Rolls-Royce streamlines jet engine design with Isight

  • 18-Jul-2011 02:49 EDT

The design iteration loop for an aircraft engine compressor blade. Process automation and optimization with Isight can be applied at any stage of the cycle.

There is perhaps no more complex engineering challenge than designing a jet engine. The intake, fan, compressor, combustion chamber, turbines, and exhaust must all operate in tandem throughout a vast range of altitude, weather, and temperature conditions. Further complicating the design task is that important, invisible player—aerodynamics.

While it once took up to 10 years to develop a new aircraft engine, the industry average has now shrunk to about two. Rolls-Royce is working to condense that time even further. “Our customers expect consistent performance, fuel efficiency, and short delivery cycles,” said Dierk Otto, Design Systems Engineer at Rolls-Royce Deutschland Ltd. & Co KG. “On our side of the equation we are looking to ensure quality and reliability while keeping design, manufacturing, and maintenance costs low.”

The solution for Rolls-Royce has been its “robust design” program, which emphasizes the leading role of design as the entry point into the company’s Six Sigma program. This focus on the impact of early design on quality has led Rolls-Royce to employ engineering resources from the fields of CAD, CAE, CFD, FEA, and more, in the pursuit of optimum product performance.

Isight, from the SIMULIA brand of Dassault Systèmes, was introduced at Rolls-Royce more than a decade ago and is now employed throughout the company. The software’s drag-and-drop capability for creating simulation process flows (known as workflows) lets engineers link—and automate—all the steps in a particular design process, allowing the simultaneous integration of multidisciplinary simulations.

As an example of the Isight toolkit in action, consider the design of an aircraft engine compressor.

The first compressor task is meanline prediction. Starting from an existing design, the engineer must find the optimum form of the annulus—the donut-shaped area of rotating “blade exits” through which the air flows—for the desired new configuration. The meanline is calculated halfway between the hub and the tip of the annulus. The total area of the annulus determines exit velocity and pressure rise for each stage. This is the principal determinant of the size and cross-sectional layout of the compressor, how many stages it should have and what the inner and outer diameters of each stage will be.

Aerodynamic parameters that have to be taken into account in meanline prediction include pressure ratios, efficiency, surge margins, form factors, and so forth. Running these analyses manually would be immensely time-consuming. But by using Isight, the engineer can integrate all the meanline tasks into an automatic process flow that works through each task sequentially, evaluating the data and applying any relevant external programs, to arrive at the optimum solution.

Although the typical Isight user at Rolls-Royce is an expert who can set up such complex simulation flows quickly, not everyone on the design team needs to work at these deepest levels of expertise. To support these more casual users of Isight, Otto and his colleagues created customized Isight components of calculation routines, data mapping, program controls and bundles, and dynamic link libraries. The casual users can now call upon whatever component they need and drop it into their own simulation flows to speed up and simplify their work, generating the same solutions that an expert would without having to work through all the subroutines.

“This user-friendly component approach gives our team a number of advantages,” said Otto. “Information can be shared easily—even with work sites in other locations—and we can standardize our process build-ups more readily as well as speed up our runtimes.”

The next step in the development of the engine compressor requires the engineering team to move from 1-D meanline analysis to throughflow (optimizing streamlines of predicted air movement inside the annulus). The final step is actual blade design. Starting with 2-D blade geometries (profile sections) that produce the flow angles and conditions predicted by the meanline and throughflow analyses, designers use CFD solvers within Isight to automatically optimize the geometry of every single blade, the cross-section profile of which is the familiar airfoil shape.

By stacking multiple airfoil profile sections on top of each other, connecting them with linear filaments, adjusting the lean of the resulting structure and giving it a root and a platform to sit on, the engineers arrive at the first 3-D shape of a blade. Using an in-house Rolls-Royce blade generator tool, these blade profiles can be modified via different design parameters such as maximum thickness, blade angles, camber style, etc.

Arriving at a CAD model of a blade where the geometry and orientation have been optimized for weight, the designers still need to prove that the blade will survive under real-world conditions. This is where static and dynamic FEA are brought into the optimization loop to perform stress analysis, study vibration and resonance behavior, and examine material creep and lifetime wear. When critical issues are identified, the engineers can return to the 2-D airfoil contours, modify the sections, build them back into a 3-D blade design, and rerun the stress analysis.

This automatic blading design process has a great deal of external I/O data flow, so selecting the correct design parameters is very important to avoid error. The Rolls-Royce team decided to create a series of templates for blading as well. The initialization step was compressed into a template containing important data and instructions for starting the analysis and then connected to a second template that runs the main process. By combining the initialization file and templates with a JAVA program, a complete Isight workflow could be created.

“Using templates for our 3-D blade optimization, much as we saw with components for meanline prediction, helps speed our process build-ups,” said Otto. “While templates reduce error-sources, they are also designed to preserve process flexibility.”

Using components and templates together in a single design iteration loop—modify airfoil, run 2-D CFD calculations, do 3-D optimization with FEA, and finally evaluate results—now takes only 13 min, whereas previously it would need about a day. “We have noticeably less redesign work now, which leads to better control over manufacturing costs,” said Otto.

Lynn Manning, Vice President, Parker Group, wrote this article for Aerospace Engineering.

HTML for Linking to Page
Page URL
Rate It
4.79 Avg. Rating

Related Items

Technical Paper / Journal Article
Training / Education
Training / Education