As seems to be the case so often, the failure was the result of multiple errors and misjudgments in design, construction and operation. The reservoir did in fact have redundant sensors, but the backup sensors (conductivity probes) ended up being above the overflow elevation of the parapet wall! And the primary (pressure) sensors were reading low because they had come loose from their mountings.
This would make an excellent case study for a class, because the basic problem is so simple. And the report touches on an interesting question for discussion: the fact that the control system could have been programmed to identify abnormal conditions - without any additional hardware - by comparing water depth readings with expected values based on pump runtime.