Rule of mixtures and Halpin–Tsai: predicting lamina properties
The formulas, a worked E-glass/epoxy example checked against a textbook lamina, and an honest account of which predictions you can trust and which are only bounds.
Two constituents, one lamina
A unidirectional composite is a two-phase material: stiff fibers embedded in a compliant polymer matrix. Homogenization replaces this heterogeneous microstructure with an equivalent anisotropic solid described by effective properties — E₁ along the fibers, E₂ transverse to them, the shear modulus G₁₂, and the major Poisson's ratio ν₁₂.
Loading along the fibers puts fiber and matrix in parallel: they share the same strain, so stiffness averages by volume fraction. This is the rule of mixtures, E₁ = Vf·Ef + (1−Vf)·Em, and it is remarkably accurate because the assumption of equal strain is nearly exact for continuous, well-bonded fibers.
Why the transverse direction is harder
Across the fibers the phases act roughly in series — equal stress rather than equal strain — giving the inverse rule of mixtures, a Reuss-type lower bound that underpredicts real laminae because stress actually concentrates between neighboring fibers.
The Halpin–Tsai relation interpolates between the bounds with a single geometry parameter ξ (≈2 for circular fibers). It tracks experimental E₂ data well up to Vf ≈ 0.65 and is the recommended estimate in the calculator. For carbon fibers, remember the fiber itself is anisotropic: its transverse modulus is far below its axial value, so an isotropic-fiber assumption overpredicts E₂.
The formulas, and how much to trust each one
Along the fibres, equal strain gives the rule of mixtures: E₁ = Vf·Ef + (1 − Vf)·Em. The same averaging holds for the major Poisson ratio, ν₁₂ = Vf·νf + (1 − Vf)·νm. Both are genuinely predictive — usually within a few per cent.
Across the fibres, equal stress gives the inverse rule of mixtures, 1/E₂ = Vf/Ef + (1 − Vf)/Em, and the same form for 1/G₁₂. These are Reuss lower bounds, not predictions. Real laminae always come out stiffer, because the phases do not sit in clean series.
Halpin–Tsai interpolates: E₂/Em = (1 + ξηVf)/(1 − ηVf), with η = (Ef/Em − 1)/(Ef/Em + ξ). The parameter ξ is usually quoted as 2 for circular fibres, but treat that as a starting guess rather than geometry — the worked example below shows why.
A worked example: E-glass/epoxy at Vf = 0.55
Take E-glass fibre (E = 72.4 GPa, ν = 0.22, ρ = 2.54 g/cm³) in a DGEBA epoxy matrix (E = 3.45 GPa, ν = 0.35, ρ = 1.20 g/cm³) at Vf = 0.55, and compare the predictions against Daniel & Ishai's measured E-glass/epoxy lamina.
E₁ predicts 41.4 GPa against a measured 39.0 — about 6% high. ν₁₂ predicts 0.278 against 0.28, essentially exact. Those are the two the rule of mixtures genuinely gets right.
E₂ is a different story. The inverse-ROM lower bound gives 7.2 GPa and Halpin–Tsai with ξ = 2 gives 12.9 GPa, while the measured value is 8.6 GPa. The truth sits between the bound and the interpolation, much closer to the bound. Solving backwards for the ξ that reproduces 8.6 GPa gives ξ ≈ 0.41 — nothing like 2. G₁₂ behaves the same way: the bound gives 2.7 GPa against a measured 3.8.
The lesson is not that Halpin–Tsai is wrong; it is that ξ is a fitting parameter calibrated against test data for a particular fibre–matrix system, not a geometric constant you can look up. Quote E₂ and G₁₂ from micromechanics as a range — bounded below by inverse ROM — and never as a single design number.
Why your prediction will not match the datasheet
The usual culprit is Vf, and there is a quick way to check it. Density mixes linearly and is easy to measure, so ρc = Vf·ρf + (1 − Vf)·ρm inverts to Vf = (ρc − ρm)/(ρf − ρm). It is the cheapest fibre-volume-fraction measurement you have, and worth doing before blaming the model.
Now use it as a consistency test, and watch what happens with the numbers above. The measured lamina stiffness E₁ = 39 GPa implies Vf ≈ 0.52. The measured lamina density of 2.1 g/cm³ implies Vf ≈ 0.67. Those cannot both be true of one panel. The disagreement is the diagnostic: it means the fibre datasheet, the resin datasheet and the lamina datasheet are describing three subtly different materials — a different glass grade, a filled resin, or a lamina measured on a panel with its own Vf and void content.
That is the single biggest practical trap in micromechanics. The models are fine; mixing constituent properties from one source with lamina properties from another is what breaks. If you need agreement, take the fibre, matrix and Vf from the same panel you are trying to predict.
After Vf, the usual suspects are: assuming an isotropic fibre when carbon is strongly transversely isotropic (its transverse modulus is a fraction of its axial one, so E₂ comes out far too high); voids, which hit transverse and shear properties hardest; and fibre waviness, which knocks down compressive stiffness in particular.
Limits of these models
All the closed-form estimates assume perfect bonding, uniform fiber distribution, and void-free matrix. Real laminae deviate: voids reduce transverse and shear properties disproportionately, and fiber waviness knocks down compressive stiffness. Treat micromechanics as a design estimate; certify with measured lamina data (ASTM D3039/D3518).