Multi-disciplinary, Multi-objective and Multi-point Optimization of a Variable Geometry Turbocharger

DOWNLOAD PDF

Introduction

The goal of this project was to optimize turbine nozzle and rotor blades, in order to maximize the turbine stage efficiency at four different operating conditions with different nozzle setting angles. Another aim was to maintain or improve the Moment of Inertia (MoI), maximum stress and vibration frequencies (first and second mode).

ADT engineers built a Response Surface Model (RSM), using Kriging, based on the design matrix obtained from a number of Design of Experiment (DoE), CFD and FEA calculations. These could then be used to quickly perform optimization using different constraints and objectives.

CFD results showed that the final optimized design provided a 1.2 percentage point improvement in total efficiency, which is defined using different weighing factors at the four different operating conditions. The results showed a 6.6% lower MoI, 2.1% lower blade stress, 8% lower backplate stress, same first mode frequency and 4% lower second mode frequency compared to the baseline, which was already a highly optimized design.

The results showed a 6.6% lower MoI, 2.1% lower blade stress, 8% lower backplate stress, same first mode frequency and 4% lower second mode frequency compared to the baseline, which was already a highly optimized design.

The client required that turbine mass flow had to be fixed for different operating conditions (ER, T01, RPM). Therefore, the nozzle setting angles for any new design had to be varied several times to match the target mass flow for the different operating conditions. An automatic CFD and FEA workflow which integrates 3D Inverse Design code TURBOdesign1 with Ansys Workbench, including CFX and Mechanical, has been developed as shown in Figs. 1 and 2. The nozzle blade geometry can be rotated automatically to meet the mass flow specifications using only four CFD calculations.

 

Automatic TURBOdesign1 and CFX workflow for CFD

 

Automatic TURBOdesign1 and Ansys Mechanical workflow for FEA

 

The nozzle and rotor blades are parameterized using meridional, blade loading, thickness and stacking parameters. In the first step, ADT ran a DoE with linear approximation composed of 30 design cases and varying the initial set of 22 design parameters. The first DoE was used to provide initial sensitivities values for the various design input parameters and select the most relevant ones for the selected performance parameters.

The final optimized design achieved a 1.2 percentage point improvement in total stage efficiency based on the client’s different weighing factors at the four different operating conditions.

Final Results

A second, more detailed, DoE comprising 11 selected design input parameters was run for a total of 80 design cases analysing the aerodynamic and mechanical performances of the entire stage on the four operating points. The surrogate model was initially validated comparing the performance predictions and both CFD/FEA results, showing extraordinary accuracy level.

The final optimized design achieved a 1.2 percentage point improvement in total stage efficiency based on the client’s different weighing factors at the four different operating conditions. It showed a 6.6% lower MOI, 2.1% lower blade stress, 8% backplate stress, same first mode frequency and 4% lower second mode frequency compared to the baseline which as already a highly optimized design.

Get the Latest Updates Direct to Your Inbox