Traditionally, reliability analysis is performed on a system design to estimate a reliability number with a given confidence level. Such a conventional method analyzes the failure rates of the components with an exponential distribution assuming a constant failure rate over time. However, in reality, for a drilling system reliability changes over the drilling period owing to component degradation, varied processes, and the drilling problems involved. Hence reliability becomes a function of not only component failure rates but also of the drilling operations, problems encountered, and depth of drilling or time.
In this paper an example of Dual-Gradient Drilling (DGD) technology is evaluated. The subsea equipment studied in the analysis includes the drill string, drill string valve, blow-out preventer, subsea rotating diverter, and the subsea mud-lift pump. A dynamic reliability analysis is performed on the DGD system to account for multiple states. These states are analyzed using the Markov process. However, using only Markov chains does not analyze the failure mechanisms and their effects on other components. Hence to overcome this limitation, a Failure Modes and Effects Analysis (FMEA) is performed to better understand how a failure affects the system. Then the probabilities for the Markov process are based on these system interactions. During the FMEA analysis it was recognized that the change in failure rate of an equipment failure mode over time is a function of different factors and hence should be assessed using different failure density distributions.
A limited drilling scenario is created with drilling rates and trip times calculated to provide the necessary data. There were two types of perturbations or disturbances modelled in the simulation that occur during the drilling scenarios. These are (1) the influx of formation fluids into the drilling mud and (2) equipment failures. A Monte Carlo simulation for the entire system is developed to assess the reliability of the drilling system at every 30' depth interval. The Monte Carlo method used different probability distribution models for different failure modes. Thus, the dynamics of system behavior and how equipment failures interact and change over time and damage accumulations and reversals (repairs/replacements) are incorporated into the model.
Ten simulation runs were carried out for the scenarios and perturbations which gave results for a total of 70000' of drilling data over approximately 109,174 hours. This paper presents the observed influxes and equipment failures. The dynamic reliability results show reliability varying with depth which is a realistic case as compared to the system's conventional reliability results. The results database can be updated with new field data as drilling progresses. This will allow the analyst to compare anticipated reliability with the revised anticipated reliability going forward in the drilling operation. Thus, the paper introduces a methodology for applying the concept of ‘reliability-informed drilling’.