Rooftop solar fire investigation, DC arc fault originating at a module junction box

A peer reviewed engineering investigation, reported in 2024, examined a fire at a rooftop photovoltaic installation in southern China. The investigators combined on site inspection of the fire scene, electroluminescence (EL) imaging of recovered modules and laboratory replication of suspected failure modes. They identified the ignition source as a single module junction box and attributed the fault mechanism to a DC arc at an internal contact within that box. No information on the extent of damage or the emergency response is recorded in the material reviewed. The case is included here because the mechanisms described, junction box contact degradation, mechanical stress introduced during installation and DC arc escalation, are also relevant to ground mount and utility scale systems using similar module construction.

As stated by the investigators: DC arc ignition originated from stress induced poor contact within a module junction box. EL testing of recovered modules across the site indicated a very high defect rate, with approximately 99% of inspected modules showing defects, mainly cell cracks. Inherent thermal failure modes, bypass diode heating and reverse current overload, were excluded as causes because laboratory replication of these mechanisms did not reproduce the observed ignition behaviour. Correlation analysis across the field tested modules associated the ignition with installation quality factors, specifically the gap between modules, cable tension and crimping quality. The investigators considered that mechanical tension arising from the installation method caused minute loosening of metallic contacts within the junction box, which then sustained a DC arc until ignition.

1. Treat installation workmanship as a primary determinant of long term fire risk, since a design that performs safely in laboratory conditions may behave differently once mechanical stresses are introduced in the field.
2. Recognise module junction boxes as a recurring weak point, combining high DC voltage, mechanical stress at the cable entry and limited internal inspectability.
3. Verify cable tension and crimp quality during installation by measurement and sampling rather than visual inspection alone.
4. Use EL imaging at commissioning and at periodic intervals to identify defective modules, since a site wide defect rate approaching the level reported here would not be detected by visual inspection.
5. When investigating a photovoltaic fire, use laboratory replication to test and exclude thermal failure modes, which strengthens conclusions about arc fault causation.
6. Combine DC side protection, such as arc fault detection and rapid shutdown, with installation quality assurance, as the two mitigations address different stages of the same failure path.
7. For asset owners, include a contractual right to EL imaging at handover and at intervals during the asset life, which is a low cost check against systemic installation defects that may otherwise emerge years later.

sciencedirect.com/science/article/abs/pii/S0038092X26001088