Post-Match Ensemble Analytic Techniques to Investigate Posterior Forecast Uncertainties

Ensemble history matching is an effective tool which allows for uncertainty quantification, but it could be very time consuming. We show that some basic techniques can be used to reduce involved workflows complexity and computational time. The objective is to present how the statistical analysis techniques, commonly used for initial ensemble quality check, can be applied to the ensemble of matched models to highlight the major source of uncertainty and post-match forecast impact. With ensemble history matching an initial set of different models is conditioned to production history. To create the initial ensemble, all relevant uncertainty parameters are selected after sensitivity and included using a workflow, which can become very complex and time-consuming, inside a geo-modeler. More realizations are requested until the initial ensemble passes quality check analysis. An iterative algorithm, requiring a preset number of iterations, has been used to match the ensemble of model realization. The matched models are then used to determine the forecast production profile range. Different methods are then applied to understand the most impacting parameters on forecast outcomes. The uncertainty analysis workflow is applied to a tight gas field with some years of production history (early development phase) that is going to be integrated with other fields as a part of an LNG project. Initial sensitivity analysis helped to reduce the included uncertainty parameters preserving the original forecast uncertainty range but reducing the workflow complexity and the needed computational time. All updated ensemble realizations have been used for further analysis on the forecast. Analyses performed on matched ensemble (including correlation analysis and clustering) were used to highlight the most impacting parameters on forecast outcomes and to select most representative cases to be inserted in integrated asset models. The choice of the cases has been guided by the most relevant outcomes for the projects (like total field gas production and gas plateau length) and most relevant impacting parameters. The presented uncertainty quantification workflow and associated analytic techniques have led to a roughly halving of the standard time needed for the uncertainty analysis (after history match) of the tight gas case. Further analysis on the posterior uncertainty helped in risk mitigation and its consecutive decision-making by changing the original drilling schedule to accelerate data acquisition that can reduce the forecast uncertainty range.