Petroleum refineries and chemical plants periodically need to clean storage tanks to check for corrosion, to allow the tank to be repurposed, or for other maintenance work. Part of the process for cleaning a tank includes gas freeing – removing flammable or toxic vapors from the tank – to provide a nonhazardous atmosphere in the tank for the cleaning crew.
A petroleum refinery had a large storage tank that had been out of service for several years that they wanted to repurpose. This tank was protected by the plant’s vapor recovery system that used natural gas (composed mostly of methane) to fill the headspace of their tanks to prevent air ingress into the tanks. They needed to clean the tank prior to putting it into service. As part of the cleaning process, they arranged for a third party to gas-free the tank.
The gas-freeing company planned to do the gas freeing by opening a ground-level manway to allow air inflow while using a thermal oxidizer to draw air and vapors from the tank through a second ground-level manway. The oxidizer included monitors to measure the concentration of methane in the discharge gas stream. The local authority limited the air flow rate through the oxidizer, so the gas-freeing plan needed 48 hours to provide the number of air exchanges required to dilute the methane to a safe concentration. However, after 7 hours, the oxidizer monitors measured the methane concentration below the target level, so the operator shut down the gas-freeing operation. The day after the operation was suspended, gas in the tank ignited and the deflagration destroyed the tank. ESi was retained to investigate the incident to determine how this occurred.
ESi Approach
The thermal oxidizer measured both the inlet methane concentration and the total amount of energy produced by the oxidizer’s burner. This allowed an estimate of the total amount of gas from the tank that was burned. ESi’s analysis showed that only a fraction of the total methane in the tank was burned by the oxidizer. Further analysis using computational fluid dynamics (CFD) showed that the locations of the air inlet and the gas discharge at ground level resulted in stratified layers in the tank headspace with a methane layer at the top of the tank and air short-circuiting through the bottom of the tank rather than completely mixing with the gases in the tank headspace. The result was that only about 20% of the tank headspace was mixed with air and this was the portion the oxidizer detected as having no methane remaining after 7 hours.
The Outcome
The API standard RP 2016 provides guidelines for vapor and gas freeing of tanks. They recommend using 5 air changes per hour when gas freeing a tank and recommend positioning inlets and outlets to take advantage of density differences between the gas or vapor in the tank and air. The gas-freeing plan used in this incident did not follow the API guidelines in the locations of the inlet and outlet, and the restriction on maximum air flow rate prevented them from flowing air at 5 changes per hour. The air flow was not capable of breaking the stratified layers. The investigation showed that ESi’s client was not responsible for the deflagration because they were not involved in planning or completing the gas-freeing operation.
ESi expanded the analysis for this incident to cover additional scenarios and included heuristic knowledge and some published correlations for analyzing whether a specific set of conditions will be likely to successfully gas-free a tank when the air flow rate that can be used is less than the API recommended rate. This analysis was presented at the 2017 AIChE Global Conference on Chemical Process Safety, and can be downloaded from https://www.researchgate.net/publication/315771757_Stratification_During_Vapor_or_Gas_Freeing_of_Storage_Tanks.
Summary
ESi was retained to investigate a deflagration that occurred shortly after a large storage tank had been gas-freed. ESi found that the gas-freeing plan did not consider the density of the gas in the tank (methane) and so the air inlet and outlet were not properly located. The gas freeing operation was suspended when the readings in the air discharge from the tank showed a low methane concentration. However, the duration of the gas freeing operation was significantly too short to completely remove the methane from the tank even if the air inlet and outlet were properly located. ESi estimated that approximately 20% of the tank was gas freed and the remainder contained methane. This subsequently ignited, resulting in a deflagration and failure of the tank.
