Conditioning an underground gas storage site to well pressures

This paper is concerned with the integrated modelling of an underground gas storage (UGS) site, considering geological data as well as production data. The production history for this UGS, located in an aquifer close to Paris, France, consists of two periods. The first one is a 7-year filling period for which the cushion gas was injected. The second one is a 10-year cycling period for which gas was injected during the summer to replenish the reserves and withdrawn during the winter according to the demand. Gas storage was performed through a dozen wells. During these two periods, pressures were recorded in 11 observation wells located all around the site. The first period data only are used to constrain the reservoir model. The second period data are used for comparison purposes. An assisted history-matching process based on the gradual deformation method was implemented to integrate production history jointly with geological data into reservoir models. By use of this innovative technique, both stochastic and deterministic parameters were accounted for. We show that the most influential parameters are related to the petrophysical properties (porosity means, coefficients defining the relations between porosity and permeability). The stochastic parameters are of second order only: they did not impact the matching process. The optimal reservoir model deduced from the matching procedure reproduced the static data and the pressures recorded for almost all the wells. The assisted history-matching methodology developed is especially powerful for managing a workflow integrating every steps from fine grid model creation to fluid flow simulation, handling rapid preliminary history-matching tests, analyzing the influence of several deterministic and stochastic parameters, determining fine grid reservoir models consistent with all the available data, and lastly reducing uncertainty in predictions.