Presently in the vehicle industry full engine system simulations are performedusing dierent one-dimensional software programs in order to assess the eectof dierent geometrical and part changes on the system as a whole. Thesesimulations are usually fast and multiple parameters can be monitored andanalysed.In this report GT-Power simulations have been performed on a completeengine designed by Volvo Car Corporation. The investigation was performedin order to gain basic knowledge about the internal combustion engine andspecically about the gas exchange system and the turbocharger. A parameterstudy was performed and the responses on the turbine eciency and breaktorque were analysed.The trends in the simulation results follow the background theory well, i.e.increasing the turbine eciency increases the engine eciency and reduces thetime to torque. The eect of the valve opening times and durations on thebreak torque and the turbine eciency can be studied. There is an intricaterelationship where the optimal conguration is dependent on the engine speedas well as the opening angles and times.GT-Power is a very powerful tool for simulating complete engines. Howevercare must be taken when analysing the results. The code only uses one directionand time, meaning that the ow will always be uniform in the cross sections.Whereas in many parts of the real engine the ow eld is three dimensional andfar from uniform in the cross-sections.

In this study comparisons have been performed between gas-stand experiments,Unsteady Reynolds Averaged Navier-Stokes (URANS) simulations, and LargeEddy Simulations (LES) of the ow through a radial turbine of a turbochargerdesigned for an internal combustion engine. The long term goal for the projectis to improve the prediction capabilities for the simpler computational modelsused by industry in the research and development of new products. At thepresent stage the eects of using simplied geometries and methods for turbineperformance predictions are assessed. Additionally, data obtained from URANScalculations is compared against experimental data and against unsteady LESresults. The comparisons are made in order to evaluate the employed methodologies,to know when to use simplied models with reasonable accuracy, as wellas to justify the use of more advanced methods when the models are inadequate.It was found out that the LES results are closer to the gas-stand experimentsthan the URANS predictions are.

In this study the ow out of pipe after a 90pipe bend has been investigated byperforming Large Eddy Simulations. Comparisons of velocity proles, ow eldsand power spectral densities have been performed between numerical simulationsand experiments. Pipe bends exists in many industrial applications especiallyin the internal combustion engine. Performing accurate and predictive pipe owsimulations is becoming more important when more of the development work isbeing performed by simulations. The long term goal of this project is to helpwith development of simpler models that are faster to run but give the sameaccuracy. During this study it was found that the method used works well bothin predicting mean values and in resolving dynamic motions such as vortexswitching. LES calculations without a SGS model has been proven to work wellwith ows of this type as long as the grid resolution is suciently ne.