This paper illustrates the impact of various complex behaviors on the long-term decline of unconventional wells. These are illustrated through numerical simulations and field production data which are compared to the various forms of expected decline behavior. The simulations are based on modeling and history matching of unconventional systems developed with horizontal wells and multi-stage hydraulic fractures. This paper focuses on tight oil systems. There are many instances where idealized plots are useful for diagnosing and comparing well performance (e.g. hyperbolic decline (Arps), square-root of time, power-law exponential, other hybrid methods and type curves). However, in many cases such decline curves do not adequately describe long-term production behavior. Early in the well life it is especially difficult to determine which relation will best fit the long term performance. Also as new wells are added interference can cause significant deviations from the earlier behavior. This paper will illustrate the causes and characteristics of long-term decline in complex systems to aid as a guide in estimating long-term decline behavior.
The results of this work illustrate the reasons for deviation from ideal behavior such as changing compressibility and total mobility during multi-phase flow, heterogeneous matrix properties, presence of natural fractures, fracture and matrix compaction and non-uniform hydraulic fracturing. Also, some methods are not applicable for variable rate or variable pressure production scenarios (the most common state of actual well production).
The paper addresses the following issues
Illustrates non-ideal decline behavior of unconventional reservoirs based on fluid characteristics;
Illustrates non-ideal decline behavior of unconventional reservoirs based on reservoir complexities;
Limitations of rate transient analysis and decline curve analysis methods.
Traditional decline curve analysis uses simple empirical curve fitting methods to graphically model the historical performance of the wells and provide production forecasts by extrapolation. The most common decline curves are Arps equations (Arps, 1945). These equations are simple enough to form linear graphs using certain plots. Generally decline curves are empirical in nature and have no theoretical basis. However, under certain circumstances the theoretical performance of the wells would follow the simple decline equations. For example, an exponential or semilog decline is theoretically expected for an ideal volumetric reservoir with constant compressibility and constant productivity index, producing at a constant bottomhole pressure.