Abstract : A good knowledge of geological reservoirs and their evolution is crucial for many applications (oil or gas prospecting/extraction, gas or waste storage, rehabilitation of industrial sites, control of soil and ground water pollution). The complexity of the coupling between chemical reactions and fluid transport, the non-linearity of the equations and the diversity of the spatial and temporal scales render the modeling compulsory. During acid gas injection in geological structures, the most reactive primary minerals dissolve due to ground water acidification. As a result, the fluid gains chemical elements and may reach a state of oversaturation with respect to other primary minerals, or with respect to minerals not initially present in the assemblage (secondary minerals); so that these minerals can precipitate. A good accuracy of the predictions is critical to have a proper assessment of gas storage capability and help prevent mechanical risks; the program ability to choose the right minerals that will precipitate, and to know precisely the kinetics of primary minerals growth and of secondary minerals precipitation is very important.
In ARCHIMEDE code (developed by E.N.S.M-S.E, Ecole Nationale Supérieure des Mines de Saint-Etienne, in collaboration with the I.F.P., Institut Français du Pétrole), only the geochemical reactions are considered, ruling out transport aspects. The purpose of the present thesis was first to solve some problems that could not allow the computation of volume variations caused by chemical reactions ; these were solved by rewriting the program main lines; the second purpose was to elaborate and implement a nucleation/ripening/growth routine for precipitation of secondary minerals.
The considering of the appearance of minerals in natural assemblages gave us hints in this task. Numerical simulations showed the shortcomings of representing precipitation by mere crystal growth for secondary minerals. We thus defined a new model for secondary minerals precipitation. The nucleation step, that is to say the appearance of first crystals, and its kinetics was taken into account. Then, the competition between nucleation and growth was computed and this required to consider crystals with various sizes and to predict their Ostwald ripening through a parallel calculation.
The outcome of this work is an entirely new algorithm able to manage, for each secondary mineral, the initial stage of nucleation/ripening and the transition to crystal growth. The sensitivity of this model with respect to the different parameters could be analyzed by help of numerical simulations. Different tests showed the model to be satisfactory.