Robert Francis Kunz

Robert Francis Kunz

Auto-generated profile, based on publicly available information

Robert Francis Kunz

Pennsylvania State University

Research Focus & Metrics

Academic Career Length

24Years (2003-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 December 2019 in Physics of Fluids
Verified
Journal Article
Closed access
PDF file
XHXinyi L. D. Huang
Xinyi L. D. Huang
XYXiang I. A. Yang
Xiang I. A. Yang
RKRobert F. Kunz
Robert F. Kunz
We develop wall modeling capabilities for large-eddy simulations (LESs) of channel flow subjected to spanwise rotation. The developed models are used for flows at various Reynolds numbers and rotation numbers, with different grid resolutions and in differently sized computational domains. We compare a physics-based approach and a data-based machine learning approach. When pursuing a data-based approach, we use the available direct numerical simulation data as our training data. We highlight the difference between LES wall modeling, where one writes all flow quantities in a coordinate defined by the wall-normal direction and the near-wall flow direction, and Reynolds-averaged Navier-Stokes modeling, where one writes flow quantities in tensor forms. Pursuing a physics-based approach, we account for system rotation by reformulating the eddy viscosity in the wall model. Employing the reformulated eddy viscosity, the wall model is able to predict the mean flow correctly. Pursuing a data-based approach, we train a fully connected feed-forward neural network (FNN). The FNN is informed about our knowledge (although limited) on the mean flow. We then use the trained FNNs as wall models in wall modeled LES (WMLES) and show that it predicts the mean flow correctly. While it is not the focus of this study, special attention is paid to the problem of log-layer mismatch, which is common in WMLES. Our study shows that log-layer mismatch, or rather, linear-layer mismatch in WMLES of spanwise rotating channels, is not present at high rotation numbers, even when the wall-model/LES matching location is at the first grid point.

14 June 2021 in Journal of Fluid Mechanics
Verified
Journal Article
Open access
PDF file
HXHaosen H.A. Xu
Haosen H.A. Xu
SASamuel J. Altland
Samuel J. Altland
XYXiang I.A. Yang
Xiang I.A. Yang
RKRobert F. Kunz
Robert F. Kunz
Cube arrays are one of the most extensively studied types of surface roughness, and there has been much research on cubical roughness with low-to-moderate surface coverage densities. In order to help populate the literature of flow over cube arrays with high surface coverage densities, we conduct direct numerical simulations (DNSs) of flow over aligned cube arrays with coverage densities . The roughness are in the d-type roughness regime. Essential flow quantities, including the mean velocity profiles, Reynolds stresses, dispersive stresses and roughness properties, are reported. Special attention is given to secondary turbulent motions in the roughness sublayer. The spanwise-alternating pattern of the thin slots between two neighbouring cubes gives rise to spanwise-alternating regions of low- and high-momentum pathways above the cube crests. We show that the strength and spanwise location of these low- and high-momentum pathways depend on the surface coverage density, and that the high-momentum pathways are not necessarily located directly above the roughness elements. In order to determine the physical processes responsible for the generation and the destruction of these secondary turbulent motions, we analyse the dispersive kinetic energy (DKE) budget. The data shows that the secondary motions get their energy from the DKE-specific production term and the wake production term, and lose energy to the DKE-specific dissipation term.

1 November 2022 in Physics of Fluids
Verified
Journal Article
Open access
PDF file
JLJia-Qi J. L. Li
Jia-Qi J. L. Li
XYXiang I. A. Yang
Xiang I. A. Yang
RKRobert F. Kunz
Robert F. Kunz
Estimates of grid-point and time-step requirements exist for many canonical flows but not for stratified wakes. The purpose of this work is to fill in this gap. We apply the basic meshing principles and estimate the grid-point and time-step requirements for Reynolds-averaged Navier–Stokes (RANS) and large-eddy simulation (LES) of stratified wake flows at high Reynolds numbers, as arise in many geophysical, aircraft, and undersea vehicle systems. Scales representative of a submarine operating in a stably stratified ocean environment are considered, and the quantitative conclusions reached here can be adapted accordingly for particular applications. For a submarine, typical wake conditions are Re0=108 and Fr0=102, and wakes extend to Nt = 1000, where Re0 and Fr0 are the initial Reynolds number and the internal Froude number of the wake, respectively, and N is the buoyancy frequency. We consider both spatially developing and temporally evolving wakes. We show that the grid points required for LES and RANS do not depend on the Reynolds number. The ratio of the grid points needed for LES and RANS is proportional to (Nt2,LW)2/3, where t2,LW marks the end of the late wake and the end of a computational fluid dynamics calculation. According to the present conservative estimates, 0.36×1012 and 0.7×109 grid points are needed for LES and RANS of a spatially developing wake. The numbers are 8×109 and 3×106 for LES and RANS of a temporally evolving wake.

19 April 2021 in Journal of Turbomachinery
Verified
Journal Article
Open access
PDF file
SMStephen T. McClain
Stephen T. McClain
DHDavid R. Hanson
David R. Hanson
ECEmily Cinnamon
Emily Cinnamon
JSJacob C. Snyder
Jacob C. Snyder
RKRobert F. Kunz
Robert F. Kunz
KTKaren A. Thole
Karen A. Thole
Because of the effects of gravity acting on the melt region created during the laser sintering process, additively manufactured surfaces that are pointed upward have been shown to exhibit roughness characteristics different from those seen on surfaces that point downward. For this investigation, the roughness internal flow tunnel (RIFT) and computational fluid dynamics models were used to investigate flow in channels with different roughness on opposing walls of the channel. Three rough surfaces were employed for the investigation. Two of the surfaces were created using scaled, structured-light scans of the upskin and downskin surfaces of an Inconel 718 component which was created at a 45 deg angle to the printing surface and documented by Snyder et al. (2015). A third rough surface was created for the RIFT investigation using a structured-light scan of a surface similar to the Inconel 718 downskin surface, but a different scaling was used to provide larger roughness elements in the RIFT. The resulting roughness dimensions (Rq/Dh) of the three surfaces used were 0.0064, 0.0156, and 0.0405. The friction coefficients were measured over the range of 10,000 < ReDh < 70,000 for each surface opposed by a smooth wall and opposed by each of the other rough walls. At multiple ReDh values, x-array hot-film anemometry was used to characterize the velocity and turbulence profiles for each roughness combination. The friction factor variations for each rough wall opposed by a smooth wall approached complete turbulence. However, when rough surfaces were opposed, the surfaces did not reach complete turbulence over the Reynolds number range investigated. The results of inner variable analysis demonstrate that the roughness function (ΔU+) becomes independent of the roughness condition of the opposing wall providing evidence that Townsend’s hypothesis holds for the relative roughness values expected for additively manufactured turbine blade cooling passages.

8 February 2019 in Journal of Fluids Engineering
Verified
Journal Article
Closed access
PDF file
MKMichael P. Kinzel
Michael P. Kinzel
JLJules W. Lindau
Jules W. Lindau
RKRobert F. Kunz
Robert F. Kunz
This effort investigates advancing cavitation modeling relevant to computational fluid dynamics (CFD) through two strategies. The first aims to reformulate the cavitation models and the second explores adding liquid–vapor slippage effects. The first aspect of the paper revisits cavitation model formulations with respect to the Rayleigh–Plesset equation (RPE). The present approach reformulates the cavitation model using analytic solutions to the RPE. The benefit of this reformulation is displayed by maintaining model sensitivities similar to RPE, whereas the standard models fail these tests. In addition, the model approach is extended beyond standard homogeneous models, to a two-fluid modeling framework that explicitly models the slippage between cavitation bubbles and the liquid. The results indicate a significant impact of slip on the predicted cavitation solution, suggesting that the inclusion of such modeling can potentially improve CFD cavitation models. Overall, the results of this effort point to various aspects that may be considered in future CFD-modeling efforts with the goal of improving the model accuracy and reducing computational time.

14 July 2023 in Journal of Fluids Engineering
Verified
Journal Article
Closed access
PDF file
XYXiang I. A. Yang
Xiang I. A. Yang
WZWen Zhang
Wen Zhang
JYJunlin Yuan
Junlin Yuan
RKRobert F. Kunz
Robert F. Kunz
This work compares various existing rough-wall models on a large collection of rough surfaces with different characteristics and studies the potential of these models in accommodating new datasets. We consider three empirical roughness correlations, two physics-based models, and one data-driven machine-learning model on 68 rough surfaces inside and outside the Roughness Database1. Results show that correlation-type models and machine-learning models do not extrapolate outside the dataset against which they are calibrated or trained. In contrast, the physics-based sheltering model performs well in extrapolation. Recalibrating a roughness correlation against a large dataset proves unfruitful. However, retraining a machine learning model yields good results. We do not pursue further retraining and recalibrating of a physics-based model, as it requires new physical insights. Overall, our findings suggest that a universal rough-wall model is yet to be found. The capability of extrapolation will likely come from incorporating physics. Data, on the other hand, benefits machine learning models.