A gas leak ignited around 2 pm on the low-density polyethylene unit at a chemical plant. At 12:45 pm, the unit was activated by the low suction pressure on the hyper compressor. The pre-compressor started up, and the technician observed that the gas transition in the average pressure return (APR) was occurring normally. At 1:50 pm, the hyper compressor was re-started, but the drum disc broke during the pressure rise phase (with the reactor at 500 bar); 15 to 20 seconds after the break, gas ignited upon exiting the chimney. Emergency services were called and plant personnel confined in the control room. The operator activated the facility’s internal emergency plan. The fire was brought under control by onsite resources around 2:30 pm. Once the ethylene supply line was shut, the operator decompressed the unit and allowed the gas to burn. Next, the machinery was cooled in order to avoid any resumption; however, polymer combustion due to the decompression was still ongoing. A portion of the APR had remained at 40 bar; a water curtain was installed to decompress this portion safely. No environmental pollution could be observed. Onsite traffic was reopened around 3 pm, except for the ethylene polymerisation unit. The extinction water was collected inside a retention basin. A neighbouring firm had to halt its activity on the part adjacent to the hazardous zone. All risk of explosion was ruled out at 5:20 pm; the unit was decompressed and isolated by 6 pm, then blanketed with nitrogen. No element sensitive to plant operations had been affected.

The pressure surge in the APR circuit was tied to polymer movement from within. This scenario, included in the safety report, was confirmed upon opening the circuits (booster segmentation problem). This step caused a loss of flow rate on the booster’s first stage, used in low-pressure recycling (LPR). The technician tried to compensate for this loss by modifying the booster operating point, but did not anticipate polymer movement in the APR circuit since no alarm had been triggered regarding the LPR ratio. In reality, the control system calculates the LPR ratio (LPR flow rate/production flow rate) in order to bleed excess gas from the average pressure separator with a low flow rate alarm. 2 ratios based on the same ethylene intake flow meter actually existed. Following its inspection, the flow meter was found to be offset.

Before re-starting the unit, the operator planned on changing the segmentation booster material to reduce flow rate loss due to internal recycling. The protocol was modified: in case of low flow rate on the booster, polymerisation was stopped and unit operations assessed for eventual shutdown if the booster required attention. Installing a backup ethylene intake flow meter on the unit would be studied over the medium term.