During operation of a succinic acid plant, hydrogen leaked from a mounting joint on a safety valve at the upper part of a reactor, which generated a hydrogen flame. Prior to the incident, the safety valve was removed and reattached during an inspection at a turnaround shutdown. An incorrectly sized, smaller gasket was installed on the joint, and the tightening force on the bolts was inadequate. Therefore, a gap was generated as time went by and un-reacted hydrogen leaked.
In the case of many leak tests after construction, a leak is checked by a soap test after pressurizing piping and facilities for the test. (A soap test is conducted by pouring soap suds at the place to be checked (mainly a joint part) after pressurizing. If bubbles are found, that is the evidence of a leak.) The gasket and bolt torque applied to a given piece of equipment is specified according to physical properties of the flowing liquid, temperature, long-term use, etc., yet, passing of a leak test is not guaranteed for long-term use. Moreover, regarding a bolt tightness check, the tightness is usually checked by striking a bolt lightly with a small hammer (called a hammering test). In this particular incident, a torque measurement test was never performed. Also, it is difficult to find a defect from a general examination after completion of construction.
A hydrogenation reaction of maleic anhydride for manufacturing succinic acid was finished. While an operator went to the site for confirmation, he found a flame jetting from a flange that joined a safety valve to a reactor.
On May 9th during a turnaround shutdown, a safety valve of a reactor was removed for inspection.
On May 16th the safety valve was attached again with a new gasket.
On May 25th an air tightness test at the joint was conducted.
On June 8th at 12:15 the manufacture of succinic acid started. Raw materials were automatically charged to the reactor, and a reaction started.
At 15:05 the feed rate of hydrogen fell, and the pressure rose. A low-temperature alarm sounded simultaneously.
At 15:10 an operator went to the site to open a steam valve. He found a pale flame from the upper part of the reactor.
A hose was immediately connected to the water service faucet and the fire was extinguished by spraying water.
The operation of the reactor was stopped and nitrogen gas was introduced to the reactor. Occurrence of the fire was reported to the office.
An incorrectly sized gasket, which was smaller than the original gasket was used in a joint of a safety valve at the time of a turnaround shutdown. This caused an unbalanced force across the valve and an unequal tightness of the connecting bolts. Due to the small size of the gasket and looseness of bolts, the weight of the piping was applied locally to the joint, and the joint inclined gradually. Eventually un-reacted hydrogen blew out from a gap due to the inclination, and it was ignited by static electricity sparks.
This incident illustrates several problems. During the construction process, procedures for managing component parts and for securely tightening bolts were inadequate. After construction was completed, the construction check process and the leak test were not adequate in identifying and correcting the problems and avoiding the incident.
Water was sprayed using a faucet.
Operation of the plant was stopped.
Nitrogen gas was introduced to the reaction system.
Construction errors are difficult to detect once construction is complete. It is important to develop and use a systematic oversight process for minimizing construction errors during the construction process.
Thorough control of component parts during the construction process is required.
Bolts should be tightened equally and fully.
A new support for distributing the weight of piping is installed.
Thoroughness of checks after construction is going to be initiated.
Additional discussion about working with reactive metal-hydride materials in the laboratory can be found in the Lessons Learned Corner on this website and in the Hydrogen Safety Best Practices Manual.