Abstract
Intelligent wells have transformed reservoir management through real-time optimization, but their dependence on downhole sensors imposes high costs, complex maintenance, and reliability concerns. This study introduces a surface-sensor-based methodology that predicts critical downhole parameters-including permeability, pressure, temperature, phase flow rates, and water holdup-without downhole instrumentation. Our approach uses only surface measurements and well geometry, providing a cost-effective and operationally robust alternative to conventional monitoring systems. We developed two integrated models: Model I, validated against the industry-standard PUNQ-S3 reservoir simulation, achieves strong predictive accuracy with average relative errors of 2.42% for water flow rates, 2.34% for oil flow rates, and 2% for pressure. Building on this foundation, the Horizontal Well Simulation Model extends the methodology to practical field applications, yielding exceptional accuracy with errors below 0.4% for permeability, pressure, water holdup, and temperature, and under 0.6% for flow rates. Performance is enhanced through integration of multiphase flow correlations (Duns & Ross for vertical sections; Beggs-Brill for horizontal sections) with Ensemble Kalman Filter-based data assimilation, ensuring reliable real-time predictions under heterogeneous reservoir conditions. The proposed methodology eliminates downhole sensors while maintaining exceptional accuracy, substantially reducing hardware and intervention costs. Beyond cost savings, it enables proactive production strategies, mitigates water breakthrough risks, and extends well life. This research demonstrates the feasibility of surface-driven downhole prediction and establishes a quantitative, field-ready framework for next-generation smart well technologies, paving the way for safer, more efficient, and sustainable hydrocarbon recovery.