How to Improve Efficiency in Multiple History Matching:A Gas Field Case Study

摘要

This paper proposes an innovative integrated workflow based on advanced history matching techniques for finding multiple history-matched models. The proposed workflow is demonstrated on a condensate gas field.The deltaic depositional environ- ment enhanced by normal faults created strong vertical and horizontal heterogeneities.This compartmen- talization,added to the production in commingle,makes the history match particularly difficult.The proposed workflow begins by an intensive dynamic synthesis and a multidisciplinary teamwork definition of the objects to match(production rates,RFT,MDT,PLT).The potentially influent parameters and the associated ranges of uncertainty are also defined. Early cooperation is crucial,as it sets up the scope of the history match strategy.This meticulous synthesis of static and dynamic observations enables to build the geological model and the associated history match strategy.Then,the following steps are performed: –Screening:experimental design is used to highlight the most impacting parameters.Given the ranges of uncertainty defined in the synthesis,this step allows the reduction of uncertain parameters (32 in our case)to the most influential ones(13 parameters); –Modeling:an innovative space filling design technique is used to define a limited number of simulations using only key uncertain parameters; –Objective function formulation:different formulations are proposed to evaluate the history match quality; –Assisted history matching:Proxy modeling is introduced in the framework of the objective function minimization and multiple history-matched models are obtained and then validated by the team. Results on this field case show the necessity of a strong and coherent dynamic synthesis.The proposed workflow enables to evaluate the different impacting parameters of the model and thus finding different history-matched models.Provided physics and geology are clearly understood,this assisted history match technique is highly helpful in particular for complex fields with a large amount of data to match.