Sealing the Wells
Shortly after the cementing was done, workers started sealing the well. This is done by pumping the "killing fluid" (a concrete mixture) down into the casing pipe, through the kill line shown in Figure 2. In this phase of the operation, until the concrete sets, the maintaining of the killing fluid pressure high enough to always exceed the formation pressure is essential, and this pressure can be controlled by keeping the concrete mix density high enough to provide the required hydrostatic pressure that prevents the blowout.
Instead, the following occurred at BP. Cementing was completed on April 19, 2010. The next day, on April 20 at 7 a.m., BP cancelled the test required to determine if the bonding of the cement was strong enough in the annulus, nor did engineers check the blow-out preventer (BOP) and just started sealing the well. They were in a hurry because the crew performing the sealing was to lave on an 11:15 a.m. flight.
Between that time and the time of the blowout some 10 hours passed, during which time no corrective action was taken. After that, at around 9:40 p.m., a jolt was felt on the bridge followed by the rig shaking and alarms being activated, because the most dangerous level of combustible gas intrusion was detected. Yet electricity was not turned off and at 9:45 p.m. the gas exploded, and oil and concrete was blown of the well onto the deck and ignited. The rest we know.
So what happened? Obviously, the cement plug was not strong enough to stop the oil and gas from blowing out. So was that predictable and would automation have prevented it? The answer is a definite yes! Testing indicated the presence of a leak somewhere in the well. In spite of that, because everyone was in a hurry, BP decided to use a low-density cement plug and seawater behind it, instead of keeping the concrete mix density high enough to provide the required hydrostatic pressure that would have prevented the blowout.
If this operation were automated, the balancing pressure required to exceed the formation pressure would have been automatically calculated and applied. In other words, the control system would have kept the hydrostatic pressure high enough to prevent the gas from entering the well and would have prevented the use of low-density cement or sealing fluid.
Therefore, in order to protect against the repetition of the BP accident, it is essential to have reliable (redundant) sensors and fully automatic response to unsafe conditions and mistakes made by either untrained operators or by ones willing to cut corners. This requires using reliable sensors and "smart annunciators," which not only inform the operators about the existence of unsafe conditions, but either gives them instructions on what to do about them or, preferably, automatically does it.
In case of the BP accident, this would have not only prevented the application of the weak concrete plug that allowed the blowout, but would have stopped also the whole operation until the blowout preventers were tested and would have automatically disengaged the rig from the well as soon as fire was detected.