Mean-time-to-failure (MTTF) enhancement is key task for oil production operator. To achieve this goal they increasingly engage remote well stock performance monitoring systems. Regular acquisition of the detailed information on the well performance and the downhole equipment condition facilitate routine well control for the production engineers as well as enables to predict some well performance deviations before they actually arise, for example, oil production declines. Such predictions might initialize actions on the well performance optimization before a failure occurs or operation conditions change. In many cases it could prove to be economically feasible due to less well shutdown period while waiting for the repair teams to arrive.
This report specifies criteria which are crucial for practical implementation of the downhole equipment failure prediction system. The main criterion is the failure time prediction accuracy. In this report "prediction accuracy" notion is described based on the statistical figures reflecting the number of successful predictions. A comparison of economical effectiveness of the various accuracy prediction systems application and existing practical approaches such as "change broken" application is provided as well.
There are applicability maps for the failure prediction systems compiled in terms of Rosneft well stock example. It is shown that the range of application of such systems exists, but is significantly limited.