ESDU TN 08008:2009

INTRODUCTION

In this Technical Note, the ESDU CFD validation studies are
presented for the prediction of friction losses and flow
characteristics in straight circular pipes with smooth walls.

Understanding of pipe friction and associated flow
characteristics in different flow regimes is fundamental to the
understanding of boundary layer development and turbulence
modelling in internal flow engineering applications such as viscous
drag and heat-transfer processes.

Typical industrial CFD applications of pipe flow include systems
where pipes are part of a more complex domain, such as aircraft
ducts, pipe fittings, heat exchangers, various components of power
generation plants, etc. In more general CFD applications, straight
long pipes are often used to set inlet and outlet boundary
conditions. At inlet boundaries, they are used as an artificial
extension of the inlet to the investigated flow domain to allow the
natural development of duct boundary layers. At outlet boundaries,
straight pipes are used as an artificial extension to the outlet of
the investigated flow domain to reduce the influence of the outlet
local fluid dynamics on the CFD predictions. In all such
applications, it is essential to reduce the sensitivity of CFD
predictions of friction losses to mesh density and distribution,
pipe length required for fully-developed flow, boundary conditions
and turbulence modelling.

Accurate prediction of the friction losses and flow
characteristics in straight pipes is one of the most challenging
problems in CFD. Although direct numerical simulation (DNS) and
large eddy simulation (LES) are used in research, in industry
Reynolds-averaged Navier-Stokes (RANS) methods are largely
used.

The purpose of this work is to provide guidance on the CFD
modelling of friction losses and flow characteristics in:

• straight circular pipes with smooth walls,

• laminar, transitional and turbulent flow regimes,

• swirl-free, uniform inlet velocity conditions,

• fully-developed exit flow conditions,

• steady-state flow conditions,

• incompressible flow of single-phase, Newtonian liquids
and gases.

• employing commercial CFD packages typically used in the
industry, i.e. finite volume RANS code with
low-Reynolds-number turbulence and transition correlation-based
‘bypass' modelling capabilities.

Extensions to rough pipes and to compressible flow will be
considered in future work.

Three types of meshes were considered: 2D-axisymmetric
hexahedral meshes, 3D purely tetrahedral meshes and 3D tetrahedral
meshes with prismatic layers. Pipes with different lengths were
tested in laminar, transitional and turbulent flow at Reynolds
numbers ranging 1 less than Re less than
10⁷.

Three different turbulence models were considered with different
near-wall treatments: the k – e with SCALABLE near-wall treatment,
k – ? with AUTOMATIC near-wall treatment and SST with AUTOMATIC
near-wall treatment. Three transition models were considered: the
SST specific-?, SST ? and SST ?-Re_? models.
Detailed CFD validation studies on transitional flow are presented
in ESDU TN 08009.

A brief description of the fundamental fluid mechanics in
straight pipes is given in Section 3. In this section,
correlations, measurements and DNS predictions in the literature
for friction losses and flow characteristics in straight pipes are
reviewed. The methodology used in the CFD predictions is described
in Section 4.

The ESDU CFD predictions for the different meshes, pipe lengths,
Reynolds number and turbulence models are discussed in Section 5
and compared to experimental data and correlations in Section 6.
Best Practice Guidelines on mesh density and distribution, setting
of boundary conditions and turbulence modelling are given in
Section 7. Overviews on the CFD modelling of turbulent and
transitional flows are given in Appendices A and B,
respectively.