A new approach for production forecasting from individual layers in multi-layer commingled tight gas reservoirs

Abstract

Estimating future production in multi-layer tight gas reservoirs is necessary, but problematic. Initial production looks promising but soon decreases to levels that are much lower than first anticipated. Boundary-dominated flow to which common production data analysis can be applied to, takes up to 8 years to establish, yet future production estimations must be made at the time of drilling. Common industry software does not fully assess individual layer contribution because large uncertainties exist in layer-specific contributions to total production. Many engineers tend to use analysis methods designed for single-layer wells, which leaves future production estimations with large errors, and this can have potentially long ranging consequences for a company. This paper presents a new way of estimating future production in multi-layer tight gas reservoirs by incorporating the uncertainty that can be found in individual layer contributions. Our approach utilises single tank material balance and matches total well production data by changing layer-specific properties. A workflow has been created for the programming language Python that includes a Bayesian element to honour the uncertainty in individual layer effective permeability and individual layer gas in place. Applying this workflow to many wells in a specified area allows probability distribution functions for layer effective permeability and layer gas in place to be generated. This will result in attaining more realistic production forecasts in the form of P10, P50 and P90 for multi-layer tight gas reservoirs, and allowing the engineer to make better-informed early assessments of the future production of wells.