Two-phase flow simulation within injection cavity of cryogenic rocket engines

The L. code is a multiphase, transient, 3D Computational Fluid Dynamics (CFD) code used in the Airbus Safran Launchers combustion department to simulate the filling of cryogenic rocket engine injection cavities. The aim is to study the flow topology and property at the injection elements outlet in order to be predictive on the rocket engine starting phase. This code asks for numerous inputs. Most of them are classical of CFD codes, case dependent and are easily set up. One of these inputs yet is hardly accessible: the initial droplets diameter. Indeed, when it comes to deal with heat transfer, the liquid phase is considered as a collection of spherical objects of equal size within a volume element. The size of these droplets set the total exchange surface and then evaporation and condensation. An error on the initial droplets size can entail very large error on the amount of the gas/liquid created making the code non predictive. The main goal of this thesis was to find a methodology to set up properly the initial droplets diameters so that the physics of the filling phase is correctly restituted. More generally, the objective was to make the heat transfer module more realistic. In order to tackle the droplets size determination problem, two methodologies have been investigated. The first one relied one test campaigns performed by a sub-contractor laboratory. Exploiting the results, the goal was to perform a parametric study based on the initial droplets diameter to make the simulation converge incrementally towards the real-life tests results. Alongside, another methodology has been investigated: recalibrate the code based on a physical, measurable and unambiguous criterion: the droplets outlet diameter. The study of the droplets stemming from atomization has for long entailed numerous modeling that have been exploited in this thesis in order to find a formulation consistent with the physics of the problem for the diameter of the droplets streaming out the injectors. Then, the goal was to set up the initial droplets diameter so that the computed outlet ones match the theoretical value. This methodology has been used in a comprehensive parametric study based on the droplets diameter that came up with interesting results about its impact on the solution of the simulation. In particular, heat transfer effectiveness regimes and their location have been discriminated, making possible for the user to explore limiting cases, the real solution being somewhere in-between. Another side of this problem was about the modeling of the droplets heating/cooling. Indeed, heat/mass transfer are not only dependent to the total exchange surface but also to the rate at which the liquid/gaseous phase is heated/cooled down. The temperature profile within the droplets reveals then to be a very sensitive aspect of the modeling. Four temperature profiles have been studied and implemented, each relying on different physical arguments. On the whole, this thesis has allowed a better understanding of the effect of the initial droplets diameter on the results of the simulations and to obtain interesting qualitative results that can be used to bound the real solution of the problem. At last, this thesis demonstrated the crucial need of qualitative experimental results so that the problem is definitely closed.