Improved abnormal formation pressure detection from drilling performance data in a partially observable setting

Abstract

Pore pressure prognosis during drilling is one of the most impactful pressure related information gathering activity for operational decision making. It enables the establishing of an appropriate overbalance, or differential pressure with bottom hole pressure (BHP) exerted by the drilling fluid, and has a direct bearing on the safety, efficiency, and cost of drilling the well. There are multiple approaches for detecting formation pressure but penetration rate (or drilling data) models are the only wholly internal, and most direct approach available in the partially observed downhole drilling operations system. In addition, they are unique in that they also encapsulate the variances from the uncertainty of the changes in the pressure data within the drilling performance optimization system. They however have been historically challenged because of the perceived limitation with the degree of their accuracy, largely as a result of conflating factor effects on drilling rate of penetration (ROP). This paper investigates the evolution of differential pressure in the drilling wellbore for the purpose of detecting abnormal formation pressure changes essential for improving drilling performance. The modified d exponent ( d m o d ) was identified as an applicable source of diagnostic data for analyzing the downhole drilling system, and factorial design experiments were conducted to study the response of the d m o d variable to changes in factor effects and systemic error. The results and analysis of variance from the experiments established the interactions of the two main exogenous factors - differential pressure and lithology - that contribute to the variations in ROP, and the unique effects of these often coalescing factors were extracted. A generalized stochastic model from the analysis of variance (ANOVA) extended with a randomized complete block design (RCBD) was thereafter developed for the evolution of differential pressure in the wellbore with uncertainty in operational time. The fitted model was used to calibrate abnormal pore pressure detection from the synthetic drilling data of the deepwater field case study. The results showed an improved detection accuracy, compared to classical drilling prognosis methods. This illustrates the effectiveness of our improved method in detecting abnormal formation pressure changes from drilling performance data and modeling the effect of overbalance transitions on ROP needed for optimal decision making in real time. • Identification of “ d m o d ” as an applicable source of diagnostic data for analysing the downhole drilling system. • Distillation of effect of differential pressure ROP from other exogenous factors for drilling optimization decisions. • TDevelopment of a fitted oilfield RCBD model, to evaluate differential pressure evolution in the wellbore in operational time. • Creation of a confidence basis for making drilling optimization decisions based on minimizingrisk of system failure. • Utilization of design experiments and stochastic modelling as a viable approach for solving downhole drilling problems.