With the emergence of liquid rich shale(LRS)plays like Eagle Ford and Northern Barnett,the petroleum industry needs a simple,easily applied technique that provides reliable estimates of future production rates in this kind of reservoir.There is no guarantee that methodology that has proved to work in gas reservoirs will necessarily be appropriate in LRS reservoirs.In this work,we found that without corrections of early data,the Stretched Exponential Production Decline(SEPD)model,designed for transient flow,usually produces pessimistic forecasts of future production.The Duong method,another transient model,may be reasonable during long term transient linear flow,but notably optimistic after boundary-dominated flow (BDF)appears.For wells in BDF,the Arps model provides reasonable forecasts,but the Arps model may not be accurate when applied to transient data.A hybrid of early transient and later BDF models proves to be a reasonable solution to the forecasting problem in LRS. In addition,use of diagnostic plots(like log-log rate-time and log-log rate-material balance time plots) improves confidence in flow regime identification and production forecasting.In some LRS’s,BDF is observed within 12 months.In any case,it is essential to identify or to estimate the time to reach BDF and to discontinue use of transient flow models after BDF appears or is expected. We validated our methodology using“hindcast analysis”;that is,matching the first half of production history to determine model parameters,then forecasting the second half of history and comparing to observed production data. We also found that application of pressure-corrected rates in decline curve analysis(DCA)may substantially improve the interpretation of data from unconventional oil wells flowing under unstable operating conditions. Fetkovich(hydraulically fractured well)type curve analysis can be added to improve confidence in flow regime identification from diagnostic plots and to estimate the Arps hyperbolic exponent b from the matching b stem on the type curve,which can then be extrapolated to determine estimated ultimate recovery.
展开▼
