Numerical and experimental studies of gas/liquid two-phase flow in a turbocharger
The turbocharger is a performance enhancing device which uses the hot exhaust gas mixture from the engine exit to spin the turbine blades for driving the compressor blades to compress the ambient air from the atmosphere. During the operation, the turbine blade surfaces will have severe contaminated deposit accumulations due to the usage of the low-grade diesel fuel for the engine combustion. Hence, this will adversely impact on the turbocharger operating efficiency with some significant performance losses. In this aspect, the development of efficient and cost-effective deposit removal techniques has become one of the urgent tasks for the turbocharger design and manufacture industries. Apparently, an engineering solution to the problem is to apply a built-in “online” water washing system for diesel turbocharger deposits removal. While this method has been used for a while now, the system performance is not entirely satisfactory and further improvements are very demanding. The intention of this study is to review the current know-how industry practice of deposit removal techniques, in particular the “online” water washing technique, in order to understand the underlying mechanisms. The objective is to provide detailed information and flow characteristics for turbocharger designers to improve the system performance. To achieve these targets, a research programme of combining numerical simulation and experimental test has been proposed and carried out on a medium sized diesel turbocharger model to investigate both single-phase gas mixture flow and two-phase gas/liquid flow characteristics. In the latter case, the study is focused on the droplet trajectory, the size distribution and the water droplet coverage area on the blade ring. The experimental tests were based on an in-house test rig at Napier Turbochargers. The measurements were conducted over a wide range of working conditions with and without water washing injections. The test results were used for data analysis and validation of numerical predictions. It was found that in general the water coverage area decreases with the increase of turbocharger loading speed. By using three water injectors evenly distributed in a circumferential direction, maximum ten blades have been washed (i.e. “wetted” blades), while the remaining fourteen blades stayed “dry”. This indicated that more water injectors might be needed to increase the water coverage area on the blade ring. It was also noted that the test rig employed has some limitations of test loading speeds and temperature range, preventing its application to wider operation conditions. The numerical simulations of single-phase gas mixture flow and two-phase gas/liquid flow have been carried out using a commercial computational fluid dynamics (CFD) software ANSYS-CFX with ‘ad hoc’ subroutines of the CFX Command Language (CCL). For a single-phase gas mixture flow, the numerical predicted blade leading-edge temperatures were in good agreement with the experimental measurements with max 1% discrepancies. The results also revealed some influences of three upstream guide vanes on the downstream flow field. For a two-phase flow, the computation has been carried out using the coupled Eulerian/Lagrangian method. Similar to the single-phase flow, simulation results of two-phase flow were generally in good agreement with the experimental observations qualitatively and quantitatively, e.g. the water coverage increases with the decrease of the turbocharger loading speed. For all simulation cases, the water droplet movements in the computational domain have been visualised by the particle trajectory tracking and the 3D iso-surface plotting, etc. A standard conical shaped of water spray cones were clearly seen. Other aspects of the studies were the validation and the assessment of various models adopted in the simulations. The numerical optimisation of the water nozzle configuration was also performed by parametric studies, such as the injector geometry, the injection location and the inclination angle. For a water injector at two inclination angles of 30[sup]o and 90[sup]o against the mainstream flow with different loading speeds, simulation results were in good agreement with the experimental measurements at corresponding test conditions. It was found that for the 90[sup]o inclination angle with the standard nozzle, the number of “wetted” blades was similar to that from the 30[sup]o inclination angle case, but the water coverage area was shifted to the upper half of the blade ring. For the nozzle injection angle at 90[sup]o with a spacer inserted, the numerical predictions have been performed with success and the water coverage has shown some regular patterns as those seen in the standard nozzle simulations. However, surprisingly, this did not entirely agree with the experimental observations, where the water droplet particle distributions on the blade surfaces were randomly distributed without any clear patterns. To understand the underlying cause, some further studies are needed, e.g. the nozzle injection point and the droplet break up model influences. Similar to other numerical simulations, limitations of CFD work are mainly due to the influence of turbulence models and other sub-models used in this study. Further improvement could be done through model refinement study and validation against wider range of test data. In conclusion, the present study has shown that numerical predictions using CFD techniques are generally in good agreement with the experimental measurements in terms of the blade leading-edge temperature distributions and the blade ring water coverage. The flow details such as the water droplet particle trajectory and the size distribution are obtained and they are valuable for the design engineers to improve the current “online” water washing system. Furthermore, the simulation models developed in this study can be used for modelling other conditions that are difficult to perform on the current test rig (mainly due to the safety constraints).