Composite DFOS Sensors — Manufacturing, Calibration and Deployment in Composite Materials

Distributed fibre-optic sensors (DFOS) embedded directly into composite materials turn the structure itself into its own measurement system. We cover fibre selection, embedding techniques, strain-transfer modelling, calibration against reference extensometry, and typical pitfalls when deploying in CFRP, GFRP and aramid composites.

Abstract

Composite structures fail from within. Delamination between plies, matrix cracking around a dropped tool strike, fibre breakage under cyclic bending — none of these announce themselves on the outer surface before they have progressed well beyond design limits. Surface-bonded strain gauges and ultrasonic inspection catch the symptoms late. Distributed fibre-optic sensors (DFOS), embedded directly between the plies during layup, give the structure an internal nervous system: a continuous strain and temperature profile along the fibre route, read by an interrogator at the end of the laminate. This article covers the four decisions that determine whether a composite DFOS installation will still deliver usable data after a thousand load cycles: fibre and coating selection, embedding method, strain-transfer modelling, and the calibration chain that ties the fibre reading to a traceable mechanical reference. We close with the failure modes we have seen in practice and a short set of recommendations for the first deployment.

Keywords: composite DFOS sensors, distributed fibre-optic sensing, CFRP, GFRP, structural health monitoring, fibre embedding, strain transfer, Rayleigh backscatter.

1. Why Embed Fibres into Composites

Composite materials — carbon-fibre reinforced polymer (CFRP), glass-fibre reinforced polymer (GFRP), aramid-reinforced polymer — are heterogeneous by design. Mechanical load is carried primarily by the reinforcement; stiffness and damage tolerance are dominated by the matrix and by the interface between the two. The failure modes that matter in service — interlaminar delamination, matrix microcracking, fibre pull-out — all originate below the top ply.

Three measurement strategies are available:

• Surface-bonded strain gauges and surface-attached FBG patches are the simplest to install and the easiest to replace, but they see only the outer-ply strain. Interlaminar separation at depth produces a surface signal that is both delayed and attenuated.
• Periodic non-destructive testing (ultrasonic, thermographic, X-ray CT) detects damage after it has reached a measurable extent. For critical structures with scheduled inspection intervals this is acceptable; for structures with sudden onset of load — wind turbine blades in an unexpected gust, aircraft structures under bird strike — it is not.
• Embedded distributed fibre-optic sensors place the measurement inside the laminate. A single polyimide-coated single-mode fibre, routed between plies during layup, becomes thousands of effective strain gauges distributed through the structure. The information is continuous in space, available in real time, and resilient to electromagnetic interference.

The trade-off is clear: embedding is more expensive at manufacture and demands discipline at the ingress and egress points, but produces a monitoring capability that cannot be bolted on later.

2. Fibre Selection

The sensor is a telecommunications-grade fibre adapted to a mechanical role. Three attributes determine whether the fibre survives the composite environment: core geometry, cladding diameter, and coating chemistry.

2.1 Single-mode vs. Polyimide-coated

Standard single-mode fibre (ITU-T G.652, SMF-28 compatible) with an acrylate coating is the default in telecoms. For composite embedding, two problems arise:

• Acrylate coatings degrade thermally above about 85 °C. Standard CFRP cure cycles reach 120–180 °C, and some aerospace-grade systems exceed 200 °C.
• Acrylate is chemically compatible with most resins but exhibits noticeable creep under sustained strain, which translates into drift of the strain reading.

Polyimide-coated fibres solve both: thermal stability to approximately 300 °C, far lower creep, and better adhesion to epoxy and bismaleimide matrix systems. The total fibre diameter remains close to the acrylate variant (approximately 155 µm vs. 245 µm), which matters when the fibre is placed between plies of 0.125–0.25 mm thickness — a lower profile means less local disturbance of the reinforcement geometry.

For temperatures above 300 °C (ceramic matrix composites, metal matrix composites), copper or aluminium metallised coatings are available, with corresponding increase in cost and reduction in the practical bend radius.

2.2 Coating Compatibility with the Matrix

The coating–matrix interface is a shear-transfer joint. Good adhesion is desirable for strain transfer, but perfect adhesion introduces a stress concentration at the fibre surface that can nucleate matrix cracks. The compromise found by most industrial deployments is polyimide + silane coupling agent for epoxy matrices, which gives shear strength sufficient to transfer microstrain-level deformation without initiating cracks around the fibre at typical service loads.

Ormocer-coated fibres are a specialised alternative with higher temperature tolerance and tunable mechanical properties, used predominantly in research and in hot-wet aerospace applications.

2.3 Bend Radius and Routing

Single-mode fibres have a minimum long-term bend radius of about 15 mm — below this, macrobending losses accumulate and the fibre becomes unreliable. In a composite layup this is less restrictive than it sounds for straight routing, but it is a hard constraint around egress points, where the fibre leaves the laminate at a controlled angle.

Specialised bend-insensitive fibres (ITU-T G.657) reduce the practical radius to 5–7 mm and are worth the price premium wherever the fibre has to turn tightly inside the structure.

3. Embedding Techniques

The embedding method is dictated by the composite manufacturing process. Three families cover almost all current deployments.

3.1 Between Plies in Prepreg Layup

For prepreg-based parts (hand layup, automated fibre placement, tape laying), the fibre is laid directly on a cured or uncured ply, oriented parallel to the reinforcement direction of the surrounding plies, and covered with the next ply. Three details matter:

• Fibre orientation. The optical fibre must run parallel to the surrounding reinforcement fibres, not across them. A transverse orientation creates a resin-rich region around the optical fibre that acts as a stress concentrator and nucleates matrix cracks under cyclic load.
• Fibre routing avoids ply drops and taper regions. These are already strain concentrators in the composite itself; embedding a fibre there measures a local anomaly, not the structure’s intended loading.
• Ingress and egress through a dedicated strain relief. The fibre leaves the laminate through a soft silicone or PTFE sleeve that absorbs the step change from fully embedded to free-standing. The transition region, if left unprotected, is where more than half of all embedded fibres fail.

3.2 Infusion Processes (RTM, VARTM, Infusion Moulding)

In resin transfer moulding and vacuum-assisted variants, the dry fibre preform is laid up, the optical fibre is placed between layers of the preform, and resin is drawn through. The fibre survives the infusion pressure and the subsequent cure well, provided two conditions are met:

• Protection at the preform edge. The fibre crosses the vacuum bag or the mould parting line through a dedicated port. A capillary tube bonded into the port protects the fibre from pinching between the mould halves and from being torn by resin flow.
• Avoidance of race-tracking. The optical fibre, being smaller than the preform tows, can create a channel for preferential resin flow. For large parts, this manifests as a visible resin-rich line following the fibre, with locally reduced fibre volume fraction and reduced stiffness. The remedy is to run the sensor fibre in close proximity to the preform’s natural flow channels (seam allowances, edge regions) and to validate on a sacrificial part before scaling up.

3.3 Pultrusion

Continuous pultrusion of rods and profiles allows the optical fibre to be fed in alongside the reinforcement at the die entrance. The fibre becomes a permanent instrument built into every metre of produced profile. Typical pultrusion speeds of 0.3–2 m/min are compatible with standard polyimide-coated fibres; the main constraint is the die geometry, which must accommodate the fibre without pinching it.

Pultruded sensor-embedded profiles are an emerging product category — used in bridge decking, wind-turbine spar caps, and composite reinforcement bars (CFRP rebar).

4. Strain Transfer Modelling

An embedded fibre does not experience the exact strain of the surrounding composite. The transfer is mediated by the coating and by the local resin region around the fibre, and this transfer has a characteristic length over which the fibre strain converges to the composite strain. Below this length, the fibre reads a smoothed, attenuated version of the real strain field.

The canonical treatment is the shear-lag model developed for optical fibres by Her and Huang [1], adapted from the classical fibre-matrix pull-out analysis. The key output is the effective gauge length:

$$ \lambda = \sqrt{\frac{E_f r_f^2 \ln(r_c/r_f)}{2 G_c}} $$

where $E_f$ is the fibre Young’s modulus, $r_f$ the fibre core radius, $r_c$ the outer coating radius, and $G_c$ the coating shear modulus. For standard polyimide-coated single-mode fibre embedded in epoxy, $\lambda$ is on the order of 1–5 mm.

Practical consequence: a strain feature in the composite narrower than about $2\lambda$ (a sharp crack-tip field, a localised impact-damage zone) will be read by the fibre as a wider, lower-amplitude disturbance. The interrogator’s own spatial resolution (1 cm for standard OFDR, 0.5 m for BOTDA) sets a second lower bound, and the real spatial resolution is the larger of the two.

For design purposes:

• Strain features > 10 mm wide: fibre reads the composite strain faithfully.
• Features 2–10 mm wide: fibre reads a smoothed version; magnitude is correct within 20–30 %.
• Features < 2 mm wide: fibre registers that something is there, but amplitude is not directly interpretable.

For crack detection this is acceptable — the appearance and growth of a strain anomaly is the operationally relevant signal, not its precise magnitude.

5. Calibration

Calibration of an embedded fibre is complicated by the fact that the reference instrument cannot be at the same place as the sensor. Three approaches are used in combination.

5.1 Interrogator Calibration

The interrogator itself is calibrated against traceable wavelength and length standards at the manufacturer or at an accredited calibration laboratory. A typical OFDR instrument is characterised in terms of wavelength accuracy (±1 pm), distance accuracy along the fibre (±10 cm over 100 m), and strain repeatability (±1 µε in a thermally stable environment). This calibration is renewed on a defined interval (annual for most industrial users) and documented against national standards (NIST, PTB, NPL).

5.2 In-situ Reference against Mechanical Extensometry

During the commissioning of a composite structure, the embedded fibre is read simultaneously with a surface-bonded strain gauge or a mechanical extensometer under known applied load. The two readings must agree in the region where the fibre is near the surface (within the strain-transfer length of the surface instrument’s gauge). Agreement validates the entire chain: fibre, coating, embedding, interrogator.

A commonly accepted criterion is agreement within 5 % at 1 000 µε applied strain under isothermal conditions. Larger deviations point to a problem — poor embedding, coating damage, interrogator offset — which must be identified and corrected before the structure enters service.

5.3 Thermal Compensation

Rayleigh and Brillouin backscatter respond to both strain and temperature. In a composite structure the two are not independent — a thermal gradient produces a thermal strain, which the fibre reads as a mechanical strain unless compensated.

Three compensation strategies are standard:

• Dummy loop. A section of fibre embedded in the structure but mechanically isolated from the reinforcement (housed in a slack-loop capillary tube) reads temperature only. Its signal is subtracted from the active sensor fibre.
• Co-located distributed temperature sensing. A separate Raman DTS fibre runs parallel to the strain fibre and provides the local temperature profile. Suitable for larger structures where a separate fibre is affordable.
• Multi-measurand interrogation. Combined Rayleigh/Brillouin interrogators can separate strain and temperature from a single fibre at the cost of increased noise. Available in high-end research instruments; mainstream industrial installations still prefer the first two approaches for their simplicity.

6. Case: Wind-Turbine Blade (CFRP Spar Cap)

A 60-metre offshore wind-turbine blade with a CFRP spar cap illustrates the full chain. A single polyimide-coated single-mode fibre is routed along the spar cap during prepreg layup, running parallel to the main carbon reinforcement. The fibre enters the blade at the root through a dedicated capillary port, runs the full length of the spar cap, and returns through a second capillary — forming a loop-back topology that doubles the interrogator’s effective signal-to-noise.

Installation effort: approximately 2 % added to the layup time per blade. Interrogator location: at the blade root, inside the hub, with the data link running down the tower to the nacelle controller.

Operational use: * Continuous strain profile along the spar cap at 1 Hz sample rate. * Baseline recorded during the first week of operation under representative wind conditions. * Deviations from baseline flagged automatically when any 1-metre section exceeds a threshold deviation sustained over 10 minutes. * Early detection of trailing-edge debonding, which produces a characteristic strain redistribution well before visual inspection would reveal the problem. * Over the design life (20–25 years), the fibre is read as part of condition-based maintenance; blades flagged by the DFOS data are prioritised for drone-based visual inspection.

The economic case rests on avoided unplanned downtime. For an offshore turbine, a single avoided catastrophic blade failure justifies the DFOS installation for the entire fleet.

7. Common Failure Modes

The failures we have observed, in order of frequency:

• Fibre breakage at the ingress or egress point. Approximately half of all failures. Caused by inadequate strain relief at the laminate-to-free-fibre transition, by sharp bending under the vacuum bag during cure, or by handling damage during final assembly. Remedy: a dedicated capillary tube bonded into the laminate, with a defined bend radius and a mechanical strain relief on the free-fibre side.
• Loss of coupling after thermal cycling. The coating–matrix bond degrades after hundreds of thermal cycles between service extremes (e.g. −40 °C to +80 °C for aerospace). The fibre still transmits light but its strain reading drifts. Remedy: coating selection validated against the specific cycle profile; periodic re-reference against surface gauges in regions where this is possible.
• Resin-rich pocket around the fibre. During cure, resin flows preferentially around the fibre and creates a locally weak region. Not a sensor failure directly, but a structural weakness introduced because of the sensor. Remedy: parallel fibre routing and validation of flow behaviour on sacrificial parts before full-scale production.
• Interrogator drift. The reference wavelength standard inside the interrogator drifts with temperature and age; sustained operation outside the specified ambient range produces strain offsets up to 10 µε per year. Remedy: climate-controlled interrogator housing and annual recalibration.
• Connector contamination. A single fingerprint on a connector endface attenuates the return signal by several dB. Remedy: inspection protocol with a fibre scope at every reconnection, and sealed connectors in the operational environment.

8. Recommendations for the First Deployment

For teams installing composite DFOS sensors for the first time, four practical recommendations stand out:

1. Validate on a sacrificial part before the production run. The first laminate instrumented with a sensor fibre will reveal all the practical issues — ingress protection, resin flow, cure cycle effects. A representative coupon or sub-element should be instrumented, cured, and sectioned to inspect the embedded fibre and its surroundings.
2. Keep the fibre routing simple. Straight runs parallel to the main reinforcement direction fail less often, read more accurately, and are easier to interpret. Complex geometries can wait for the second installation.
3. Record the installation. Fibre position, splice locations, interrogator channel assignment and calibration references all go into a monitoring passport that lives with the structure. Without this, the data from year five cannot be compared to the data from year one.
4. Decide the data pathway early. Raw interrogator output is high-bandwidth and not immediately interpretable. Decide at the start which features (peak strain, event rate, threshold crossings) are recorded for long-term archival, and which are only kept as rolling windows. A plan at commissioning is cheaper than a schema migration five years later.

9. Conclusion

Composite DFOS sensors transform a passive composite structure into one that continuously reports its own strain and temperature state from the inside. The technology works — a decade of deployments in wind energy, aerospace and civil infrastructure confirms this. What determines success is not the physics of the fibre or the interrogator but a short list of integration decisions: polyimide coating over acrylate, parallel routing with the reinforcement, dedicated ingress capillaries with a defined bend radius, in-situ calibration against surface instrumentation, and a thermal compensation strategy chosen at the design stage.

Teams who treat the sensor fibre as part of the structural design — not as an instrumentation afterthought bolted on at the end — get sensors that still deliver useful data a decade after they were embedded. Teams who treat it as a cable to be routed around the composite layup discover, usually at the first thermal cycle, that composite sensing is not forgiving of shortcuts.

IOAS supports industrial teams in composite instrumentation, from fibre and interrogator selection to in-situ calibration and long-term data interpretation. For enquiries: director@ioas.pro.

References

[1] Her, S.-C.; Huang, C.-Y. (2011): Effect of Coating on the Strain Transfer of Optical Fiber Sensors. Sensors 11(7), 6926–6941. DOI: 10.3390/s110706926.

[2] Güemes, A.; Fernández-López, A.; Soller, B. (2010): Optical Fiber Distributed Sensing — Physical Principles and Applications. Structural Health Monitoring 9(3), 233–245. DOI: 10.1177/1475921710365263.

[3] Kinet, D.; Mégret, P.; Goossen, K. W.; Qiu, L.; Heider, D.; Caucheteur, C. (2014): Fiber Bragg Grating Sensors toward Structural Health Monitoring in Composite Materials: Challenges and Solutions. Sensors 14(4), 7394–7419. DOI: 10.3390/s140407394.

[4] Luyckx, G.; Voet, E.; Lammens, N.; Degrieck, J. (2011): Strain Measurements of Composite Laminates with Embedded Fibre Bragg Gratings: Criticism and Opportunities for Research. Sensors 11(1), 384–408. DOI: 10.3390/s110100384.

[5] Di Sante, R. (2015): Fibre Optic Sensors for Structural Health Monitoring of Aircraft Composite Structures: Recent Advances and Applications. Sensors 15(8), 18666–18713. DOI: 10.3390/s150818666.

[6] Lecompagnon, J.; Grosse, C. U. (2020): Distributed Fibre Optic Sensing for Monitoring of Composite Structures — Review and Case Study. Sensors 20(14), 3832. DOI: 10.3390/s20143832.

[7] Chandrasekaran, S. (2019): Structural Health Monitoring with Application to Offshore Structures. World Scientific. (Chapters on wind-turbine blade monitoring).

[8] Fernando, G. F.; Hameed, A.; Winter, D.; Tetlow, J.; Leng, J.; Barnes, R.; Mays, G.; Kister, G. (2003): Structural Integrity Monitoring of Concrete Structures via Optical Fiber Sensors: Sensor Protection Systems. Structural Health Monitoring 2(2), 123–135. DOI: 10.1177/1475921703002002004.

[9] Rajan, G.; Prusty, B. G. (2016): Structural Health Monitoring of Composite Structures Using Fiber Optic Methods. CRC Press.

[10] Minakuchi, S.; Takeda, N. (2013): Recent Advancement in Optical Fiber Sensing for Aerospace Composite Structures. Photonic Sensors 3(4), 345–354. DOI: 10.1007/s13320-013-0133-4.