Analysis of production data is a critical component for performance estimation in ultra-low permeability unconventional reservoirs. Diagnosis of early-time infinite-acting transient flow is important for accurate estimation of future unobserved time-rate data. However, well clean-up effects and lack of pressure data may mask the earlytime flow regimes, prohibiting accurate diagnosis of the end of the transient regime by graphical techniques. A model-based approach can improve forecast accuracy by bounding the range of possible interpretations of production performance.
The familiar Arps hyperbolic model is re-cast in terms of the colloquially named "linear flow parameter", relating the initial rate and initial decline parameters with a functional relationship. Computation of initial rate can be performed from a diagnostic fit of a ½-slope trend line that corresponds to transient linear flow. Additionally, the early-time masking of linear flow due to well clean-up may be quantified by computing a diagnostic ½-slope trend from the hyperbolic model parameters. Functional relationships are provided to compute additional diagnostics on the basis of material balance time and a novel transient superposition rate function. Parameterization of linear flow duration allows creation of a coupled GOR or CGR forecast for secondary phase yields.
Long well clean-up periods and short duration transient regimes present a challenge for making consistent production performance interpretations, because graphical techniques assume ideal conditions that may not exist in field data. Moreover, diagnosis of flow regime change must take place rather late after-the-fact, as confidence of visual deviation from a straight-line trend is required. An empirical model-based workflow that maps the fluid flow solution for linear reservoirs to a simple empirical model assists with the creation of performance estimates that achieve greater consistency with those from more rigorous rate-transient analysis workflows. The additional diagnostics, which are computed from the model parameters, alleviate some of the difficulties caused by noisy data sets.
The workflow is simple and robust, while respecting the physics of fluid flow theory. Various successful rate-transient analysis concepts and current best-practices for production performance estimation are encoded into the model parameters, and therefore may be easily applied. The diagnostics that result are not simply additional plots or manipulation of scales, but functional computations mapped to common Arps hyperbolic parameters. Given the presentation of numerous diagnostics - for a given model input, an exact set of interrelated diagnostics is provided - forecast accuracy should improve. We present the workflow in terms with which the typical reservoir engineer is already familiar. This enables direct application in day-to-day work for estimation of future well performance.
