Four rear-mounted TEGs use differences in thermoelectric voltage to estimate hotspot positions in two dimensions
Outdoor tests recorded a 7.7°C reduction in local point temperature and a 5.4°C reduction in average temperature across the monitored area
The harvested energy supported approximately 45 wireless transmission cycles during daytime testing
Hotspots are a major concern in PV module reliability. They are localized overheating events that can occur due to partial shading, electrical mismatch, surface contamination, cell microcracks, soldering defects, encapsulant degradation, etc. Detecting hotspots can be challenging in distributed PV, where modules are spread across rooftops or sites that are not easy to access.
Researchers from Southeast University in China and the University of Technology Sydney have developed a PV thermoelectric generator (PV-TEG) that can detect the location of hotspots. It can also reduce localized heat buildup and convert heat into electricity to power wireless monitoring of these modules.
The solution is based on the Seebeck effect, in which a thermoelectric generator produces a voltage when a temperature difference exists between its hot and cold sides. The researchers placed a 2 × 2 array of TEGs on the module’s rear side, with the hot sides connected to the module backsheet and the cold sides to an aluminum heat sink. Each TEG consisted of 288 p-n thermocouple pairs.
When a hotspot develops, it creates a non-uniform temperature distribution on the module’s rear side. As the four TEGs are positioned at different locations, they experience different temperature differences and produce different voltage responses. The system compares these responses and estimates the hotspot’s position in two dimensions. The same arrangement also provides a heat-transfer path to dissipate heat from the module.
The team first studied the relationship between hotspot position, TEG voltage response, and local cooling through a multiphysics model. They evaluated nine hotspot positions arranged in a 3 × 3 grid in a 500 × 500 mm PV region. The different voltage patterns from TEGs were used to map voltage responses to hotspot coordinates. The simulated local temperature reduction ranged from 8.15°C to 12.40°C across nine positions, averaging 9.93°C.
The team tested it outdoors at a rooftop installation at a telecommunications facility in Taizhou, China. They used black tape on the module’s front surface to create a repeatable local heating anomaly. The tape increased solar absorption and increased local temperature. Based on the voltage responses from TEGs, the system reconstructed the position of the anomaly with a localization error of about 9.93 mm across three tested positions.
Infrared measurements showed that the TEG-integrated region had a local point temperature of 41.7°C, compared with 49.4°C in the reference PV region without TEG integration. This represents a reduction of 7.7°C. Across a 500 × 500 mm monitored area, the average temperature was 45.2°C for the PV-TEG region and 50.6°C for the reference region, a difference of 5.4°C.
It is important to note that this controlled test does not reproduce the electrical and thermal behavior of an actual PV hotspot caused by reverse-bias operation, electrical mismatch, interconnection defects, etc. Their heat generation and temperature distribution patterns may differ from those produced by a tape-induced anomaly.
The study also included operating the monitoring system using electricity generated by the TEG array. The harvested energy was regulated and stored in a 0.47 F, 5 V supercapacitor, which powers the data acquisition circuits, controller, and wireless communication module. The system was designed to activate and return to charging when the supercapacitor voltage reached 4.0 V and 2.8 V, respectively. During outdoor testing from 08:00 to 18:00, the TEGs repeatedly charged the capacitor within this voltage range, supporting approximately 45 wireless transmission cycles. The charging rate varies with the temperature difference across the TEGs, which slowed down during morning and evening hours.
The researchers describe the system as a local sensing and thermal-management node rather than a solution covering an entire PV module. Larger installations would require multiple nodes, with further work needed on their spacing, calibration, and interaction. This study is published in Applied Thermal Engineering.
