Real-time prediction of formation pressure gradient while drilling.

Area of Science:

• Petroleum Engineering
• Artificial Intelligence in Geosciences
• Drilling Operations Optimization

Background:

• Accurate real-time pore pressure prediction is vital for drilling operations, impacting cost, time, and safety.
• Existing pore pressure prediction methods often rely on logging data or complex formation analyses.
• The need for efficient, real-time predictive models using readily available drilling parameters is significant.

Purpose of the Study:

• To develop and evaluate artificial intelligence (AI) models for real-time pore pressure gradient estimation during drilling.
• To compare the performance of Artificial Neural Networks (ANN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) for this task.
• To introduce a practical, data-driven approach for pore pressure prediction using drilling data.

Main Methods:

• Application of Artificial Neural Networks (ANN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) models.
• Utilizing drilling parameters such as rate of penetration (ROP), mud flow rate (Q), standpipe pressure (SPP), and rotary speed (RS).
• Training and validation using datasets from vertical wells to assess model accuracy and reliability.

Main Results:

• Both ANN and ANFIS models demonstrated strong predictive performance with high correlation coefficients (R) during training and testing.
• The developed models achieved a low Average Absolute Percentage Error (AAPE) of less than 2.1%.
• Validation confirmed the models' accuracy in estimating pore pressure gradients in real-time.

Conclusions:

• The study validates the reliability of ANN and ANFIS models for estimating pore pressure gradients using real-time drilling data.
• These AI-driven models offer a cost-effective and efficient alternative to traditional methods.
• An ANN-based correlation is provided for direct application when drilling parameters are available.