Inverse Rule of Mixtures — Transverse Modulus

Turn the ply ninety degrees. The load now has to cross the fibres, and it can only do that by passing through the resin between them. The two phases are no longer side by side sharing a strain; they are in series sharing a stress, and compliances in series add the way resistances do: $1/E_2 = V_f/E_f + (1 - V_f)/E_m$. Reuss published this dual of Voigt's average in 1929, and it is a lower bound in exactly the sense that Voigt's is an upper one.

The consequence is severe and it is the point of the whole subject. Because the compliances add, the answer is dominated by whichever phase is softer, and in a fibre composite that is always the resin. Take a 230 GPa fibre at 60% by volume in a 3 GPa epoxy: $1/E_2 = 0.6/230 + 0.4/3 = 0.0026 + 0.1333$. The fibre term is two percent of the total. Sixty percent of the volume is contributing two percent of the compliance, and the transverse modulus comes out at 7.4 GPa — from a lamina that is 139 GPa the other way up. The same piece of material is nineteen times stiffer along its fibres than across them.

That ratio is the material, and it is the entire difficulty of designing with composites. A metal has one modulus and you can stop thinking about direction. A lamina has two that differ by an order of magnitude, plus an independent shear modulus, plus two Poisson's ratios that are not equal to one another. Everything about laminate design — the ply angles, the balanced and symmetric stacking rules, the whole apparatus of classical lamination theory — exists to manage that anisotropy rather than to exploit it.

The honesty problem here is the opposite of the rule of mixtures. Where the Voigt average is an optimistic upper bound, this one is a pessimistic lower bound, and real transverse moduli come out above it — typically twenty to fifty percent above. The reason is a piece of physics the iso-stress assumption throws away: the stiff fibres are bonded to the resin, so as the ply is stretched transversely they prevent the resin between them from contracting freely in the fibre direction. That constraint stiffens the resin locally, and none of it appears in a model that treats the two phases as compliances in a chain. Halpin–Tsai is the semi-empirical form that puts the answer where measurements actually land, and it is what you should design to.

So what is this equation for? Three things. It is a floor — a transverse modulus measured below the inverse rule means something is wrong with the panel, most likely a poor fibre-matrix bond or a lot of voids. It is the clean statement of why the transverse direction is weak, in a form you can do in your head. And it is genuinely accurate in the one case where the constraint it neglects does not exist: a laminate whose interface has failed, or a dry-fibre preform, really does behave as compliances in series.

Two practical notes. Solving this backwards for $E_m$ is the well-conditioned direction and it is a good way to get an in-situ matrix modulus — the resin as it actually cured between the filaments, complete with whatever the sizing did to the interphase and whatever moisture the part has since picked up. Expect it to read a little below a neat casting, and expect it to fall further hot and wet, which is the condition most composite allowables are cut against. Solving it for $E_f$ is the opposite: transverse stiffness barely depends on fibre modulus, so the inversion magnifies every error and frequently returns nonsense. And there is a subtlety underneath even that — the fibre itself is anisotropic. A PAN-based carbon filament is around 230 GPa along its axis and perhaps 15 GPa across it, because the graphitic planes run lengthwise. The $E_f$ that governs transverse behaviour is not the one that governs longitudinal behaviour, and neither of these two equations can tell them apart.