Based on the empirical knowledge, the fault diagnosis model is built adopting fault tree analysis method which considers the uncertainty.
Based on such knowledge, the fault parallel trend of σ 1 axis (fault normal trend of σ 3 axis) suggests a low coefficient of friction.
Based on this knowledge, the fault parallel trend of the σ1 axis (fault normal trend of σ3 axis) suggests a low coefficient of friction on the planes that accommodated the aftershock activity of the Yalova-Çınarcık cluster.
This scheme allows isolating specific faults in the closed-loop system using a purely data-based approach without requiring prior knowledge of fault history.
The presented method uses the knowledge of fault effects on the measured output, when the inputs are successively excited quasi-stationary, to determine the location of the fault.
This is, however, challenging due to nonlinearities, noise, uncertainties, as well as the combination of qualitative and quantitative process knowledge and fault specifications.
An alternative approach with the following features is required: ability to interpret and use the vast quantities of process data, respond with high speed to process inputs, mapping capability/pattern recognition, adaptivity, learn and capture knowledge, and fault tolerant and robust.
The ratio of these two probabilities, G = Pr EOPF|Δ tOPF, EFF /Pr(EOPF|Δ tOPF), represents the probability gain due to the knowledge of fault interaction.
A design of a multivariate knowledge-based fault diagnosis system is described in this paper.
Improved knowledge on faults and hydrocarbon seal put pressure on geologists and reservoir engineers doing reservoir modeling.
