Session: 15-04: General Internal Cooling

Paper Number: 153089

Impinging Jet Flows – CFD Model Validation Using 3-D Magnetic Resonance Velocimetry Data 

Impinging jet flows are an efficient means for cooling critical turbomachinery components, yielding much higher heat transfer coefficients than convective cooling devices with flow parallel to the hot surfaces. Therefore, many turbomachines use impingement cooling devices. Examples are the internal cooling of turbine blades, radial compressor and turbine wheels and combustor walls.

The design and sizing of cooling devices is a challenging task. It requires a delicate balance between the necessary amount and temperature of the cooling fluid to prevent overheating of critical parts and the efficiency losses associated with branching cooling fluid off the mainstream. Because of the complexity of the task, numerical simulations with computational fluid dynamics (CFD) codes often support the sizing of cooling devices. However, CFD codes contain empirical models describing the influence of turbulence on the mean flow and the heat transfer. Additionally, the mesh topology and size and the discretisation schemes of the code can influence the accuracy of the results. Because of these uncertainties, CFD software needs thorough validation before optimising the design of cooling devices.

Magnetic Resonance Velocimetry (MRV) is a non-optical experimental technique that measures mean velocity fields and turbulence statistics in complex flow geometries. The 3-D measurements are typically conducted in water channels with flow similarity to real-world conditions maintained through the Reynolds number. The MRV technique offers short measurement times and gap-free, high-precision data, making it an ideal tool for efficient CFD validation.

In this study, the authors used MRV data to systematically evaluate the capabilities of modern CFD codes when applied to single-jet and triple-jet impingement flows. Experimental parameters, such as Reynolds numbers and crossflow ratios, were selected to match conditions in cooled turbomachinery components. MRV provided high spatial resolution velocity and Reynolds stress tensor fields across the entire flow domain, producing a valuable dataset for CFD validation. Additionally, the data sets contained detailed information about boundary conditions, such as the inlet flow profiles. A data-matching algorithm was employed to quantitatively compare the experimental data and the CFD results.

In the CFD part of this study, the mathematical model of the flow simulations consisted of the Reynolds-averaged mass, momentum, and energy equations complemented by modern eddy viscosity turbulence models, like the SST two-equation model of Menter (1994), the - -based four-equation models with elliptic blending described by Lardeau and Billard (2016), and an eddy diffusivity model for the turbulent energy fluxes.

Assessing the accuracy of the physical turbulence and heat transfer models is only possible after numerical errors and uncertainties have become negligible. The authors generated a sequence of three scalable meshes to achieve this goal with an effective three-dimensional refinement factor of 1.6 between them. The term “scalable” implies that mesh quality regarding mesh angles and volume ratios remained constant during mesh refinement. The applied formal quality assessment procedure included eliminating iteration errors and quantifying discretisation errors using the Grid Convergence Index (GCI). The results on the fine meshes showed that both turbulence models captured the flow features well, even though the experimental data suggested negative eddy viscosities in parts of the flow field.

Bruschewski, M., Wüstenhagen, C., Domnick, C., Krewinkel. K., Shiau, C.-C., Grundmann, S., Han, J.-C., 2023, “Assessment of the Flow Field and Heat Transfer in a Vane Cooling System Using Magnetic Resonance Velocimetry, Thermochromic Liquid Crystals, and CFD”, ASME Journal of Turbomachinery, Vol. 145, p. 3

Lardeau, S., Billard, F., 2016 “Development of an Elliptic-Blending Lag Model for Industrial Applications”, 54^th AIAA Aerospace Sciences Meeting

Menter, F. R., 1994, “Two-Equation Eddy-Viscosity Turbulence Models for Engineering Applications”, AIAA-Journal, Vol. 32, pp. 1598 – 160

Presenting Author: Thorsten Hansen ISimQ GmbH

Presenting Author Biography: Thorsten Hansen studied Mechanical Engineering at Ruhr-Universität Bochum. After completing university, he joined AEA Technology GmbH, which later became Ansys Germany GmbH, where he worked for 18 years as a customer support team leader specialising in turbomachinery applications. Thorsten and his team have successfully supported several significant turbomachinery and power generation accounts. At ISimQ, Thorsten manages and works on CFD simulation projects, assisting his customers in establishing innovative and efficient simulation workflows.

Authors:

Martin Bruschewski Universität Rostock - Lehrstuhl Strömungsmechanik (LSM)
Sven Grundmann Universität Rostock - Lehrstuhl Strömungsmechanik (LSM)
Thorsten Hansen ISimQ GmbH
Robert Krewinkel Technische Universität Graz - Institut für Thermische Turbomaschinen und Maschinendynamik
Erik Munktell Siemens Industry Software AB
Swantje Romig Universität Rostock - Lehrstuhl Strömungsmechanik (LSM)
Georg Scheuerer ISimQ GmbH
Carolin Wüstenhagen Universität Rostock - Lehrstuhl Strömungsmechanik (LSM)