Reserve estimation of unconventional formations is a new challenge to reservoir engineers due to the geological
uncertainty and complex flow patterns evolving in the multi-fractured horizontal wells (MFHWs). Some predicting
models have been presented and widely used in MFHWs exhibiting a long-term linear flow, such as stretched-exponential
production-decline (SEPD), power law exponential decline (PLE) and Duong’s model. Plenty of successful field
applications of these models seem to have demonstrated their availability and correctness especially in the transient
linear flow period.
Due to the limitation of reservoir boundaries or size of stimulated volume, any fractured tight reservoir will eventually
exhibit a boundary-dominated-flow (BDF). The models above which show “goodness of fit” in linear flow may not be
used or will cause great error when used to predict production in BDF period.
This paper compared the newly developed PLE model with the traditional Arps’ hyperbolic decline model in terms
of production historic match, decline rate and decline exponent during and after linear flow. The analysis result
demonstrated that PLE model actually cannot match production decline characteristics as previously thought when
only linear flow appears and it is a model which should be used in the transition period rather than linear flow period
as applied in the past few years. The wrong usage of the model will cause great error to reserve estimation. The
modified steps to predict production in different flow pattern are given in this work. The outcome of this work should
help the industry to forecast production and ultimate reserve more accurately in tight oil and shale gas reservoirs.