The twin-screw supercharger uses two intermeshing rotors to trap and move air between the inlet and outlet manifolds compressing the air charge along the flow pathway. The rotors rotate at very high speeds (10000 – 20000 RPM). The automotive twin screw supercharger is of a dry type (i.e. - No lubrication is applied between the rotors. Lubrication is, however, used on the internal gears that spin the rotors). The rotors must not come in contact with each other in order to avoid excessive friction and heat generation. This would lead to rotor seizure and would damage the supercharger. The rotors are manufactured with very tight clearances to overcome the contact problem (i.e. - Nominal clearances around 0.2 mm and reaching less than 0.08 mm at the closest point between the rotors).

The gaps between the meshing rotors lead to axial and radial air leakage during the compression process, known as parasitic losses. This reduces the net output and efficiency of the twin-screw supercharger especially at partial load operation. There is a need to properly identify and quantify the locations with the highest leakage losses and focus on improving the sealing action of the intermeshing rotors at such locations. The use of advanced Computational Fluid Dynamics (CFD) simulation and visualization would give a better insight into the fluid dynamics in the flow pathways and the leakage mechanism between the rotors.

The fluid flow inside the supercharger is highly turbulent and can reach sonic speed across the leakage pathways leading to localized chocking or even supersonic flow conditions. Proper CFD modeling of the highly turbulent flow along the rotor flow pathway, and very small gaps formed by the intermeshing rotors, requires the use of a very fine grid with hundreds of thousands of elements. The high speed of the rotors results in time increment restrictions in the CFD simulation. This translates into significant computation power and storage requirements.

Satisfying these requirements can only be achieved by using a High Performance Computing (HPC) facility with advanced parallel processing and storage capabilities. Current research at RMIT University (Dr. Bassam Abu-Hijleh, Associate Professor Jiyuan Tu, Associate Professor Aleksandar Subic) focuses on the use of state of the art CFD computation for the simulation and optimization of twin-crew supercharger as well as to provide guidelines for future twin-screw designs with enhanced efficiency and performance over the entire operating regime of the engine. To this effect, the current research project has at its disposal a wide range of computing facilities ranging from standard Pentium 4 PCs and a dedicated dual Xeon CPU workstation all the way to a 128 CPU Compaq Alphasever supercomputer courtesy of the Victorian Partnership for Advanced Computing (VPAC).

The project is divided into the CFD simulation of several simplified models of the twin-screw supercharger before attempting to model the operation of an actual twin-screw supercharger. Each model is designed to gain experience in one or more of the many specific areas and expertise needed to be able to fulfil the final goal of full 3D dynamic simulation and visualization of the airflow within a twin-screw supercharger. Models successfully simulated include a stationary 2D version of the twin-screw rotors (see Figure 1 below). This model was used to determine the proper CFD parameters needed to be able to simulate the high-speed compressible flow in the leakage pathways.

Fig. 1a - Velocity contours (m/s) in a 2D twin-screw supercharger. Operating pressure ratio of 3.34 (i.e. - 237 kPa/34.4psi pressure boost). Flow from left to right.	Fig. 1b - Close-up of the velocity contours in the circular region shown in Fig. 1a.

A model of a generic 2D two-rotor dynamic configuration was also simulated successfully (see Figure 2). This model was used to gain the required expertise of using a dynamically changing mesh, the latest in CFD capability available from FLUENT software. In dynamic CFD simulation of a twin-screw supercharger, the mesh used to simulate the fluid flow needs to “adapt” to the changes in the geometry that result from the rotors motion from one time step to another.

Fig. 2a - Dynamically changing mesh with rotors’ motion. Right click here to download movie. (16.9 MB)	Fig. 2b - Pressure contours variation with rotors’ motion. (Note the pressure waves travelling in the inlet and outlet manifolds). Right click here to download movie. (7.54 MB)	Fig. 2c - Velocity contours variation with rotors’ motion. Right click here to download movie. (8.87 MB)

This is not an easy thing to accomplish, as it requires extensive experience in initial mesh generation and is extremely computer intensive. The experience gained from these two simplified models will be combined, currently underway, and then advanced into 3D before working on the full 3D dynamic simulation of a twin-screw supercharger early in 2003.

Stay tuned for more!