Sector modeling using iterative boundary pressure estimation

Abstract

Simulation of large-scale fields with high-level of heterogeneity are commonly modeled using high-resolution grid cell, which can be time-consuming to run the entire reservoir numerical model. One way to deal with these cases is to split the model into sectors and individually simulate them using artificial boundary conditions to represent the fluid flow among the sectors. However, the representation of the boundary condition can be affected by operational or strategic changes, leading to poor-quality forecasts. This work aims to present a methodology to update the pressure in the boundary of sector models using virtual wells. The methodology consists of balancing the pressure of adjacent sector models over the simulation run. The update procedure is carried out observing boundary block pressure and applying an estimation of the flow rate between sectors and using a pair of virtual wells (producer/injector) to perform it. The methodology was tested in the Brugge Field Benchmark Case, which is a complex case that presents high-interaction activity among the wells. A tree of tests was created to study the parameters of the method, as a different number of times to update the boundary pressure and a different number of pairs of virtual wells. Furthermore, two different ways to split up the reservoir model were also tested: using streamlines and not considering any dynamic flow information, to evaluate the importance of using a technique to select the sectors. The quality of the results was quantified using the indicator called normalized quadratic deviation with sign, comparing the sector simulation and the entire reservoir model simulation. The results show that the methodology was able to improve the quality of the response to all cases. The case splitting the reservoir using streamlines achieved better results, although the improvement of the response was lower than found for the random split. The number of times to update the model was the parameter with a more significant impact on the quality of the results. A higher number of update times tended to achieve better outcomes. The number of virtual wells was the parameter that caused a lower impact on the results, and significant gains in increasing the number of virtual wells were not observed. Therefore, just a few numbers of virtual wells can balance the boundary pressure. To conclude, the proposed methodology can improve the quality of the results for a sector simulation model. The results also show an improvement when the methodology is combined with a splitting procedure of the reservoir using streamlines. The number of times to update the model can significantly affect the results, but the number of virtual wells had a lower impact.