio of the surrounding matrix — not on the moduli contrast. This makes it the central object in inclusion-based homogenization.

For a circular cylindrical fiber (infinite aspect ratio, fiber axis along direction 1) embedded in an isotropic matrix with Poisson ratio $\nu_m$, the non-zero independent components are [4]:

Component Value
$S_{2222} = S_{3333}$ $\dfrac{5-4\nu_m}{8(1-\nu_m)}$
$S_{2233} = S_{3322}$ $\dfrac{4\nu_m-1}{8(1-\nu_m)}$
$S_{2323}$ $\dfrac{3-4\nu_m}{8(1-\nu_m)}$
$S_{1212} = S_{1313}$ $\dfrac{1}{4}$
All others $0$

The zero entries for components involving the fiber axis (index 1) reflect that an infinitely long cylinder imposes no axial constraint — the problem reduces to plane strain in the 2–3 plane. The in-plane components $S_{2222}$, $S_{2233}$, and $S_{2323}$ fully characterize the transverse response.

See also: Eshelby tensor — full derivation for all aspect ratios

Mori–Tanaka

The Mori–Tanaka (MT) method [2] uses the Eshelby tensor to account for interactions between fibers. Each fiber sees the average matrix strain rather than the applied strain. The full tensor treatment accounts for all transverse Eshelby components ($S_{2222}$, $S_{2233}$, $S_{2323}$) and involves solving for the effective transverse plane-strain bulk modulus $k_T$ and shear modulus $G_{23}$ separately:

\[k_T^{\rm MT} = k_m + \frac{v_f}{\dfrac{1}{k_f - k_m} + \dfrac{1-v_f}{k_m + G_m}} \tag{4}\] \[G_{23}^{\rm MT} = G_m + \frac{v_f}{\dfrac{1}{G_f - G_m} + \dfrac{(1-v_f)(k_m + 2G_m)}{2G_m(k_m + G_m)}} \tag{5}\]

Then $E_2^{\rm MT} \approx 4k_T^{\rm MT}G_{23}^{\rm MT}/(k_T^{\rm MT}+G_{23}^{\rm MT})$. Comparing with eqs. (7)–(10) shows that full Mori–Tanaka coincides exactly with the Hashin–Shtrikman lower bound when fibers are stiffer than the matrix ($k_f > k_m$, $G_f > G_m$). The two are therefore plotted as a single curve.

See also: Mori-Tanaka — derivation, MT = HS lower bound identity

Halpin–Tsai

The Halpin–Tsai equations [1] are semi-empirical, with a reinforcement parameter $\xi$ fitted to geometry:

\[E_2^{\rm HT} = E_m \frac{1 + \xi\,\eta\, v_f}{1 - \eta\, v_f}, \qquad \eta = \frac{E_f/E_m - 1}{E_f/E_m + \xi} \tag{6}\]

For circular fibers, $\xi = 2$. The formula is convenient and accurate at moderate $v_f$ but has no rigorous micro-mechanical derivation.

See also: Halpin-Tsai — short fiber generalization (ξ = 2l/d)

Hashin–Shtrikman Bounds

Hashin and Shtrikman (1963) [5] derived tighter bounds using a variational principle with a polarization field. For a fiber composite the bounds apply separately to the transverse plane-strain bulk modulus $k_T$ and the transverse shear modulus $G_{23}$, where for an isotropic phase:

\[k = \frac{E}{2(1+\nu)(1-2\nu)}, \qquad G = \frac{E}{2(1+\nu)}\]

HS bounds on $k_T$ (assuming $k_f > k_m$, $G_f > G_m$):

\[k_T^{-} = k_m + \frac{v_f}{\dfrac{1}{k_f - k_m} + \dfrac{1-v_f}{k_m + G_m}} \tag{7}\] \[k_T^{+} = k_f + \frac{1-v_f}{\dfrac{1}{k_m - k_f} + \dfrac{v_f}{k_f + G_f}} \tag{8}\]

HS bounds on $G_{23}$:

\[G_{23}^{-} = G_m + \frac{v_f}{\dfrac{1}{G_f - G_m} + \dfrac{(1-v_f)(k_m + 2G_m)}{2G_m(k_m + G_m)}} \tag{9}\] \[G_{23}^{+} = G_f + \frac{1-v_f}{\dfrac{1}{G_m - G_f} + \dfrac{v_f(k_f + 2G_f)}{2G_f(k_f + G_f)}} \tag{10}\]

The transverse Young’s modulus follows from $k_T$ and $G_{23}$ via the exact relation for a transversely isotropic solid [6]:

\[\frac{1}{E_2} = \frac{1}{4G_{23}} + \frac{1}{4k_T} + \frac{\nu_{12}^2}{E_1} \tag{11}\]

The last term couples to the longitudinal modulus $E_1$ and Poisson ratio $\nu_{12}$; for stiff fibers ($E_1 \gg k_T$) it is negligible, giving $E_2 \approx 4k_TG_{23}/(k_T + G_{23})$.

The HS bounds are the tightest possible bounds given only phase moduli and volume fractions. Mori–Tanaka coincides with the HS lower bound for $k_T$ when fibers are stiffer than the matrix.

See also: Hashin-Shtrikman bounds — K, G formulation, E₂ conversion, all models compared

Interactive comparison

Adjust fiber and matrix properties; the chart updates live. The shaded region between HS lower and upper bounds shows the range consistent with the given phase moduli.

Python

import numpy as np

def phase_k(E, nu): return E / (2 * (1 + nu) * (1 - 2 * nu))
def phase_G(E, nu): return E / (2 * (1 + nu))

def transverse_modulus(Ef, Em, nuf, num, vf):
    """Returns E2 (MPa) from all models for a given fiber volume fraction."""
    # Voigt and Reuss
    voigt = vf * Ef + (1 - vf) * Em
    reuss = 1 / (vf / Ef + (1 - vf) / Em)

    # Halpin-Tsai (xi=2 for circular fibers)
    xi  = 2
    eta = (Ef / Em - 1) / (Ef / Em + xi)
    halpin = Em * (1 + xi * eta * vf) / (1 - eta * vf)

    # Hashin-Shtrikman bounds (HS lower = full Mori-Tanaka)
    kf, km = phase_k(Ef, nuf), phase_k(Em, num)
    Gf, Gm = phase_G(Ef, nuf), phase_G(Em, num)

    kT_lo  = km + vf / (1/(kf-km) + (1-vf)/(km+Gm))
    G23_lo = Gm + vf / (1/(Gf-Gm) + (1-vf)*(km+2*Gm)/(2*Gm*(km+Gm)))
    hs_lo  = 4 * kT_lo * G23_lo / (kT_lo + G23_lo)

    kT_hi  = kf + (1-vf) / (1/(km-kf) + vf/(kf+Gf))
    G23_hi = Gf + (1-vf) / (1/(Gm-Gf) + vf*(kf+2*Gf)/(2*Gf*(kf+Gf)))
    hs_hi  = 4 * kT_hi * G23_hi / (kT_hi + G23_hi)

    return voigt, reuss, halpin, hs_lo, hs_hi

References

[1] Halpin, J. C. & Kardos, J. L. (1976). The Halpin-Tsai equations: A review. Polymer Engineering & Science, 16(5), 344–352.

[2] Mori, T. & Tanaka, K. (1973). Average stress in matrix and average elastic energy of materials with misfitting inclusions. Acta Metallurgica, 21(5), 571–574.

[3] Eshelby, J. D. (1957). The determination of the elastic field of an ellipsoidal inclusion, and related problems. Proceedings of the Royal Society A, 241(1226), 376–396.

[4] Mura, T. (1987). Micromechanics of Defects in Solids (2nd ed.). Martinus Nijhoff.

[5] Hashin, Z. & Shtrikman, S. (1963). A variational approach to the theory of the elastic behaviour of multiphase materials. Journal of the Mechanics and Physics of Solids, 11(2), 127–140.

[6] Hill, R. (1964). Theory of mechanical properties of fibre-strengthened materials: I. Elastic behaviour. Journal of the Mechanics and Physics of Solids, 12(4), 199–212.

[7] Christensen, R. M. (1979). Mechanics of Composite Materials. Wiley.

At high fiber volume fractions ($v_f > 0.5$), the Ponte Castañeda–Willis (PCW) scheme improves accuracy by encoding the spatial distribution of fibers through a security zone aspect ratio, not just the volume fraction. See PCW bounds post.