Load transfer through fibres and interfaces

A fibre must receive force

A short fibre embedded in a matrix gains axial load through interfacial shear along its length. Ends carry less axial force in a simple shear-transfer picture. Length, adhesion and the matrix response determine how effectively reinforcement is used.

Imagine pulling a matrix containing a short straight fibre whose ends are buried inside it. The grips act on the body, not directly on each buried fibre end. The surrounding material must pull tangentially along the fibre’s side. This is interfacial shear: force parallel to the interface, distributed over its area. Axial fibre stress is a different component, acting across a section normal to the fibre axis.

Adhesion and a bonded interphase can transmit shear before separation; surface interlocking or friction can transmit it in other states. The surrounding matrix must sustain the associated deformation and force. A nominally strong chemical bond cannot supply unlimited transfer through a matrix that has already yielded, cracked or detached. Which mechanism is active depends on the actual interface and loading history.

Constituents and composite architecture identifies these roles. Applied force and stress over an area prepares stress as force divided by the stated area. Here we follow one axial fibre path; a directly gripped continuous fibre, angled fibre, woven tow, layer separation or transverse load has different boundary conditions.

Surface shear changes the force from one section to the next

Use a straight circular fibre of constant diameter d. Its cross-sectional area is A_f = πd²/4; the curved surface over a short axial distance Δs has area πd Δs. The circumference πd, not the cross-sectional area, converts surface shear into an axial force increment. Units, scales and orders of magnitude prepares consistent length and area units.

Measure s inward from a traction-free left end. In the left transfer region the surrounding matrix pulls the fibre surface toward that end, while the larger tensile force at the inward cut balances the smaller force at the outer cut and this distributed surface force. Calling the shear magnitude τ, equilibrium gives N(s + Δs) − N(s) ≈ τπd Δs. N is the tensile force carried across a fibre section; there is no acceleration in this quasistatic balance.

Divide by A_f and by Δs: Δσ_f/Δs ≈ 4τ/d, or dσ_f/ds = 4τ/d for a smoothly varying field. The derivative means the spatial rate at which axial stress builds inward; it is not a stress–strain slope or a time rate. If τ varies, add its contributions along the path. On the other half of a symmetrically loaded fibre, the shear direction reverses and stress decreases toward the other free end.

The magnitude and shape of an actual shear field require a constitutive and compatibility model for the matrix, interface and fibre. Elastic shear transfer generally has a curved stress profile. Equilibrium alone does not determine that profile, a composite modulus or the strain needed to mobilise a specified shear.

Make the transfer-distance assumptions visible

For an original teaching comparison, stipulate a uniform limiting shear magnitude τ_lim = 10.0 MPa wherever transfer is fully mobilised, diameter d = 20.0 μm, and zero force at both fibre ends. Assume a straight aligned fibre, no end-face load, no residual stress, no bending, constant diameter and no matrix spalling or fibre breakage. This is a limiting-transfer illustration, not the elastic solution at every applied load and not measured data for a named composite.

Supply a target axial fibre stress σ_t = 600 MPa for a sufficiently long interior under the compared loading condition. It is an input to this picture, not a tensile strength. The mobilised end regions build linearly until they reach that target; a central plateau then needs no further axial shear. A concise profile for full length ℓ is σ_f(x) = min[σ_t, (4τ_lim/d)x, (4τ_lim/d)(ℓ − x)], with x measured from the left end and 0 ≤ x ≤ ℓ. The min operation selects the smallest of its three entries.

One end needs distance a_t = σ_t d/(4τ_lim) to reach the target. With d = 0.0200 mm and 1 MPa = 1 N/mm², the gradient is 4 × 10.0/0.0200 = 2000 MPa/mm. Thus a_t = 600/2000 = 0.300 mm. Both ends need room: a full fibre must have ℓ ≥ 2a_t = 0.600 mm to reach 600 MPa anywhere in this model. A one-end distance and a full two-end length differ by a factor of two.

For ℓ = 0.400 mm, each end has only 0.200 mm to the centre. The maximum is 400 MPa, and the triangular profile’s length-average stress is 200 MPa. The fibre is still carrying load; being too short to reach the chosen target does not mean zero reinforcement.

For ℓ = 1.200 mm, each 0.300 mm end region reaches 600 MPa and a 0.600 mm central region remains at that target. The length-average is 450 MPa. Calculate it as the area under stress versus position divided by ℓ: a 600 MPa rectangle loses two end triangles, together equivalent to a 0.300 mm-wide rectangle. Hence the average is 600 × (1 − 0.300/1.200) = 450 MPa. This is an average along one fibre, not an average across the composite section.

Now separately stipulate a fibre tensile threshold σ_b = 900 MPa. The full strength-based critical length is ℓ_c = σ_b d/(2τ_lim) = 0.900 mm. Below it the fully mobilised two-end transfer cannot attain that threshold in this idealisation. Above it enough length exists in principle, but reaching it still needs sufficient applied loading and the assumed transfer state. The 1.200 mm fibre in the supplied 600 MPa-target comparison has not broken merely because ℓ > ℓ_c. Do not clip a predicted stress at 900 MPa and keep calling the broken fibre intact.

Maximum bonding is not every goal

A weak interface can slip prematurely; an appropriate debonding or pull-out process can also absorb fracture energy. The useful interface depends on whether stiffness, strength, toughness or another response governs the design.

For a load-bearing uncracked region, early interfacial slip may reduce how much fibre stress can be developed and change the deformation. Near a matrix crack, a fibre that remains connected across the opening can bridge it. If the interface debonds and then resists sliding, continuing pullout can consume mechanical work over a substantial displacement. If the fibre breaks early, that long sliding path can be lost.

Those statements do not make the weakest interface best. With almost no sliding resistance, a fibre can leave its socket while consuming almost no frictional work. With excessive force demand, the fibre or surrounding matrix can fail before the assumed pullout is completed. The useful balance depends on direction, fibre length and strength, matrix damage, surface state and the required service response. Damage, delamination and interface failure develops competing composite damage paths.

Debond initiation and sliding resistance are separate properties. A larger force peak can arise from more contact area or friction as well as from a stronger initial bond. It does not alone identify an intrinsic local adhesion strength or prove a larger total work. We therefore calculate the next sliding example separately from the preceding transfer-capacity comparison.

Work comes from the entire force–displacement path

Use a second original model: one side of an aligned fibre has already debonded over a remaining embedded length a = 0.600 mm. Pull it axially from its socket. Set d = 20.0 μm again, but now stipulate a uniform sliding resistance τ_f = 5.00 MPa over the remaining curved contact. The subscript f here labels friction, not the earlier limiting-transfer input. Assume controlled slow extraction, negligible fibre extension and inertia, no rupture or matrix spalling, and no change in diameter or sliding resistance.

After outward displacement u, the remaining embedded contact is a − u, where 0 ≤ u ≤ a. Its surface area is πd(a − u), so the applied resisting-force magnitude is F(u) = πdτ_f(a − u). This force falls as contact is lost, even though the frictional shear value is unchanged. At u = 0 the force is 0.188496 N; at u = 0.300 mm it is 0.094248 N; at full extraction it is zero.

Mechanical work is the sum of force times displacement increments. The force graph is a triangle, so W_slide = ½F(0)a = πdτ_f a²/2. Numerically W_slide = 0.0565487 N·mm = 56.5487 μJ, about 56.5 μJ. One N·mm is 10⁻³ J, or 1000 μJ. This is work for one already-debonded fibre side. It excludes the preceding work to create the debond, stored elastic energy in the fibre and machine, matrix damage, other fibres and any new crack-area normalisation.

Check the intact-fibre assumption before accepting that triangle. The largest tensile stress at the socket entry is F(0)/A_f = 4τ_f a/d = 600 MPa, below the supplied 900 MPa threshold. The model is not contradicted by that screen; this is not proof that a real fibre will pull out intact. If τ_f were doubled to 10.0 MPa with the same a and d, the nominal initial entry stress would be 1200 MPa. The unchanged intact-pullout premise would then fail before the calculated complete sliding path. Do not claim twice the realised work by integrating an unavailable path.

Strength, toughness and fracture resistance distinguishes specimen work from resistance per new crack area. Our microjoules cannot become a bulk composite fracture energy in J/m² without a defined population of bridging fibres, their geometry and histories, a crack-area basis and all other relevant work. Neither transfer-capacity arithmetic nor one pullout curve establishes composite tensile strength or a safe component load.

Measure the path, not just the peak

A useful interface comparison records fibre diameter and shape, actual embedded length, orientation, matrix/fibre state, temperature and conditioning, loading history, local versus machine displacement, and the observed outcome: slip, pullout, fibre rupture or matrix damage. Surface treatment can change more than one property. An experiment on one polymer fibre in cement is evidence for that system, not a universal shear value for a glass-fibre polymer composite.

The convenient ratio τ_app = F_max/(πda) uses a peak force and the initial lateral embedded area. It is an apparent average measure for that specimen and analysis. Debonding and friction can both contribute to F_max, while shear and geometry can vary locally; τ_app is not automatically the local stress that initiated separation. Changing the embedded length or angle can alter both force and failure mode. A model used to infer an interface parameter needs its own geometry and assumptions checked.

The conditional diagrams here help expose those assumptions: zero end force, a stated shear-transfer region, a supplied target, or an already-debonded sliding side. Direction dependence and mixture estimates asks when uniform-strain or uniform-stress approximations are appropriate. Keep that idealisation separate from a fibre whose axial force demonstrably changes along its length.

Worked examples

Try the reasoning

1. Why can longer fibres help in a shear-transfer model?

   Show the answer and reasoning

   More length permits force to build up.

   The improvement still depends on alignment and interface behaviour. Surface shear acts over more distance from an unloaded end, so greater axial force can be built before a target is reached. The actual profile still needs a transfer model.

2. Is all interfacial separation always undesirable?

   Show the answer and reasoning

   No.

   Controlled separation can contribute to toughness. Frictional sliding after a suitable debond can consume work while a fibre bridges an opening; premature slip with little resistance or early rupture can instead defeat the intended function.

3. A calculation calls 0.450 mm the full critical length for d = 20.0 μm, τ_lim = 10.0 MPa and σ_b = 900 MPa. Where is the missing factor, and what maximum does a full 0.450 mm fibre attain in the fully mobilised triangular state?

   Show the answer and reasoning

   0.450 mm is the one-end distance to 900 MPa. The full critical length is 0.900 mm. A full 0.450 mm fibre reaches only 450 MPa at its centre.

   Each free end has 0.225 mm to the centre. Multiply it by the 2000 MPa/mm inward gradient. A single embedded side and a free-ended fibre embedded on both halves have different geometry.

4. Keep τ_lim = 10.0 MPa, σ_t = 600 MPa and full length 0.400 mm, but use d = 10.0 μm. Find the end transfer distance, plateau length and length-average stress. Has a universal composite-strength improvement been proved?

   Show the answer and reasoning

   a_t = 0.150 mm; the plateau is 0.100 mm long; the length-average is 375 MPa. No universal composite-strength conclusion follows.

   The gradient doubles to 4000 MPa/mm. Each end triangle is 0.150 mm wide, leaving 0.400 − 0.300 = 0.100 mm at 600 MPa. Area divided by full length gives 600 × (1 − 0.150/0.400) = 375 MPa. Fibre population, fraction, actual properties, matrix and damage were not supplied for a composite strength.

5. For d = 20.0 μm and full length 0.400 mm, what fully mobilised shear magnitude would be needed just to reach 600 MPa centrally? Does improving chemical adhesion alone establish that value?

   Show the answer and reasoning

   15.0 MPa in the assumed constant-shear geometry; chemical adhesion alone does not establish that the matrix/interface path can sustain it.

   The available one-end distance is 0.200 mm, so τ = σ_t d/(4s) = 600 × 0.0200/(4 × 0.200) = 15.0 MPa. This rearranges equilibrium under its assumptions; it does not supply a matrix shear response, a treatment recipe or a verified interface parameter.

6. A different 0.400 mm fibre has a 0.050 mm end zone at each end that transfers no shear and carries no end-face load. Elsewhere use d = 20.0 μm, τ_lim = 10.0 MPa and target 600 MPa. What is the central maximum and the average over the complete fibre? Assume no friction in those inactive zones.

   Show the answer and reasoning

   The active span is 0.300 mm, the central maximum 300 MPa and the whole-fibre average 112.5 MPa.

   The two inactive spans carry zero axial force under these stipulated boundaries. Each active half is 0.150 mm, giving 300 MPa. The triangular stress-area is ½ × 0.300 × 300 = 45.0 MPa·mm; divide by the complete 0.400 mm length. This is not a rule that every debonded zone has zero transfer—sliding friction can still act there.

7. In the already-debonded pullout model, after u = 0.300 mm what force remains? How much sliding work has been done, and why is that not half the complete work?

   Show the answer and reasoning

   F = 0.094248 N; work so far is about 42.4 μJ, or three quarters of the complete 56.5 μJ.

   The remaining embedment has halved, so force halves. The first half-displacement sweeps a trapezoid of area ½(0.188496 + 0.094248) × 0.300 = 0.0424115 N·mm. The remaining triangle is one quarter of the original area. Force was greater during the first half.

8. A separate specimen has d = 20.0 μm, initial one-sided embedment a = 0.400 mm and measured peak force 0.160 N. Calculate τ_app = F_max/(πda). Does this identify the local debond-initiation stress or prove friction is absent?

   Show the answer and reasoning

   τ_app ≈ 6.37 MPa. Neither local debond stress nor zero friction is established by that ratio.

   The initial lateral area is π × 0.0200 × 0.400 = 0.0251327 mm². Divide 0.160 N by it. The ratio averages a peak over an initial geometry; local stress variations, changing debond length, friction and specimen response remain unresolved.

What to carry forward

• Trace how force enters reinforcement.
• Judge interfaces against the intended mechanism.
• Use the curved interface area to balance shear transfer and the fibre cross-section for axial stress.
• State whether a distance is one-sided or a full two-ended fibre length.
• A fibre strength threshold is not automatically reached and is not a composite strength.
• Integrate only a mechanically possible pullout path; one-fibre work is not bulk fracture energy.