One of the biggest challenges in the gas turbine industry is predicting aeromechanics. This is concerning because the aeromechanical failure of an engine’s blades could damage an aircraft and risk lives.
Engineers could use the geometry of a turbine’s full wheel to create 3D simulations that investigate multirow effects and aeromechanics. However, these wheels contain complicated blade geometries that make the gas turbine simulation computationally expensive and time-consuming.
Some methods can reduce the geometry to a few passages, but they involve approximations that reduce accuracy. These approximations are often not acceptable in aeromechanics because they change the frequency content of the flow, which is critical to calculating the aeromechanical response of blade rows.
To address this challenge, a team consisting of the GUIde consortium, Duke University’s Aeroelasticity Group and the Zucrow labs in Purdue University started a project to simplify the process needed to understand how multirow interactions cause aeromechanical failure in a gas turbine’s compressor. The idea is to reduce the geometry while maintaining the accuracy of a full wheel simulation.
Duke created 3D simulations in the time domain to study the aeromechanics of a compressor rig. These simulations were created using ANSYS CFX and an in-house harmonic balancing code named MUSTANG 2.0. The results from these simulations were then verified experimentally at Purdue.
How to Simplify Gas Turbine Simulations Studying Aeromechanics
Duke took advantage of a couple of model reduction techniques in CFX to simplify the aeromechanical model of the gas turbine simulation.
One technique is the time transformation method which enabled Duke to model the unsteady aerodynamic interactions on specific blade rows. These simplifications use a small part of the compressor’s geometry — rather than the entire a 3.5 stage full-wheel compressor.
The time transformation method’s model reduction capabilities significantly minimized the computational time required to obtain a solution without affecting the accuracy needed to study the physics of each blade row. As a result, Duke could run more cases and investigate solution variation.
The model reduction also minimized the memory requirement to complete the simulation. Thus, less hardware was needed to complete the simulations
Time Transformation Simplifications Yield Highly Accurate Gas Turbine Simulations
Duke conducted a series of studies on the turbine using three, four and five rows and only a few passages modelled for each row. These numerical experimentations enhanced Duke’s understanding of how each row contributed to the forces on the rotor-blade located in the 2nd-row of the compressor.
The accuracy of the solution obtained was systematically improved, compared to data simulated by Duke 5 years ago. In fact, Duke was able to predict modal forces with up to 10% accuracy, and also time-averaged wakes.
To learn of other tricks in CFX, and other ANSYS software, check out ANSYS’ Academic Support Resources.