This was a single-run comparison, however, not a validated warning time before ignition, and the study did not establish that particle sensors generally outperform gas detectors. The experiment addressed a monitoring problem in which an abnormal signal may be difficult to recognize before a cable fire becomes conspicuous. Cable trenches are often narrow, elongated and only partly enclosed, which can restrict ventilation and heat dissipation.

Temperature sensing is comparatively resistant to interference, but the study notes that it can lag as a material heats. Smoke detection is aimed mainly at light scattering from dense smoke and large particles associated with later stages of fire, while gases released during initial thermal degradation may be present in very small quantities and become rapidly diluted. Gas measurements can also be affected by ambient temperature, humidity and background impurities.

The researchers therefore examined whether airborne particles associated with PVC heating could provide an additional early-warning signal. Wu and colleagues constructed a straight laboratory trench 6 m long, with an internal width and height of 0.8 m by 0.8 m. Its geometry was informed by a State Grid general design for cable lines, and the flame-retardant wooden enclosure was intended to reproduce the elongated, confined character of a distribution trench.

It did not reproduce all the conditions of an operational installation, including the same enclosure materials, cable occupancy, electrical loading and ventilation. Five laser particle sensors, designated N1 through N5, were mounted vertically at 1 m intervals along the upper central axis. A PTC heating plate supported a PVC coupon measuring 40 by 40 by 10 mm, and a PT100 temperature sensor monitored the heated material.

The sensors detected particles through laser-light scattering interpreted using Mie-scattering theory. Air carrying suspended particles entered an optical measurement chamber, where particles intersected the laser beam and scattered light. Photodetectors captured the weak scattered signal, which underwent filtering, amplification, digitization and conversion into a concentration value reported by the instrument.

This value was treated as an estimated mass concentration rather than a chemically specific measurement. For a fixed optical geometry, the signal depends on both particle abundance and optical properties, including size, shape and refractive index. The sensors had a stated display resolution of 1 microgram per cubic metre and a range extending to 2,000 micrograms per cubic metre, but the resolution was not the same as a measured minimum detectable concentration.

The researchers also used electrochemical sensors for carbon monoxide, methane, ethylene, acetylene and HCl. In a separate qualitative comparison, a standard particle counter and a laser particle sensor were operated under the same thermal and environmental conditions. Both outputs remained near baseline during the first three minutes at approximately 20 °C.

About six minutes after heating began, both instruments increased as the temperature approached 100 °C. The counter enumerated particles of different diameters, whereas the laser sensor produced a continuous mass-concentration estimate, so the researchers did not compare their numerical outputs directly. Their matching trends supported the laser sensor’s responsiveness to a heating-related change in that setup, but did not demonstrate equivalent sensitivity, a statistically significant correlation or a universal detection temperature.

Thermogravimetric analysis was used separately to characterize the PVC sample. In air, with a heating rate of 10 °C per minute, the specimen entered its first main mass-loss stage at approximately 200–350 °C. The derivative mass-loss curve had a peak at 286 °C, and the corresponding mass loss was 43%.

A second stage occurred at approximately 400–500 °C, with a peak at 447 °C and a further loss of 21%. Below 200 °C, the sample lost about 1.6% of its initial mass, taking its thermogravimetric curve from approximately 99.8% to 98.2%. This small loss was consistent with the release of moisture, residual processing compounds or formulation additives, but the experiment did not identify the emitted species, distinguish evaporation from chemical reaction or prove that the released material formed optically detectable particles.

Three experimental temperature bands were used: 100–110 °C, 150–160 °C and 200–210 °C. These were research conditions selected to examine weak low-temperature emissions, an intermediate condition and a range approaching the onset of the major thermogravimetric degradation stage; they were not standard operating limits or prescribed alarm settings. In the end-source configuration, the PVC was positioned below N1 and the gas-sensor unit was also at N1, so transport was expected to proceed mainly along one longitudinal direction.

No sensor response was reported at 100–110 °C. At 150–160 °C, the particle-sensor response onsets were 23.40 minutes for N1, 26.06 for N2, 26.97 for N3, 28.56 for N4 and 33.85 for N5. At 200–210 °C, the reported times were 18.31, 19.67, 20.56, 21.34 and 24.49 minutes, respectively.

Peak particle concentrations did not rise uniformly at every position as the heating band increased. In the end-source runs, the N1 maximum changed from 448 micrograms per cubic metre at 150–160 °C to 444 micrograms per cubic metre at 200–210 °C, while N2 through N5 increased to between 653 and 674 micrograms per cubic metre. The reported traces included an initial rise, a decline and a later increase.

The researchers said this nonmonotonic pattern could reflect changes in emission rate, local mixing, deposition and transport, but the signals could not uniquely identify a transition from release of surface additives to deeper polymer degradation. Those estimates combine emission, transport and instrument response; they were not direct measurements of particle diffusivity or local airflow. A second configuration placed the heated PVC below N3 near the middle of the trench, allowing transport toward both ends, while the gas unit was positioned at N2, 1 m from the source.

No response was reported at 100–110 °C. At 150–160 °C, N3 responded first at 18.95 minutes. The remaining onset times were 20.28 minutes for N4, 20.60 for N5, 21.48 for N2 and 21.50 for N1.

At 200–210 °C, N3 again responded first, at 16.02 minutes. N4 followed at 16.91 minutes, then N2 at 17.50, N5 at 17.73 and N1 at 18.24. The higher band was thus associated with earlier onsets overall in these traces, although the spatial sequence was not identical in every location.

The N3 peak declined from 590 micrograms per cubic metre at 150–160 °C to 363 micrograms per cubic metre at 200–210 °C, showing that a higher temperature did not produce a larger peak at every sampling position. These findings remain exploratory because each combination of temperature and source location was represented by one reported trace. The five sensors sampled different positions within a shared run and were not independent experimental replicates, so standard deviations, confidence intervals and statistical significance could not be estimated from the reported data.

Operational trenches also contain dust and aerosols unrelated to PVC heating, while humidity, contaminated inlets and optical-surface drift can alter particle measurements. The study proposes defining baseline-adjusted concentration and persistence criteria, testing them on separate overheating and nuisance-source trials, and confirming candidate alarms with neighboring particle sensors or gas and temperature channels. It also proposes comparison with intact cable sections at matched jacket temperatures and subsequent loaded-cable testing with separate conductor and surface temperatures.

The 1 m sensor spacing was therefore an experimental configuration rather than a deployment recommendation. Independent repeats, quantitative detection thresholds, interference testing and field validation are needed before the reported response advantage or any safe spacing and warning time can be established.

Subject of Research: Experimental evaluation of laser particle sensing for detecting PVC-related emissions in a simulated 10 kV cable trench

Article Title: Early warning of PVC overheating in a simulated 10 kV cable trench using thermally induced particulate matter

Article References: Wu, Z., Wang, Q., Xiao, Q., Wang, S., Shao, Y., Guo, W., Long, Y., & Huang, H. (2026). Early warning of PVC overheating in a simulated 10 kV cable trench using thermally induced particulate matter. Frontiers in Sensors, 7, Article 1953610. https://doi.org/10.3389/fsens.2026.1953610

Image Credits: AI Generated

DOI: 10.3389/fsens.2026.1953610

Keywords: 10 kV cable trench, PVC overheating, laser particle sensor, thermally induced particulate matter, hydrogen chloride sensing, airborne particles, Mie scattering, distributed sensing, cable insulation, early fire warning, Early, warning