Crack-propagation gauges (CPGs) are series-parallel conductive tracks etched into a polymer substrate, bonded across a surface-breaking fatigue crack so each track snaps as the crack extends, giving a step-wise resistance rise that maps directly to crack length [S1][S2]. They are specified after a visual crack has been identified on steel girders, eyebars, truss members, or weld toes, where the goal is to confirm growth and quantify rate, not to find the crack in the first place [S1][S2].
Data sets commonly stretch over weeks to years because the application is permanent structural health monitoring, with milliohm-class ohmmeters or standard DAQ voltage channels serving as the readout [S1][S2]. Vendor options include the HBM RDS pattern and Micro-Measurements CPGs read through QuantumX, MGCplus, or Gantner Q.series A105 front-ends [S5][S7].
Sensor Construction and Operating Principle
A CPG is a chain of narrow conductive tracks photo-etched onto a thin, non-conductive polymer carrier, then adhesively bonded to the specimen so the track array sits across the expected crack-tip line [S1][S2]. As the fatigue crack advances, each track it crosses is severed; the number of severed tracks drives a discrete jump in gauge resistance that the readout converts to crack-extension length [S1][S2].
Geometry is matched to the suspected crack origin: rectangular patterns are used on plate and girder surfaces, while circular or semicircular patterns wrap bolt-hole and eyebar-bore features where cracks initiate radially [S1][S2]. Bonding quality is critical, because delamination lets the substrate stretch without track fracture and silently under-reads growth; FHWA guidance calls out surface preparation as a prerequisite to track-fracture behaviour [S2].
Readout Hardware and Data Pipeline
Reading a CPG is intentionally low-end: a milliohm-sensitive ohmmeter resolves the resistance step directly, while a standard DAQ voltage channel is the more common route on permanent bridge and tunnel installations [S1][S2]. The HBM portfolio explicitly lists the QuantumX and MGCplus DAQ families for the RDS crack-propagation gauge, and Gantner's knowledge base shows a Micro-Measurements CPG wired to a Q.series A105 module for the same application [S5][S7].
Because the output is a step waveform tied to discrete track breaks, data processing is minimal, and interpretation reduces to counting severed tracks against a calibrated track-pitch table [S1][S2]. Sampling cadence is set by anticipated crack-growth rate rather than by sensor bandwidth, with deployments that routinely log for years on a single bonded gauge [S1][S2]. The same kind of low-speed, high-stability signal chain is what makes condition monitoring systems economical for bridges, where the alternative is repeated rope-access visual inspection.
CPG vs DIC, AE, Compliance, and DCPD
Where a project needs spatial resolution, full-field strain, or detection of cracks before they are visible, the literature now groups the four principal non-CPG techniques: digital image correlation (DIC), acoustic emission (AE), compliance-based methods, and direct current potential drop (DCPD) [S4]. A 2025 review reports DIC in-plane displacement accuracy of roughly 1/100,000 of the field of view, strain resolution near 100 microstrain, and around 20 microstrain achievable with careful speckle and calibration; an aircraft-wing prototype ran a 1440x1080-pixel IMX296 camera at one frame every 15 s across a 25-hour cyclic test [S4].
The trade-off is concrete: DIC and DCPD deliver high spatial or electrical resolution but need camera rigs, current injection, and controlled lab conditions; AE captures initiation events across a structure but cannot easily quantify crack length; compliance methods derive length from load-displacement slope and need a load frame [S4]. CPGs sit at the opposite end of that matrix, cheap, passive, and bond-and-forget, but limited to a known, already-visible crack path and to track-by-track resolution rather than continuous length. On a remote bridge or offshore wind support structure, that combination is exactly why a vibration condition monitoring scheme will still be paired with a bonded CPG over any crack that visual inspection has flagged.
Welded Joints and Thick-Section NDT Campaigns
Fatigue cracks in thick welded specimens are the canonical hard case, because weld toes and weld metal mask the crack mouth and accelerate growth under cyclic load [S3]. A 2024 BAM-led study combined strain gauges, thermography, ultrasound, and BAM's crack-luminescence method in a synchronised data-fusion stack to follow initiation, propagation, and final fracture on the same specimen, with the explicit goal of feeding inspection programs on offshore wind support structures [S3].
CPGs were not the headline sensor in that campaign, but they remain the cheapest way to bracket a known weld-toe crack between inspection cycles, and the multi-method logic is the same: one technique for detection, another for quantification, and a data layer that timestamps both. A 2025 paper on real-time monitoring and propagation prediction goes further, building a system that covers crack identification, propagation visualisation, and crack-width quantification in one pipeline, which is the direction permanent monitoring is moving [S6].
Where CPGs Fit and Where They Do Not
CPGs are for the case where a crack is already known to exist, the operator needs to know whether it is dormant or growing, and the budget does not allow a camera rig, a function generator for DCPD, or an AE sensor network on every member [S1][S2][S4]. They are not for finding new cracks, not for sub-track-resolution length measurements, not for through-thickness or sub-surface cracks, and not for high-cycle components where growth between track pitches can skip events.
For new-build spec writing, the practical rule is: use CPGs to instrument flagged details on bridges, tunnels, eyebars, and welded girders where inspection access is expensive; use DIC, AE, DCPD, or compliance when the crack has not been found yet, or when the test article is in a lab under controlled cyclic loading [S3][S4]. The two strategies are complementary, not competing, which is why a layered condition monitoring system will typically include both bonded resistive gauges on flagged members and a wider sensor net across the structure.
Specification Checklist for a CPG Installation
Selection starts with geometry: rectangular pattern for plate and girder faces, circular or semicircular pattern for bolt holes and eyebar bores, and a track pitch finer than the smallest crack extension the asset owner needs to resolve [S1][S2]. Readout is chosen next, with a milliommeter acceptable for spot checks and a QuantumX, MGCplus, or Q.series A105 channel the standard route for permanent data logging [S5][S7].
Surface preparation, adhesive selection, and gauge orientation relative to the expected crack plane are the three installation variables that decide whether a CPG reports true growth or silent under-reading [S2]. Documentation should record track count, track pitch, bonded area, adhesive type, readout channel ID, and the inspection cycle that flagged the original crack, so future readings can be diffed against a known baseline [S1][S2]. On multi-asset sites, this is the same documentation discipline used in flow meter and pressure transmitter datasheets, where the calibration record is what makes the data defensible years later.
The next signal to watch is the move from passive CPG-only monitoring to fused stacks that pair bonded resistive gauges with DIC, AE, or DCPD under a single DAQ timestamp, as already prototyped for aircraft wings and offshore wind monopiles [S3][S4][S6]. Trackable follow-ons include the BAM crack-luminescence method moving from lab specimens to in-service welded structures, and the integration of real-time crack-width quantification with structural health-monitoring platforms on bridges and wind turbine supports [S3][S6].
For related coverage, see Film vs Digital Radiography: Per-Exposure Cost and Chemistry Stack.