(193b) A Series of Small Changes

A 6,000-gallon, stirred tank batch reactor was equipped with a one-inch, piston-type sample valve at the lower margin of the straight sidewall, just above the joint with the bottom dished head. When closed, the piston in the sample valve sits flush with the inside wall of the vessel. This type of valve is necessary to prevent plugging of the valve when high melt point products are produced in the vessel.

A one-inch diameter, nine-inches long pipe nipple was threaded into the sample valve outlet, to extend the sample discharge point to the chest level of the operator obtaining the sample. When producing high melt-point products, this extension pipe would often plug with product due to the residual material on the pipe walls solidifying as it cooled. When this occurred, the operator would unscrew the pipe nipple from the valve, and either rod it out or replace it with a new pipe.

During a risk assessment of the vessel, it was recommended that all hand valves on the reactor be fitted with position indicators. These position indicators would be interlocked to hazardous materials charging valves on the reactor to prevent charging flammable and/or toxic materials to the vessel whenever a hand valve was inadvertently left open. For example, if the position indicator on the atmospheric vent valve indicated the valve was open, the automated charge valves for flammable and/or toxic materials would not open. This effort included installing a position indicator on the sample valve. However, because a position indicator was difficult to install on the piston-type sample valve, it was decided to install a quarter-turn ball valve on the end of the nine-inch sample valve extension pipe. This ball valve would be equipped with a position indicator, such that even if the sample valve itself were left open, the operator could be assured that the ball valve was closed before flammable/toxic materials were permitted to the charged. This modification was reviewed and approved using the facility's Management of Change system.

Soon after installing the ball valve and position indicator, the operators realized that they could no longer remove the extension nipple, as the position indicator on the ball valve would first have to be removed. Episodes of sample line plugging became more frequent due to the addition of the ball valve on the end of the sample extension. Furthermore, plugged sample lines were more difficult to clear, because they could no longer be removed for cleaning. Operators resorted to applying steam, using a hand-held steam hose, to the sample assembly in order to melt the material plugging the line.

A department manager observed an employee unplugging a sample valve using a steam hose, and decided that steam tracing the valve and extension piping would be preferable to using a hand-held steam line. A work order was written, and ¼? steam trace tubing, supplied with 150 degree C steam, was installed on the sample valve and extension pipe. The installation of steam tracing was not reviewed using the Management of Change system.

Within a month of the steam tracing installation, an operator opened the ball valve on the end of the sampling line, and witnessed a small, pressurized discharge of vapor. He then opened the sample valve, drawing a sample into an 8-oz. glass sample jar. Two to three seconds after flow from the sample valve began, a six-inch diameter fireball erupted from the end of the sample extension pipe. The flash fire lasted less than one second, and diminished to a small (1? long) flame. The operator blew out the flame and closed the sample valve.

The ensuing investigation revealed that operators routinely charge a solution of 20 % sodium chlorite in water through the sample valve. 60 minutes before the fire, sodium chlorite had been charged through the sample valve. After the charge, both the sample valve and ball valve had been closed to allow for charging of a hazardous material. The investigation team theorized that residual sodium chlorite solution, trapped between the sample valve and ball valve, was heated by the 150-degree C steam tracing, boiling/evaporating off the water and leaving behind concentrated sodium chlorite. When the operator opened the ball valve, the observed discharge of vapor was the water vapor from the chlorite solution. When the sample valve was subsequently opened, the organic product contacted the concentrated chlorite and ignited, as foreseen by the warning on the sodium chlorite MSDS.

The conditions that permitted this incident to occur developed over a one to two-year period. A perfectly safe sampling arrangement was modified, with the ?blessing? of an approved MOC, to make the reactor system safer. The MOC reviewers, however, did not realize the modification would increase the frequency of sample line plugging, and that installation of the position indicator would make clearing a plugged sample line more difficult.

The hazard created by the first modification (clearing plugged lines using steam) was noted by an alert department manager. The facility had experienced a number of thermal burns, sensitizing the manager to potential burn hazards. The manager asked a subordinate to write a work order to install steam tracing on the sample valve. Neither the manager nor the subordinate thought the change was ?worthy? of an MOC. The maintenance mechanics that installed the tracing had been trained to identify work orders that should have an MOC, but did not flag this work order as requiring one. While it is easy to include this oversight in the causes of the incident, the four MOC approvers interviewed as part of the incident investigation indicated they did not believe they would have realized the hazards of the steam tracing in combination with sodium chlorite charging.

The incident investigation team initially considered calling for a more robust review of MOCs. However, they also noted that the hazards of concentrated sodium chlorite were not well known in the facility, and in fact were not noted in the chemical hazards section of the facility's procedures. Further, an incident had occurred in the facility four years prior involving sodium chlorite residue igniting a wooden pallet. This incident, and the hazards of concentrated sodium chlorite, had not been communicated to the employees as the investigation recommended. The investigation team felt that had an MOC been written for this change, and had the hazards of sodium chlorite been well known in the facility, the potential for this incident to occur would have been recognized.

The final corrective actions from the investigation included:

Water-flushing lines and pumps used for sodium chlorite charging.

Water-flushing lines and pumps used for hydrogen peroxide charging (a similar oxidizer to sodium chlorite).

Investigate practices of charging materials, including:

1. Should the Manufacturing Packet specify what method/equipment/nozzle to use for each charge?

2. Should the sample valve be used for charging?

3. Are chemical incompatibilities considered when charging multiple raw materials through the same nozzle/lines?

4. Is it acceptable to charge materials to the liquid phase of a product?

Conduct employee training on the hazards of sodium chlorite.