Computation of dissipation rates in turbo pumps using different simulation methods

Abstract

The dissipation rate in a flow is a measure of the transition from mechanical flow energy to internal energy and thus represents a loss for the flow. For this reason, it is useful to investigate high dissipation rates when optimizing the flow in turbo pumps, to identify areas where energy is lost in the pump flow. Nowadays, flow optimizations are carried out with the help of numerical flow computations and mostly simulation methods as URANS are used for optimization. The special feature occurs with these unsteady methods that the total dissipation rate in the flow consist of three terms, a direct, a resolved turbulent and a modeled turbulent component. In this context, our study shows how these three terms are defined for an unsteady simulation and how they contribute to the total flow loss in a turbo pump. To reach this goal, a DNS computation of a generic turbulent channel flow and a “quasi-DNS” (qDNS, LES with little turbulence modeling) of an axial turbo pump were performed at first. These simulations built the validation and comparison basis for further URANS computations with stronger turbulence modeling using a k-omega-SST and a omega-based Reynolds stress model. The total dissipation rates between the simulation methods (URANS and DNS) were only quantitatively comparable when all occurring dissipation terms (direct, resolved turbulent and modelled turbulent part) were considered by URANS. The computation of the dissipation rates showed a similar trend for both turbulence models in the turbo pump’s flow. They were able reproduce the volume-integrated, total dissipation rate with a relative deviation of to the qDNS data. Lastly, a phenomenological investigation was carried out to determine, in which flow regions high dissipation rates occur in the turbo pump.