A petroleum refinery experienced a catastrophic rupture at one bank of three heat exchangers in a catalytic reformer/naphtha hydrotreater unit because of high temperature hydrogen attack (HTHA). Hydrogen and naphtha at more than 500F were released from the ruptured heat exchanger and ignited, causing an explosion and an intense fire burned for more than three hours.
The rupture fatally injured seven employees working in the immediate vicinity of heat exchanger at the time of the incident. The workers were in the final stages of a start-up activity to put a parallel bank of three heat exchangers back in service following cleaning. Such start-up activities had resulted in frequent leaks and occasional fires in the past and should have been considered as hazardous and nonroutine.
The damage of the specific heat exchanger was the result of its carbon steel being severly weakened by HTHA, a mechanism that results in fissures and cracking which occurs when the material is exposed to hydrogen at high temperatures and pressures, severely degarding the mechanical properties of the steel. API Recommended Practice 941 (Steels for Hydrogen Service at Elevated Temperatures and pressures in Petroleum Refnieries and Petrochemical Plants) provides Nelson curves to predict the occurrence of HTHA in various materials of construction. These curves are predicated on past equipment failure incidents and are plotted based upon self-reported process conditions. An investigative computer reconstruction of the ruptured heat exchanger estimated that exchanger was operating in a safe region of the Nelson curves where HTHA could not occur. API Recommended Practice 581 (Risk-Based Inspection Technology) allows users to calculate a damage factor to determine HTHA susceptibility of various materials of construction, rather than requiring users to actually verify operating conditions when determining applicable damage mechanisms.
- Heating Equipment
- Heat Exchanger
Carbon steel Nelson curve methodology cannot be depended on to prevent HTHA equipment failures and cannot be reliably used to predict the occurrence of HTHA equipment damage. Revisions to recommended practices should be considered regarding the use of carbon steel in HTHA-susceptible service and the verification of actual operating conditions.
Given the difficulty of inspecting for HTHA because the damage might not be detected, inherently safer design is a better approach to prevent HTHA.
Process hazards analysis (PHA) and damage mechanism hazard reviews (DMHRs) need to carefully consider all assumptions, periodically if necessary, to ensure that hazard identification, safeguards and control of hazards to prevent equipment failure are effective.
Effective programs need to be in place to manage and provide oversight for hazardous nonroutine work.