Tremendous High Pressure of SF6 Gas

Some pretty nasty compounds can be formed by the decomposition of SF6. I would use caution when cleaning up.

like msamad said its due to presence of high short circuit currents and flash over voltages which has a value greater then intreruppting capacity of the breaker (ka rating) leading to arcs or flashes which will destroy the breakers along its path.............

It is not even necessary to have fault interruption conditions within the breaker bushing for it to fail catastrophically. Typical gas pressure in an SF6 breaker tank is 85-90 psig. If there is a flaw in the bushing porcelain, or a stress that causes a bond to fail (for example, a person standing on it during maintenance & testing work), the bushing can crack and explode. For these various reasons, ABB sells a kevlar (i.e., "bulletproof vest") blanket that can be wrapped around the breaker bushings when work is being done on it. It's only for use de-energized, not during normal operation.

The utility I work for does not do any breaker switching operations locally at the breaker due to the inherent danger of a bushing failing. We do all of our switching remotely from the control center, and even order our breakers without local close/trip controls. We don't want anyone standing next to the breaker when it operated, in the "line of fire."

You mentioned high pressure - do you have a record of some sort that tells you what the pressure was before failure? Was it above the manufacturer's normal range for that breaker?

As an amendement to the previous posts I would like to add the IMHO most obvious reason for the explosion of a SF6-isolated switch gear: The breaker was not capabale to interrupt the high current due to burnout of the interrupting contact. In such a case the length of the pin of the interrupting contact might have become to short and the overlap of the interrupting contact in respect to the working contact might have become critically low. In such a constellation the high current might not have been completely forced to the interrupting contact but was able to feed an electric arc across the working contact, which is absolutely unwanted and certain cause to have the breaker failing. Depending on grid impedance exposed to the breaker and speed of the other protection units only one or two additional half cycles of the AC might be enough energy to cause an explosion due to the arc, moving to the alumina case and puncture it. The reason for the above constellation might be errors in the measurement and recording of the interruption current and arcing time to which the breaker was exposed. Our protection units (breaker monitoring) measure current, voltage and arcing time of all three poles of a breaker unit and build a sum current. By means of well know ablation formulas we are able to predict when it is recommended to refurbish the interruption contact system, which is certainly cheaper then refurbishing the breaker room.