The main motivation of this thesis is the analysis of turbulent flows. Turbulence plays an important role in engineering applications due to the fact that most flows in industrial equipment and surroundings are in turbulent regime. The thesis has a double purpose and is divided in two main parts. The first one is focussed on the basic and fundamental analysis of turbulence models. In the second part the know-how acquired in the first part is applied to the study of air curtains.Regarding to the first part, the principal difficulty of computing and modelling turbulent flows resides in the dominance of non-linear effects and the continuous and wide spectrum of time and length scales. Therefore, the use of turbulence modelling employing statistical techniques for high Reynolds numbers or complex geometries is still necessary. In general, this modelization is based on time averaging of the Navier-Stokes equations (this approach is known as Reynolds-Averaged Navier-Stokes Simulations, RANS). As consequence of the average new unknowns, so-called Reynolds stresses, arise. Different approaches to evaluate them are: i) Differentially Reynolds Stress Models (DRSM), ii) Explicit Algebraic Reynolds Stress Models (EARSM), and iii) Eddy Viscosity Models (EVM).Although EVM models assuming a linear relation between the turbulent stresses and the mean rate of strain tensor are extensively used, they present various limitations. In the last few years, with the even-increasing computational capacity, new proposals to overcome many of these deficiencies have started to find their way. Thus, algebraic or non-linear relations are used to determinate the Reynolds stress tensor without introducing any additional differential equation.Therefore, the first part of this thesis is devoted to the study of several EARSM and EVM models involving linear and higher order terms in the constitutive relation to evaluate turbulent stresses. Accuracy and numerical performance of these models is tested in different flow configurations such as plane channel, backward facing step, and both plane and round impinging jets. Special attention is paid to the verification of the code and numerical solutions, and the validation of the mathematical models used. In the impinging plane configuration, improvements of models using higher order terms in the constitutive relation are limited. Whereas, in the rest of studied cases these non-linear models show a reasonably good behaviour.Moreover, taken into account models convergence, robustness and predictive realism observed in the analysis of these benchmark flows, some of them are selected for the study of air curtains and their interaction with the environment where they are placed. Air curtains are generally one or a set of vertical or horizontal plane jets used as ambient separator of adjacent areas presenting different conditions. The jet acts as a screen against energy losses/gains, moisture or mass exchanges between the areas.As was indicated before, the main purpose of the second part of this thesis is to characterize in detail actual air curtains using both experimental and different numerical approaches. Semi-empirical models to design air curtains are presented. Then, an experimental set-up used to study air curtain discharge and jet downstream is explained. Experimental measurements of velocity and temperature are shown. As a result of the experiments carried out, an improved air curtain with a new design of the discharge nozzle is obtained. Furthermore, air curtain experiments are numerically reproduced and predictions validated against the experimental data acquired. Good agreement between numerical and experimental results is observed.Finally, systematic parametric studies of air curtains in heating and refrigeration applications are done. Global energetic balances are specially considered together with global parameters selected in order to evaluate air curtain performance. It is found that discharge velocity, discharge angle and turbulence intensity of the jet are the most sensitive parameters. Inadequate values for these variables can produce undesirable effects and contribute to increase energy gains/losses.