In the thesis is the structure of the numerical model and the simulation results of a water model which imitates the flow of steel during continuous casting of steel process. The simulations predict the flow behaviour in the mould, which affects the quality and the efficiency of this process in the industry. A geometric model of mould's water model and a 3D block structured spatial discretization are developed. Using the RANS approach for modeling turbulence and the VOF model to address the free surface, a three-dimensional model is developed. Together with accurately selected computational parameters, the model enables precise, robust, independent and numerically stable simulation of velocity and pressure conditions of the water model. The realizable k-epsilon and SST k-omega models are used to model the turbulence. A simulation and analysis of velocity profiles and velocity fields at selected locations is performed. The simulation results of both turbulent models are compared with each other and with the results of the particle image velocity based experimental data. The comparison of the simulation results of the individual models with the experiment shows that the k-omega model is slightly more suitable for prediction. The comparison also shows that both turbulent models can qualitatively predict the primary flow in the upper and middle section of the mould. However, in the lower section of the mould, the results of both models do not exactly match the experiment in the primary flow. Although k-epsilon and k-omega models adequately describe primary flow, their use is unsuitable for proper predicting of the secondary flow.