Abstract

In a drilling program, the design of wells has historically been an incremental process in which compiled experience from previously drilled wells is the principal driver for realizing drilling design. The efficiency of drilling in a field, and as a result, the cost, typically improves significantly initially, before reaching a point at which little improvement is achieved. The progression from the initial well to the final wells of a mature drilling program is often referred to as the "drilling learning curve." Rather than focus on getting the design right after several wells, it is possible to outperform the learning curve by developing and applying fundamental physics-based models and computational optimization engines up front to quantitatively analyze well design options. These proprietary tools have been applied extensively in field development planning, as well as in real-time, to ensure optimal drilling performance. This paper describes ExxonMobil's approach to physics-based modeling, advanced mathematical optimization, and real-time updating to achieve drilling performance that both accelerates and undercuts the historical learning curve. Specific business examples illustrating successful applications of this approach are presented.

Introduction

Optimal well design requires attention to both fundamental engineering principles and site-specific learnings. The drilling learning curve, which identifies adjustments to initial design parameters to improve overall well design, can frequently cut 50% or more from well costs over the course of a continuous program. While a variety of learning-based well design techniques have served the industry well, the biggest prize is achieved by utilizing physics and mathematics to shift the optimization process from the learning curve to the initial well design – getting it right the first time.

ExxonMobil's approach to going beyond the learning curve is based on routine application of proprietary technology that couples fundamental physics-based models with statistical analyses. The models enable the accurate prediction of expected drilling performance over a wide range of operational conditions. By accounting for downhole parameter uncertainty, engineers effectively optimize this performance over all expected scenarios.

This content is only available via PDF.
You can access this article if you purchase or spend a download.