Simulation tests could not replicate failure at similar conditions, so an ambiguity remained. The real cause remained the assembly pipe between reactors 4 and 6 since no study and no test were launched before its installation. The problem on reactor 5 should have been studied before running the unit again. In fact, the reactors had been sprayed with water to dilute the cyclohexane leaks to limit flammability, but the nitrates of the water dripped into the steel pipe and caused to corrosion