Why did BP Texas City explode?
At BP’s Texas City refinery on March 23, 2005, a distillation tower overfilled during startup. Its level indication misled the operators, and a separate high-level alarm failed to activate. When liquid reached the pressure relief system, its valves opened and discharged into a blowdown drum and atmospheric stack that could not contain it. Released hydrocarbons formed a flammable vapor cloud that ignited. Fifteen people died, all in or near office trailers close to the blowdown system. CSB, pp. 17, 21–25.
Preventing overfilling, safely disposing of an emergency release and keeping occupied buildings away from process hazards each offered a chance to protect people. The investigation found failures in all three safeguards and in the process-safety management behind them.
Follow the complete failure chain
The companion film traces the liquid from the tower through the relief and blowdown systems, then explains why the investigation looked beyond the control room.
How could the reading fall while the tower filled?
The raffinate splitter separated a hydrocarbon mixture into lighter and heavier components. During the restart, liquid entered the tower for more than three hours without any being removed. The control display did not adequately show this imbalance, leaving operators with a misleading picture of the process. CSB, pp. 17, 21, 23.
The displayed percentage represented the transmitter’s measurement span near the bottom of the tower. It did not measure how much of the entire tower was full, so a reading below 100 percent could not establish that most of the tower remained empty. The CSB reconstructed a rising liquid level while the indicated level fell. CSB, p. 56 and footnote 42; pp. 81–82.
The redundant high-level alarm did not activate. Its silence reinforced the board operator’s belief that the declining indication was accurate. A dirty, unreadable sight glass prevented visual confirmation, leaving the operator without a functioning, accurate measure of tower level that day. CSB, pp. 81–82.
A different high-level alarm, associated with the transmitter, was active during startup and had already been acknowledged. The failed redundant alarm was one specific loss of protection; the plant did give other warnings. CSB, p. 82, footnotes 77–78.
The valves opened. Where did the liquid go?
As the tower filled, liquid entered the overhead pipe. Pressure rose at the relief valves below, and they opened. Opening a relief valve can reduce pressure in one part of a system, but the discharged material still needs a safe destination.
At Texas City, it entered a blowdown drum connected to a stack open to the atmosphere. The drum was too small for the incoming liquid. The drum and stack overfilled, releasing flammable material outside the process equipment. That material fell to the ground and formed the vapor cloud that ignited. CSB, pp. 21, 24.
The system was not connected to a flare. A properly designed flare system separates and contains liquid before burning the remaining gases. The CSB found the atmospheric blowdown arrangement unsafe and documented earlier warnings and missed opportunities to replace it. CSB, pp. 21, 24 and footnote 10.
Why the location of the trailers mattered
The explosion’s consequences also depended on where people were working. Occupied trailers stood too close to a unit handling highly hazardous material. All fifteen fatalities occurred in or around those trailers; the report does not say everyone was inside them.
The CSB also found that BP had not effectively applied its pre-startup review policy to remove nonessential personnel from the area during startup. Trailer placement and the presence of people near the unit were therefore part of the safety system the investigation examined. CSB, p. 25.
A low injury rate did not measure this danger
The CSB warned against reducing the explanation to the operators closest to the event. It identified deficiencies in training, staffing, communication, instrumentation, maintenance, supervision and organizational oversight. It also found that cost-cutting, inadequate investment and production pressures impaired process safety. CSB, pp. 19, 23–26.
BP treated a low personal-injury rate as reassurance about safety more broadly. Fewer routine workplace injuries, however, did not establish that the refinery’s safeguards against a major release, fire or explosion were working.
The Baker Panel examined leadership and process-safety management across BP’s five US refineries. It was commissioned to assess corporate oversight and safety systems, rather than reconstruct the Texas City accident or allocate blame. The panel also lacked sufficient information to conclude whether BP intentionally withheld safety resources for budgetary reasons. Baker Panel, printed pp. viii–ix, xii.
What did the investigation call for?
The response extended beyond replacing an instrument. The CSB called for safer siting of occupied trailers, better treatment of atmospheric blowdown hazards, and stronger oversight of major-accident prevention. The Baker Panel recommended clearer process-safety leadership, accountability, expertise and performance monitoring. CSB, pp. 27–28.
The reports document historical findings and recommendations; they do not establish that every recommendation has since been implemented everywhere. Accurate indications, safe relief-system discharge and separation of people from hazards each need to work if the refinery is to contain a failure and protect those nearby.
For another case involving misleading control-room indications, read what the Three Mile Island control room could not see. The physical mechanisms differ, but both investigations show why operators need reliable information about the condition of the equipment.
Primary sources
- US Chemical Safety Board, Investigation Report: Refinery Explosion and Fire, BP Texas City, Report 2005-04-I-TX, March 2007. Incident and findings: printed/PDF pp. 17–28; level indication: p. 56 and pp. 81–82. Read the final report.
- BP US Refineries Independent Safety Review Panel, January 2007. Scope: printed pp. viii–ix (PDF pp. 10–11); corporate process-safety findings and limits: printed p. xii (PDF p. 14). Read the Baker Panel report.
- US Chemical Safety Board, Texas City investigation page and aftermath photograph. The article hero is an authentic photograph. The episode includes clearly illustrative AI refinery scenes, technical graphics and AI narration; those scenes are not recordings of the accident.
