This paper presents an integrated workflow for generating type curves using dynamic models with limited historical production data and significant subsurface uncertainty, as well as the corresponding horizontal and vertical interference factors for forecasting well performance with different well spacings and a given stimulation design. We present a dynamic reservoir simulation workflow for four co-development horizontal wells in three landing zones within the black-oil window of the Vaca Muerta formation.
History matching included 39 reservoir, fluid, completion, and hydraulic fracture parameters. Twelve thousand simulation model runs yielded 222 unique cases that best match historical bottom-hole pressure and production rate data. A ranking method was used to identify the most representative P90, P50, and P10 individual subsurface models based on pad-level 30 year-life cumulative oil and gas production forecasts.
The core objectives of this extensive dynamic modeling exercise were 1) to provide a base P50 forecast to regress Arps parameters upon to support long-term decline curve analyses; 2) to provide a range of expected P10/P90 outcomes to quantify subsurface performance uncertainty; 3) to support vertical and horizontal interference factors that quantify a relationship between well spacing and well performance. Local grid refinements and element of symmetry methods were employed for modeling simplicity.
Selection of P90, P50, and P10 production cases was non-trivial due to the presence of multiple wells. The P50 history-match case was chosen based on its alignment with the reservoir model input parameters, i.e., the P50 case matched the P50 of the input parameter distributions. Multiple methods of comparison were used on both a pad and an individual well level to justify the P90, P50, and P10 simulation history-match cases. Further, to assess oil recovery efficiency, a well drainage area was assumed, accounting for overlap of hydraulic fracture boundaries and drainage areas.
Based on a combination of rate transient analysis of multiple wells and the four-well dynamic model, a regression was performed on each of the four wells to establish a set of Arps parameters for decline curve analysis. An updated look-back including another two and a half years of production was performed and reaffirmed the Arps parameters. Future work will consider a six well pad spanning two landing zones with two fully confined wells, following the same methodology.
