Eventos Anais de eventos
COBEM 2021
26th International Congress of Mechanical Engineering
Reservoir characterization comparing different ensembles methods
Submission Author:
Vinicius Mattoso Reis Da Silva , RJ
Co-Authors:
Vinicius Mattoso Reis Da Silva, Danmer Maza, Abelardo Barreto Jr, Marcio CARVALHO
Presenter: Vinicius Mattoso Reis Da Silva
doi://10.26678/ABCM.COBEM2021.COB2021-0249
Abstract
The estimation of reservoir properties with a high level of certainty is essential to guarantee an efficient reservoir management. To minimize the associated uncertainties with estimated properties, different data sources can individually or in combination be analyzed. Well test is an important source of dynamic data during the exploration of the reservoir, normally the bottom hole pressure (BHP) and the flow rate from the reservoir are post-processed to calibrate the model that represents the reservoir dynamics. The procedure of using the response from the reservoir to calibrate the model is called history matching. The estimated properties can be associated with a local minimum on the objective function used to find the parameters. Finding a local minimum is normal since history matching is an inverse problem that is highly non-linear, there are several combinations of parameters that can generate profiles that have good data matching with the observed data from the reservoir. Therefore, several methods can be used to create a set of models generating a distribution of the parameters. Those methods are called ensemble-based methods. In this work, we applied Genetic Algorithm (GA), Ensemble Smoother (ES), and Ensemble Smoother with multiple data assimilation (ES-MDA) to estimate reservoir properties using pressure and/or temperature dynamic data and make a probability distribution of the parameters in the analysis. The results indicate that for a cylindrical homogenous single layer reservoir and assuming the same quantity of assimilations and/or generations, the ES-MDA produced a better history matching of the data and a better uncertainty quantification of the porosity and permeability from the reservoir. It also indicates that both the pressure and the temperature data can be used separately or combined to estimate the reservoir parameters.
Keywords
Reservoir Characterization, Inverse problem, genetic algorithm, ES-MDA, Ensemble Smoother
DOWNLOAD PDF VIEW PRESENTATION
