It's estimated that about every three seconds an aircraft fitted with Pratt & Whitney Canada (P & WC) engines takes off somewhere in the world. P & WC's mandate to design, develop, and manufacture turboprop, turboshaft, and turbofan engines, auxiliary power units, and other non-aircraft applications of gas turbines means engineers at its Industrial Engine Division. must stay at the forefront of technology.
P&WC has dramatically improved its gas-turbine engine design process using 3D viscous CFD and CSM codes as primary design tools.
By using multidisciplinary computer simulations such as computational fluid dynamics (CFD) and computational structural mechanics (CSM) codes (developed at P & WC) in the design of gas turbine engines, engineers analyze the various interactions occurring between the air and the structure. These analyses produce massive amounts of data and it becomes increasingly difficult to focus on the important details of the analysis. A powerful commercial post-processing software tool called FIELDVIEW, from Intelligent Light (Lyndhurst, NJ), helps engineers quickly grasp the physical problem at hand, cutting design time while continuously improving engine designs.
As one of the world's leading manufacturers of small and medium-size gas turbine engines for the aviation industry, P & WC has delivered more than 45,000 gas turbine engines, which have accumulated more than 305 million hours on more than 18,500 aircraft operated in more than 170 countries. "The turbomachinery design industry is a demanding environment that requires the solution of complex problems with the highest accuracy," says CFD Group Leader Michel Robichaud at Pratt & Whitney Canada (Longueuil, Quebec).
"There is no margin for error when an extra 0.5% performance improvement is required from an already efficient design. The use of 3D viscous CFD and CSM codes is a key to our recent transition from a 5-year design cycle based on a succession of design-test-redesign cycles to a 3-year cycle based on analytical design methodology that eliminates rig testing from the design phase," Robichaud adds.
"One strategic decision at P & WC guides us to control the key analytical tools," says CFD Group Leader Martin Peeters at Pratt & Whitney Canada (Missisauga, Ontario). Making this vision a reality requires establishing strategic alliances with commercial software vendors, and strong industry/university collaboration for basic research, in particular with the CFD Laboratory at Concordia University, directed by Prof. W. G. Habashi.
"The control of these critical technologies allows P & WC to adapt rapidly to the challenges of designing gas turbines in a highly competitive market," says Peeters.
Engineers can create three-dimensional perspective views with hidden-line removal and light shading such as in this low-emission combustor analysis using unstructured grids.
For example, one alliance with a commercial software vendor involves CFD post-processing. P & WC engineers rely on a commercial post-processing program called FIELDVIEW Version 6 from Intelligent Light. The tools in this package allow the engineer to understand the complex flow phenomena that occur in gas turbine engines.
For example, if engineers create three-dimensional perspective views with hidden-line removal and light shading, they can trace the path of a marker traveling along with the fluid through a series of animated views. Engineers can depict an isosurface, a surface running through the points where a numeric value such as fluid velocity is constant, colored with contour plots depicting another numeric value such as temperature. These and the many other visualization techniques the program provides can be combined, such as plotting pressure contours on various surfaces while streamlines illustrate the flow.
"We have developed our own toolkit interface to FIELDVIEW, which makes it possible to use the post-processor directly with our own CFD codes," Robichaud explains. In addition, FIELDVIEW is compatible with most leading commercial CFD solver codes and NASA's PLOT3D format. Available for all major UNIX platforms such as Silicon Graphics, Hewlett Packard, IBM, SUN, and DEC, the program also runs on Intel PCs running Windows NT or 95. "We run the program mostly on Hewlett-Packard C200-C380 workstations with 256 Mbytes of RAM and other Hewlett-Packard, Silicon Graphics, and IBM workstations," says Robichaud.
Engineers can quickly understand complex flow phenomena in gas turbine engines using colors to depict numeric values such as temperature.
"The next major challenge for gas turbine engines is to significantly improve predictive capabilities through the use of multidisciplinary applications," says Peeters. "We are currently developing a wide range of multidisciplinary applications in cooperation with the CFD Laboratory. The ability to take into account the flow unsteadiness caused by the interaction of rotors and stators and diffusers and centrifugal compressors is crucial for gas-turbine design. We have developed a fully implicit unsteady stage scheme for unequal pitch configuration, with considerable impact on performance and durability. This unsteadiness is also a source of vibratory stresses that affect the structural design of blade rows. This has brought us to complement the unsteady stage capability with aero-elastic calculations in which the flow and structure interact, including a 3D viscous aerodynamic damping calculation."
Another very important physical phenomenon in gas turbine engines is heat transfer. "To optimize the life of hot end components, solid and fluid heat transfer are being coupled tightly through conjugate heat-transfer methods," explains Robichaud. "In aero-acoustics, we are coupling the flow and noise propagation in a fully nonlinear fashion that promises substantial reduction of noise from our engines. Finally, Concordia's CFD Laboratory is building an advanced and comprehensive de-icing code to predict ice accretion effects in engine intakes and to optimize bleed air."
Plotting pressure contours helps engineers perform 3D viscous aero-elastic analysis of a fan blade's unsteady pressure.
Development of multidisciplinary applications is pushing the limits of current numerical technology and has revealed areas that require further development, according to Peeters. "We have rapidly found that for complex geometries, mesh generation of multi-block structured grids is a culprit. We have extended our solver to support unstructured grids. Despite its many advantages, a serious drawback of unstructured grid generation is the lack of control on the grid away from surfaces, making them inappropriate for reliable quantitative CFD predictions. Lack of quality meshes is one of the most important sources of CFD inaccuracy and we have successfully focused on the area of anisotropic mesh optimization."
"The complexity of multidisciplinary applications using unstructured adaptive grids obviously requires a powerful and versatile post-processor," Peters continues. "The engineer is only interested in finding among the mass of data from CFD codes how his design is behaving. FIELDVIEW Version 6 has brought significant improvements towards that goal, while maintaining an easy-to-use intuitive interface." P & WC continues to work in partnership with Intelligent Light to further enhance FIELDVIEW capabilities in the context of gas turbine engines.
Engineers can combine visualization techniques such as plotting pressure contours on various surfaces while streamlines illustrate the flow.
One multidisciplinary application example involves the area of aero-elasticity. "Trying to predict the conditions when a fan experiences flutter, a well-known vibrating condition that happens at off-design conditions where the aerodynamics plays a determining role, is a challenge," Robichaud says. "Although this phenomenon is common to all fans, it is critical that it does not occur at any engine operating condition. The conditions at which a fan flutters, known as the flutter boundary, are usually measured through tedious and expensive tests, well after the blade has been designed."
"The challenge," Robichaud continues, "is to numerically predict the flutter boundary, before the design has been released. This first requires a structural analysis to be performed on the blade to identify the primary modes of vibration of the blade susceptible to exhibit flutter. The displacements obtained from each of these primary modes get mapped to a CFD grid that reproduces blade movement induced in each mode. Then we perform a 3D unsteady viscous CFD analysis to predict the reaction of the air to the imposed blade motion."
Engineers use the FIELDVIEW postprocessor to visualize the pressure force contours induced by the blade vibration onto the blade surface. While engineers can visualize the pressure force at an instant in time, only the time history of this force through a full vibration cycle reveals how the aerodynamic response occurs. That's where FIELDVIEW's flipbook capability comes in, to create a movie from a time sequence of these surface plots on the workstation monitor. Incidentally, movies created in this way can also be converted to other formats such as the NTSC and used to create a VHS video. The movie covered 50 time steps at a frequency of 3,000 Hz.
To validate the procedure, engineers performed this flutter analysis at different operating conditions on both sides of the flutter boundary. To measure if the aerodynamics response will dampen or amplify the vibration, the pressure force induced by the blade movement is integrated in time to obtain what is called the aerodynamic damping. "The analyses correlated well with physical experiments and showed positive aerodynamics damping at design condition and negative aerodynamic damping at the condition known to flutter," says Robichaud.
"Multidisciplinary analysis is not an end in itself but a means for achieving further reductions in the engine design cycle," Peeters concludes. "P & WC's vision is to create a virtual engine design system that will integrate all disciplines, from preliminary design to manufacturing, in order to realize the full potential of analytical methods."