Traditional Estimated Ultimate Recovery (EUR) forecasting often fails to guarantee accuracy while using early-time production data. The estimation precisions derived from different empirical equations can be varied. Recent studies develop EUR forecasts and demonstrate its applicability during the early stage by using Pressure Normalized Rate (PNR) as the indicator. This paper inherits the basic idea and performs a comparative study using a probabilistic PNR approach to forecast EURs based on early-time production data. The comparison is conducted using Arps, Stretched Exponential (SEPD) and Mittag-Leffler (ML) decline model with P90-P10 uncertainty quantification.
In this study, we first categorize all the variables from the shale gas reservoir in the Sichuan Basin into geology and completion classes. The representative type well is established with P90-P10 interval and then the PNR transformed production data are curve-fitted using Arps, SEPD and ML decline model. The flow regime transitions are identified and fixed in each type well curve P90, P50 and P10. Further, the early-time data is projected on each PNR decline model and performs EUR forecasting. We take two wells with long-term histories for hindcasting before four wells with limited data are applied. The applications demonstrate the applicability and efficacy of the comparative studies by showing Arps with optimistic, ML with moderate and SEPD with conservative estimates.
The successful application in field cases shows this comparative study is of necessity for early-time producing well EUR forecasting and reserves booking. The outcomes of this study are threefold: 1)The most critical takeaway while predicting EURs during early production stage is to quantify the uncertainty of the flow regime transition, and we incorporate type well concept to maintain the consistency and minimize the discrepancy; 2)The calculated decline based on each decline model differs mainly after the pseudo-steady state is achieved, the Arps, ML forecasts moderately to optimistically, while SEPD forecasts conservatively; 3)Proved Developed Reserves (PDs) can be highly-likely booked using decline model instead of using volumetric approach while providing early stage wells' reasonable and comparative forecasts based on P90-P10 interval.