Typical turbine blade design systems are based on parametric studies and engineering experience. Best practices use feedback from manufactured hardware to additionally create producible designs and establish tolerances. Often there is a large gap in understanding of true hardware variation and the ability to describe and analyze it without over simplification. Previously, hardware inspections that feed analyses were limited based on Coordinate Measurement Machines (CMM) point inspection, optical measurements or gages designed to measure specific features. Those details were input into parametric models in an attempt to quantify the impact of hardware variation. Each measurement method has its limitations including accuracy and methodology which can affect the evaluation results. With the advent of 3D scanning, the complex 3D nature of airfoil designs can be described accurately with minimal simplification or assumptions. Actual hardware can be analyzed and assessed for design requirement compliance. In this study, 3D structured light inspection technology is used to assess aerodynamic performance and mechanical durability variation on a statistical sample of actual production hardware.
A Honeywell designed auxiliary power unit second stage uncooled turbine rotor was used as a test case for this study. Thirty-five (35) blade castings randomly selected from three different casting mold lots were scanned using structured light to capture manufacturing process variation. The 3D scanning was completed using controls and robust scanning techniques certified for metrology use. Scans were assessed according to manufacturing requirements and correlated to aerodynamic and mechanical requirements using nominal machining geometry based on datum alignments.
Analyses were performed using STAR-CCM+ for aerodynamic performance and ANSYS for mechanical durability. The designed nominal geometry was used as a baseline for comparison. HEEDS was used to automate the aerodynamic analysis process — replacing the baseline model with the scanned data, merging the new geometry with the base model, meshing, applying boundary conditions, and solving in remote Linux clusters. The automation process achieved a significant reduction in cycle time from several weeks to a few days.
Similarly, the same structured light scan data was used to evaluate mechanical durability in ANSYS. Scan data was aligned to machining datums, boolean cut with nominal CAD machining geometry and directly converted to solid models using SpaceClaim. The solid model was meshed in ANSYS and merged onto the baseline FEA model for solving. Macros were used to automate the analysis process of each part scan and output stresses at critical locations for durability assessment.
Using a combination of analytical tools and actual scan hardware, the impact of manufacturing variation could be understood. Integrating scan data directly to analysis reduces engineering assumptions and gives a better understanding of reality.