ABSTRACT :

The development of efficient simulation tools requires an understanding of physical modeling, mathematical modeling and computer programming. For each of these domains it is necessary to bear in mind the intended application, because the use for a calculation code or simulation software will dictate the level of modeling, and also the programming techniques to be adopted.

This dissertation starts with a detailed description applied in the form of fluid flow calculations using the Euler equations. Then simulation of an industrial benchmark is considered using a parallel computational method. Finally, simulation of a complete industrial plant is addressed, where phenomenological relations based on experimental correlations can be used.

The first chapter deals with the determination of mesh velocity in the context of ALE (Arbitrary Lagrangian-Eulerian) methods. In the following chapter we focus on the compressible Euler equations solved using the FVCF method (Finite Volume with Characteristic Flux). In this case we consider an interface between a single fluid and a homogeneous two-fluid mixture, where one of the two mixed fluids and the single fluid have the same equation of state.

The third chapter is devoted to running high performance simulations using the FluxIC computation code based on the FVCF method with interface capturing. The focus is on sloshing phenomenon encountered during transportation of Liquefied Natural Gas by LNG carriers.

The fourth and final chapter deals with modeling of an industrial facility at system level. A systemic approach is presented that provides a level of modeling adapted to the simulation of a large number of components and their interactions. This approach enables users to combine deterministic modeling of physical phenomena with stochastic modeling in order to simulate the behavior of the system for a large set of operating conditions.