![]() ![]() Only a part of this excess in predicted production is due to the linear interpolation.99% of wells drilled in the US today are horizontally drilled unconventional wells with a bunch of man-made propped hydraulic fractures. The ND decline curve clearly results in substantially higher projections than were realized. Sanish initially outperformed, but now tends to underperform. ), while others underperform with respect to the average wells (Little Knife). Some fields systematically outperform the expected production from our deline curve (Parshall, Spotted Horn. The projection of total production for a field gives an idea how its wells compare to the average decline curve we constructed. The results for the twenty most productive fields are shown in the figure below. Some "legacy" fields will have some wells producing oil which we do not include in our model. We then aggregate the data on a per-field level. Using all wells with a peak in oil production starting July 2005, we project out production per well. ![]() We linearily interpolate the ND decline curve data points to monthly values, which will overestimate production in the first two to three years by a some amount. If many new wells are going into production, this small difference is expected to be neglegible. We will slightly overestimate production on an aggregated per-field level, which might be apparent should only few new wells be added over time. We linearly extrapolate for the 20k ft MWD curve beyond 82 months, using mean production data from months 36 to 72. We will use the average 20k ft MWD decline curve constructed above, as well as the ND decline curve, to extrapolate production values. Geological factors are believed to have a larger impact on production than technological advances, according to Theloy and Sonnenberg. This increase is likely due to a better understanding of geological structure and increasingly experienced operators. Both the Bakken and Three Forks formation likely have differences in their overall characteristics which we will ignore, thereby assuming that the variance within the two formations is larger than the differences between them.įor many fields, a clear rise in the magnitude of the initial production value can be seen in figure B, e.g. We therefore might expect a lower oil production on average, while keeping all other factors equal. įor lower measured vertical depths, the lateral extent of the well is shorter. ![]() ![]() We automatically select and label wells at around 15k/20k ft MWD via a clustering algorithm, see figure A. The Bakken/Three Forks formations are at 2 miles =~ 10k ft TVD. Horizontal drilling is employed and the measured well depth (MWD) is not equal to true vertical depth (TVD). There is prominent clustering of measured well depths when plotting vs. year, for the twenty most productive fields. measured well depth, as well as initial production vs. The two figures below show initial production vs. We can cross-reference the measured well depth for each initial production value in the enigma.io datasets. We only have production data at a monthly resolution available, and will try to give a reasonable estimate for the typical decline curve using this coarsely spaced data. We potentially introduce a selection bias doing so. We select wells with an oil production maximum within July 2009 to July 2014 to minimize a potential influence of oil price, and restrict ourselves to the twenty most productive fields in terms of total volume (as of 2015, covering ~40% of the total output). We assume that the initial production is a good indicator for long term profitability. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |