Study on pyrolysis gas generation characteristics of insulating oil under electro-thermal action based on ReaxFF...

Mineral insulating oil serves as the core insulation and cooling medium of power transformers, and its pyrolysis and gas generation behavior directly determine the safe operation of equipment. Conventional dissolved gas analysis (DGA) only targets small-molecule C1/C2 gases produced in the late pyrolysis stage, which inevitably causes obvious delay in early fault identification. In this work, ReaxFF reactive force field molecular dynamics simulations and physical experiments are jointly adopted to build a multi-component insulating oil molecular model consisting of alkanes, cycloalkanes and aromatics. We systematically investigate oil pyrolysis under single thermal stress and coupled electro-thermal stress, with C₃ and C₄ hydrocarbon intermediates as the exclusive core monitoring indicators. Simulation results reveal that higher temperature accelerates overall oil pyrolysis; C1 and C2 dominate final pyrolysis products, while C₃ accumulates massively at moderate temperatures and C4 appears in trace quantities specifically in the medium-temperature window. The superimposed electric field supplies extra reaction energy and facilitates the generation of C3, C4 as well as small-molecule gases. Consistent experimental observations further characterize the unique evolutionary patterns of C3/C4: under thermal faults, C3 and C4 intermediate concentrations follow a distinct rise-then-fall trend with growing temperature; under discharge faults, total gas yield rises with applied voltage, yet C4 content declines synchronously. Notably, significant concentration fluctuations of C3 and C4 can be captured under non-breakdown incipient fault states, long before prominent changes of conventional C1/C2 gases. This study fully demonstrates that continuous monitoring of the characteristic variation rules of C3 and C4 intermediates can effectively reflect the early pyrolysis signatures of insulating oil, offering a novel and reliable basis for early warning of latent transformer faults.