Validation of the discrete element roughness method for predicting heat transfer on rough surfaces

https://doi.org/10.1016/j.ijheatmasstransfer.2019.03.062Get rights and content

Highlights

  • Discrete Element Roughness (DERM) implementation in the context of a general-purpose CFD code.

  • Benchmarked results for a series of canonical validation cases.

  • Results indicating potential to improve heat transfer prediction on rough surfaces compared to traditional models.

Abstract

The Discrete Element Roughness Method (DERM) is evaluated as an engineering solution to the problem of convective heat transfer on rough surfaces. As part of the present work, DERM is incorporated into a general purpose compressible CFD code and explored as a way of modeling the sub-resolved roughness scales. In addition, DERM-model inputs are evaluated in detail and developed to represent sand-grain roughness (SGR). The results display good agreement in a number of validation cases. The overall results clearly indicate that DERM has potential to improve heat transfer predictions beyond the capability of SGR models, while only slightly increasing the computation time.

Introduction

Surface roughness relates to many applications such as increased head losses through pipes [1], increased convection in pipes [2], [3], heat transfer on roughened turbine blades [4], [5], [6], and ice accretion [7], [8], [9]. The present efforts aim to improve computational fluid dynamics (CFD) models for heat transfer on large-scale roughness elements in the context of CFD codes that can scale to these various applications.

Strategies to include roughness effects in CFD calculations can be divided into three broad categories, listed here in order of increasing computational cost. One approach aims to modify the no-slip boundary condition at the solid walls [11]. These models work well for small roughness (shorter than the mesh height) for conditions where the equivalent sand-grain roughness is known. The second level of modeling aims to account for the roughness within the flow volume. There tend to be two approaches on this front that include modeling roughness via the turbulence model [12] versus modeling through momentum/energy sources. In terms of the eddy-viscosity models, the models tend to focus on altering the turbulence model to predict the developed roughness boundary profile [12], [13], [14]. These approaches are reasonably well suited for conditions where the flow is fully developed over homogeneous roughness. The alternative forms add momentum and energy source in the roughness regions to emulate roughness effects, typically by a variant of the Discrete Element Roughness Method [15] (DERM). Lastly, the roughness shape can be directly resolved, either in two or three dimensions [16], [17], [18]. In addition to an increase in required computational resources, each of these levels of increasing sophistication demands more knowledge about the details of the roughness geometry. Only the equivalent sand-grain roughness (SGR) height is required in order to use a wall correction. Adding source terms may require, depending on the implementation, knowledge of the roughness statistics and how they vary in space. Directly resolving the roughness geometry requires, in two dimensions, some form of surface tracing and, in three dimensions, a surface scan (via computerized tomography (CT), for example). The present work focuses on DERM as it is both computationally efficient and has the potential to effectively model convective heat transfer on surfaces with large-scale roughness.

The basis of DERM is a decomposition of the total wall skin friction drag and convective heat transfer into components due to the presence of the roughness elements and components due to flow over the remaining smooth surface between the roughness elements if it exists. The components due to the elements are then estimated from empirical correlations as functions of the roughness geometry and the local conditions of the flow at the element location. More sophisticated models also account for the flow accelerations due to the reduction in flow area, or blockage, caused by the presence of roughness. This approach is a natural way to incorporate as many geometric properties of the real roughened surface as are necessary to fully describe the discrete elements being used while still allowing the computational mesh to conform to the underlying smooth wall. DERM is also able to naturally capture the different behaviors of smooth, transitionally rough, and fully rough surfaces without explicitly distinguishing between the three.

DERM as an idea was proposed as early as 1936 by Schlichting [19] and has been applied by several investigators to problems of surface roughness on gas turbine blades [10] and geophysical flows [20], [21], for which DERM is a natural way to capture the effects of very large roughness such as tree canopies and buildings. Pioneering work in DERM includes the research of Lewis [22], Taylor [15], [23], McClain [24], and Aupoix [25], [26], among others. Lewis [22] derived the skin friction for flow over ordered transverse ribs by modeling the flow near the ribs as a lid-driven cavity flow and assuming that, far from the wall, only the statistical properties of the roughness are important. Early investigators, such as Adams and Hodge [27], accounted for roughness effects by combining correlations for the form drag on roughness elements with a modification of the turbulence model. Lin and Bywater [28] and Finson [29], [30] included the element blockage as well as the form drag. Christoph and Pletcher [31], [32] further modified a mixing-length turbulence model to account for roughness effects. Taylor [15] derived the turbulent boundary-layer equations to include blockage effects and form drag acting on roughness elements. These investigations also involved tuning a robust functional form for the drag on the individual roughness elements from experimental data that remains in use. Aupoix [25] rigorously derived the DERM boundary-layer equations from consideration of the spatial-averaging theorem [33] rather than the control-volume approach used by Taylor. McClain in the intervening years has made considerable contributions to the DERM field including developing an understanding of how properly to incorporate knowledge of real gas-turbine-blade roughness geometry into the DERM model [24], [34] and how to deal with the temperature gradients that develop along the height of roughness elements [35]. More recently, Aupoix [26] revisited the volume-averaging approach to DERM and identified two new terms in the volume-averaged RANS equations that arise from the interaction between time-averaging and volume-averaging. To date, the primary disadvantage of DERM methods has been their limitation to boundary-layer solvers. To the author’s knowledge, the only implementation of DERM into a general CFD solver for use in aeronautics is that of K. Fleming in 1992 [36].

The present research is focused on improving the heat transfer modeling in CFD as it occurs on accreted ice, which is relevant to aircraft icing. Surface roughness caused by ice accretion is known to cause a significant increase in skin-friction drag [37], [38] due to increased boundary-layer mixing, increased surface area, and the form drag of the individual roughness elements. With respect to glaze icing, the presence of ice roughness impedes the flow of unfrozen surface water [39], potentially modifying the resulting ice shape. Because iced surface roughness tends to occur near the airfoil leading edge, the associated aerodynamics shows several complications beyond those seen in flat-plate studies. The statistical properties of the roughness change in the chord-wise, and possibly also in the span-wise, direction. In addition, the roughness is typically in a region of favorable pressure gradient and accelerating flow [40]. Further, since the boundary layer is very thin near the leading edge, the dimensionless roughness height tends to be correspondingly large. This complicates the application of wall-correction terms to account for the roughness. Although not directly evaluated in this effort, these are the underlying issues that the present work is aiming to address.

The present work aims to extend the application of DERM from boundary-layer codes, to implementation in a general-purpose CFD code relevant to aircraft ice accretion. The work is outlined as follows. The DERM theory is developed and its implementation in the context of a general CFD code is described. Specifically, the full formulation is presented and discussed in the context of its implementation into a CFD code. The resulting implemented model is then validated with respect to a series of canonical validation cases. Using these results, we discuss sensitivities to the model as well as where DERM can be improved in the future.

Section snippets

DERM background

The fundamental principle of DERM is the representation of the roughness in a flow using a sub-grid-scale model, i.e., accounting for roughness without explicitly resolving the roughness features. A sketch depicting the approach is provided in Fig. 1. The real rough surface (left) is approximated using volumetric source terms in the same region, shown in the box at the right. An intermediate step, shown in the middle panel, involves approximating the potentially very complicated real roughness

Flow solver

In this work, the flow solver is based on the commercial CFD software, Star-CCM+ [41]. Star-CCM+ is a comprehensive finite-volume fluid-dynamics simulation package. The spatial discretization schemes used in the present work are formally second-order. While Star-CCM+ is capable of using numerous discretization schemes, flow solvers, and turbulence models, only those used in the present work are mentioned in this section. The present work focused on solving the compressible flow equations in the

Results

Upon implementation of the aforementioned model, validation is required for its application. In the scope of this effort, the method is validated using a series of canonical cases for which there are ample validation experimental data and/or correlations. The goal of these efforts is to establish that the DERM method and implementation are well suited to predicting the impact of roughness on the boundary layer and the resulting heat transfer. In Section 4.1, the CHT-based thermal model of the

DERM accuracy

The validation presented for a variety of cases in Section 4 gives strong evidence that the DERM method can capture the momentum and energy losses due to roughness for a variety of roughness regimes and shapes. The predictions for SGR cases typically lie within 10% of the correlations in skin friction and 20% in heat transfer coefficient. Results in skin friction and heat transfer for large, ordered roughness elements were typically within 20%. Hence, the fundamental DERM model is considered to

Conclusions

In this work, the DERM equations are implemented into a general-purpose CFD flow solver to enable the ability to handle large-scale, complicated roughness. The overall process of implementing DERM is possible but leaves challenges with obtaining the geometric inputs relevant to a DERM model. For this reason, this work also developed the DERM-model inputs relevant to SGR and expanded the understanding of the DERM-model inputs. The DERM implementation was validated using a variety of experimental

Conflict of interest

This work has no conflict of interest.

Acknowledgments

The authors thank Richard E. Kreeger of the NASA Glenn Research Center for his technical guidance and support. This research is partially funded by the Government under Agreement No. W911W6-11-2-0011. The U.S. Government is authorized to reproduce and distribute reprints notwithstanding any copyright notation thereon. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of

References (57)

  • J.P. Bons

    A review of surface roughness effects in gas turbines

    J. Turbomach.

    (2010)
  • R.C. Henry et al.

    Heat transfer variation on protuberances and surface roughness elements

    J. Thermophys. Heat Transf.

    (1995)
  • R.J. Hansman et al.

    Modeling of surface roughness effects on glaze ice accretion

    J. Thermophys. Heat Transf.

    (1991)
  • P.E. Poinsatte, G.J. Vanfossen, K.J. Dewitt, Convective heat transfer measurements from a NACA 0012 airfoil in flight...
  • S.T. McClain, B.K. Hodge, J.P. Bons, Predicting skin friction and heat transfer for turbulent flow over real...
  • Cebeci et al.

    Momentum Transfer in Boundary Layers

    (1977)
  • H. Abdul-Sater et al.

    Predicting radiative heat transfer in oxy-methane flame simulations: an examination of its sensitivities to chemistry and radiative property models

    J. Combust.

    (2015)
  • R. Taylor, A Discrete Element Prediction Approach for Turbulent Flow Over Rough Surfaces (Ph.D. thesis), Mississippi...
  • S.-H. LEE et al.

    Direct numerical simulation of the turbulent boundary layer over a rod-roughened wall

    J. Fluid Mech.

    (2007)
  • D.R. Hanson, M.P. Kinzel, J.P. Palacion, Y. Han, Computational heat transfer modeling of ice-roughened airfoils, in:...
  • D.R. Hanson, Computational Investigation of Convective Heat Transfer on Ice-Roughened Aerodynamic Surfaces (Ph.D....
  • H. Schlichting

    Experimentelle untersuchungen zum rauhigkeitsproblem

    Arch. Appl. Mech.

    (1936)
  • P.W. Cleary et al.

    Discrete-element modeling and smoothed particle hydrodynamics: potential in the environmental sciences

    Phil. Trans. R. Soc. Lond.

    (2004)
  • B.M. Stone et al.

    Hydraulic resistance of flow in channels with cylindrical roughness

    J. Hydraul. Eng.

    (2002)
  • M.J. Lewis

    An elementary analysis for predicting the momentum-and heat-transfer characteristics of a hydraulically rough surface

    J. Heat Transf.

    (1975)
  • S. McClain

    A Discrete-Element Model for Turbulent Flow over Randomly Rough Surfaces

    (2002)
  • B. Aupoix

    Modeling of boundary layer flows over rough surfaces

  • B. Aupoix

    Revisiting the discrete element method for predictions of flows over rough surfaces

    J. Fluids Eng.

    (2016)
  • Cited by (23)

    • Analysis of temperature gradients in thin-walled structures under thermomechanical fatigue loading conditions

      2023, International Journal of Fatigue
      Citation Excerpt :

      Though CFD is advanced in dealing with complex flow fields, it had a poor performance in simulating the flow with rough walls [23]. Hanson [24] improved CFD for heat transfer problems with rough walls by the Discrete Element Roughness Method, but it required complex calculation and the determination of the friction factor of the inner wall, which was difficult to implement. Therefore, an accurate and universal model to acquire the inner boundary is still in need.

    • Optimizing the conception of hybrid PV/PCM by optimizing the heat transfer at the contact interface and by integrating two types of PCM

      2022, Results in Engineering
      Citation Excerpt :

      Some studies have gone through diverse routes, for instance, Al-Najjar et al. [22] proposed mathematical study in order to improve the efficiency of heat transfer by optimizing the design and performance analysis for a PV/PCM system. Nevertheless, the optimization is imperfect as long as the interface between the PV system and PCM system is inhomogeneous and imperfect due to the roughness of the contact surface and the inhomogeneity of material [23,24], because the medium interface can be filled by insulators as the air for example. The imperfect contact decrease the thermal conductivity which decreases heat transfer between components of solar systems that keeps the thermal energy affect the PV system.

    • Experimental characterization of heat transfer enhancement in a circular tube fitted with Koflo Blade™ inline mixer

      2021, Chemical Engineering and Processing - Process Intensification
      Citation Excerpt :

      The simulations showed that increasing the conical strips in addition to the reduction of pitch results in the augmentation of heat transfer rate and flow resistance. Other techniques for heat transfer enhancement have also been addressed in the literature, including treated and rough surfaces [27–29] as well as vortex generators [30–33]. Static or inline mixers are multifunctional motionless devices that are used in a chemical reaction and heat transfer applications in addition to mixing, either individually or simultaneously.

    View all citing articles on Scopus
    View full text