Extended exponential decline curve analysis

Abstract

Over the last few years, the hydrocarbon production from shale plays has grown dramatically around the world. These unconventional resources have presented many challenges to the oil and gas industry. One of the biggest challenges for evaluators is to predict long-term shale production performance, especially in a timely and reliable manner.

For traditional reservoirs, the most common practice of reserve evaluation is Decline Curve Analysis (DCA), a method with decided advantages; it is not only the simplest and least time-consuming method, but it also accommodates the historical field uncertainties by honoring observed performance trends. However, applying the traditional DCA method to shale wells, engineers commonly encounter the difficulties of simultaneously matching the high initial production rate, the extremely sharp decline rate in the transient flow period, and the shallow decline resulting from boundary-dominated flow (BDF) in late-life. This suggests that traditional DCA may not be suitable for evaluating shale reservoirs. As a result, numerical simulation seems to be the best solution to provide reliable results but at the expense of extensive manpower, cost, time, and data requirements.

We propose an alternate DCA approach, which overcomes the shortcomings of traditional DCA or numerical simulation, to estimate the recoverable hydrocarbons. We suggest that a mechanism of “growing drainage volume” is an excellent way to conceptualize and model the performance of shale wells. This paper presents a new extended exponential form of the production decline analytical equation. Three empirical depletion terms, β_e, β_l, and n, have been used in the equation. The parameter β_e represents the early, sharp decline in the transient period immediately after the well is put on production; the β_l parameter represents the comparatively shallow decline in late-life when the progress of “growing drainage volume” plays the dominant role on the production performance; the parameter n is an empirical exponent. The overall decline rate can be calculated by a relationship involving, β_e, β_l, n, and time t. Even though the new method is not always more accurate than peer models, it is likely as accurate, and does not require the analyst to guess when to switch to a boundary-dominated flow model nor to force a switch to exponential decline.

We have tested and verified this new empirical DCA by both extensive field data and detailed numerical simulation results for seven wells. For each of these data sets, the comparisons between traditional DCA methods, and in some cases simulations, indicate the relative advantages of this new approach. Later, we applied this method to over 2000 wells in the Eagle Ford shale. The analysis resulted in a relatively symmetric distribution for the empirical parameter n.

Introduction

The development of unconventional hydrocarbons has become a significant resource leading to material reserve growth worldwide. In the US, one of the important contributions in our industry came from the development of shale gas and oil during the past decade. However, forecasting production and estimating shale hydrocarbon reserves is still not fully understood because of the limited knowledge of flow mechanics in the ultra-low permeability rock. This paper presents a concept of “growing drainage volume” and develops an empirical formula to forecast production.

Standard reserve evaluation methods include volumetric calculations, material balance, decline curve analysis (DCA), analogy, and numerical simulation. An evaluation process typically involves a combination of two or more methods. Among them, numerical simulation is generally believed to be the most rigorous and accurate method. The drawback for using simulation, though, is the significant data requirements. For shale reservoirs, data requirements are even more demanding and uncertain because multi-stage fracture stimulation and horizontal completions increase significantly the data needed, or the assumptions that must be applied. It is very challenging for the simulation engineer to properly take into account the interference between fractures, which requires reliable estimates of the fracture half-length, width, and fracture permeability. This method also encounters other modeling problems like relative permeability effects and extremely heterogeneous rock properties, to name a few. On the other hand, when properly applied, DCA can play an effective role because it accommodates all of these factors, which may have influenced the historical production performance. Further, it has the exclusive advantages of speed, simplicity, and the inherent reasonableness in the forecasts generated.

We note that as with any new technique, there may be wells, reservoirs, or plays (formations) that might not be suitable to this technique. In some cases, we have seen that this new technique might provide overly optimistic or pessimistic results when compared to other methods. These cases are typically associated with situations where it is early in the life of the well(s). Because only additional data and time will truly validate the methodology presented herein, we caution the user to take into consideration the prior understanding of the reservoirs, the methodology employed, and the experience of the evaluator, when applying this method to previously evaluated entities.