Abstract This paper describes the aeromechanical aspects of multi-row forced response in embedded compressor rotors building upon previous research by the authors. Forced response in turbomachinery compressor is one of the grand challenge topics and has received a lot of attention in literature recently. The current paper fills a gap in literature by the authors. Previously, the multi-row analysis presented by the authors utilized the time transformation (TT) method within ANSYS CFX to reduce the number of passages in each row to a specific count. This enabled significant savings in computational resources but did not capture the rotor-rotor interaction accurately in computational space. The current paper fills that gap and models full wheel domains which indeed does capture the wave reflection, circumferential variation in pressure field, rotor-rotor interaction and its impact on mistuned forced response. The paper provides a new approach to analyzing and predicting forced response in turbomachinery which includes rotor-rotor interaction, and it provides a validation study by comparing simulations to experimental data. Mistuning in this paper is modeled from a random mistuning standpoint—because of frequency variations due to manufacturing. The computational results are validated with data from the Purdue 3.5stage compressor rig. The first part talks briefly about the unsteady CFD computational approach utilized to obtain the modal force values for each of the multirow cases. These modal forces were utilized to obtain the mistuned forced response predictions. The second part describes in detail the method used to quantify blade responses and goes onto account for a few discrepancies in the mistuned forced response predictions. The last part addresses the cause for these discrepancies and the authors conduct sensitivity studies to justify the anomaly. Some of the conclusions drawn from this study were 1) The rotor-rotor interaction was captured well by numerical methods and in-line with system identification data 2) The response was more localized at the -11ND computationally as against experimental data, which included significant responses at other ND’s although the -11ND content dominated. 3)Including the sidebands in the mistuning model from CFD predictions helped improve the mistuned blade response predictions significantly, in-line with data.
Hegde et al. (Mon,) studied this question.