Using data analytics to assess the impact of technology change on production forecasting

The nature of well completions in the oil and gas industry continues to evolve. Although the effects of completions and spacing on initial production are well reported, how they affect ultimate recovery and terminal decline is not well understood. Over the last decade, drilling on multi-well pads has become prevalent, spacing between horizontal wells has decreased, and hydrofracture intensity has increased. These developments have decreased drilling and completion costs, while increasing initial well production. Yet, the impact of the timing, spacing, and intensity of fracturing on terminal decline rates and ultimate recovery has not been systematically investigated. In this paper, Bakken well production profiles are used to evaluate the impact of differences in completion design on the nature of long-term production decline. To evaluate these effects, production for 12,000 Bakken wells were forecast using a physics-based approach. Using descriptive statistics and advanced visualization, terminal decline rate and ultimate recovery parameters are found to depend upon date of well completion, volumes of proppant and water injected, lateral length, and well spacing. We utilize a tree-based machine learning approach to test predictability of completion parameters on terminal decline rate and estimated ultimate recovery. Our analyses show that pad drilling and increased hydrofracture intensity are apparently associated with small increases in initial production rates but have led to larger terminal decline rates. For example, in the Bakken, the terminal decline rate increases by upwards of ten percentage points for wells with modern completions in multi-well pads. Since production life is dependent upon terminal decline rates, spacing and completions effects must be accounted for in type curves for wells in multi-well pads.