CFD aids Pulse Turbocharging

Dean Palfreyman, Engineer, Professional Services Department, CD-adapco
Ricardo Martinez-Botas, Thermofluids Section, Mechanical Engineering Department, Imperial College London, UK

Turbocharging is a method of increasing the power output from reciprocating engines by utilizing the waste energy in the exhaust gases. The exhaust gases drive a turbine which provides power to a compressor pressurizing the air at engine inlet, allowing more fuel to be burned. Automotive engines typically use the Pulse Turbocharging method in which the turbine inlet is closely coupled with the exhaust manifold. As a consequence the turbine is subjected to a highly pulsating flow field caused by, and synchronized with, the opening and closing of the engine valves. Figs. 3 and 4 show experimentally measured traces of instantaneous static pressure and mass flow at the inlet to the turbocharger turbine. Note the time scales.

There is a lack of understanding of the turbine aerodynamics under pulsating conditions. This is because of the difficulty in acquiring detailed experimental data for such a highly unsteady flow field and also because of the computational expense associated with predicting the full three-dimensional time-accurate flow within the volute-turbine system. As a result, turbocharger design methods rarely take into account the effect of the pulsating inlet conditions.

Research work over the pastdecade in the Thermofluids Section at Imperial College London has focused on the aerodynamics of turbocharger turbines under pulsating flow conditions. This work has been predominantly experimental but advancement of both STAR-CD and computing resources have made it possible to investigate the turbine performance under pulsating conditions in a time accurate three dimensional manner. A mesh, whose domain encompasses the volute and all turbine passages, has been built employing STAR-CD's moving mesh capability enabling turbine rotation to be modeled. Fig.1 shows a sample mesh. This work has allowed for the first time an assessment of the propagation of the pulse waveforms through the turbine passages and their interaction with the stationary as well as rotating components. The model employs a transmissive boundary condition at the inlet. This permits the application of the instantaneous inlet conditions, whilst allowing the pressure wave rarefactions to propagate out of the domain without (unrealistic) reflection. The second order spatial discretization Total Variation Diminishing (TVD) based scheme, M.A.R.S, was employed since it is particularly suited to capturing the pressure waves (discontinues) in the flow field.

The effect of the pulse waves on turbine performance has been measured at Imperial College, London together with the unsteady velocity field measured using Laser Dopplier Velocimetry (LDV) as shown in figs. 5&6. The predicted results are shown for comparison. As it can be seen, the turbine performance is far from quasi-steady (that is assuming pulse time averaged inlet conditions) and it exhibits a hysteresis type loop due to an imbalance in mass flux entering and leaving the domain (the turbine acts as a restriction). The unusual 'chaotic' efficiency trace is due to the highly disturbed flow field in the turbine caused by the rapidly varying inlet conditions due to the pulse; fig. 6 shows the trace of fluctuating tangential velocity at the turbine inlet.

Comparison between prediction and experiment is good; the velocity field is particularly well resolved and the efficiency trace exhibits the same hysteresis type loop. The experimental data occupies a slightly smaller 'swept' area due to some inertial damping of the turbine instantaneous torque, caused by the coupling of the turbine to the compressor.

In conclusion, the work is testament to a successful collaboration between experimentation and computational studies using a commercially available code. A much more detailed understanding of the highly unsteady nature of the flow in a turbocharger turbine has been achieved than would have been possible by experimental means alone.

Experimental data provided by N. Karamanis, Ph.D. Thesis, Imperial College, University of London.

For more information contact: dean.palfreyman@uk.cd-adapco.com