Mean resolvent analysis of periodic flows

Abstract: The mean resolvent operator predicts, in the frequency domain, the mean linear response to forcing, and, as such, it provides the optimal LTI approximation of the input-output dynamics of flows in the statistically steady regime (Leclercq & Sipp 2023). In this paper, we aim at providing numerical frameworks to extract optimal forcings and responses of the mean resolvent, also known as mean resolvent modes. For periodic flows, we rewrite the mean resolvent operator in terms of a harmonic resolvent operator (Wereley & Hall 1990; Padovan & Rowley 2022) to obtain reference mean resolvent modes. Successively, we propose a projection algorithm approximating those modes within a subspace of mean-flow resolvent modes. The projected problem is directly solved in the frequency domain, but we also discuss a time-stepper version that can bypass the explicit construction of the operator without recurring to direct-adjoint looping. We evaluate the algorithms on an incompressible axisymmetric laminar jet periodically forced at the inlet. For a weakly unsteady case, the mean-flow resolvent correctly approximates the main receptivity peak of the mean resolvent, but completely fails to capture a secondary receptivity peak. For a strongly unsteady case, even the main receptivity peak of the mean resolvent is incorrectly captured by the mean-flow resolvent. Although the present algorithms are currently restricted to periodic flows, input projection may be a key ingredient to extend mean resolvent analysis to more complex statistically steady flows.