History Match Case Study: Use of Assisted History Match Tools on Single-well Models in Conjunction with a Full-field History Match

Abstract

This paper describes the use of Assisted History Matching (AHM) techniques as a systematic and efficient process for integrating dynamic data (production data) into a reservoir model. As a proof of concept the AHM process was applied to obtain a history match for flowing bottom-hole pressure (FBHP) and gas-oil ratio (GOR) data in the Chayvo field (Sakhalin-1).The AHM workflow developed for this case study consisted of three key components: 1) Quality of match measures 2) Uncertainty Analysis (using Design of Experiments) 3) Static Connectivity MeasurementsTools and workflows were developed to modify the properties of the simulation model and to make static (Shortest-pathbased) connectivity measurements on them. Quantitative quality-of-match measures were developed and used as the response to a change in values of history match parameters. Uncertainty Analysis methods were used to maximize the information obtained from a limited number of single-well sector model simulation runs in order to identify key factors that drive the history match for the Chayvo field.Basic parameters thought to impact HM were the horizontal and vertical permeabilities in three major facies -Upper Shoreface (USF or EOD 3, good quality sand), Middle Shoreface (MSF or EOD 2, intermediate quality) and Lower Shoreface (LSF or EOD 1, low quality). Permeability thickness (KH) information from well tests was used as a constraint on average permeability. This reduced the total number of independent parameters to five. The ranges (maximum and minimum bounds) of these parameters were obtained in consultation with geoscientists. A series of experimental designs were executed, each successive design including more details in the model. Vertical permeability in the MSF and the permeability contrast between MSF and USF were found to be the two major factors that had the largest impact on the quality of the history match. These two factors were then varied systematically to get an improved history match for the sector model.The generation of workflow scripts allowed the history matching process to be executed efficiently, saving the time required for repeated manual input. Also, correlations were found to exist between the static connectivity measurements and the quality-of-match measures derived from simulation results. These indicated that models within a certain range of static drainage volumes for each well were more likely to yield a good quality of match. These correlations could be used to screen future simulation and geologic models. The design of experiments analysis gives a good indication of the key history matching parameters, thereby guiding an understanding of underlying mechanisms and also reducing the number of factors to be considered to improve the match.The learnings from history matching on the sector model were applied to the full-field geologic model resulting in a significant improvement in the well-by-well and the overall field-match to production data. Additional changes required to match individual wells were primarily localized.