J. Fluid Mech. (2016), vol. 795, pp. 210–240.

c Cambridge University Press 2016 210

doi:10.1017/jfm.2016.196

Comparison of turbulent boundary layers over

smooth and rough surfaces up to high
Reynolds numbers

D. T. Squire1, †, C. Morrill-Winter1 , N. Hutchins1 , M. P. Schultz2 ,

J. C. Klewicki1,3 and I. Marusic1
1 Department of Mechanical Engineering, University of Melbourne, Melbourne, VIC 3010, Australia
2 Department of Naval Architecture and Ocean Engineering, US Naval Academy, Annapolis,
MD 21402, USA
3 Mechanical Engineering Department, University of New Hampshire, Durham, NH 03824, USA

(Received 28 October 2015; revised 28 January 2016; accepted 9 March 2016)

Turbulent boundary layer measurements above a smooth wall and sandpaper roughness
+
are presented across a wide range of friction Reynolds numbers, δ99 , and equivalent
+ +
sand grain roughness Reynolds numbers, ks (smooth wall: 2020 6 δ99 6 21 430, rough
+ +
wall: 2890 6 δ99 6 29 900; 22 6 ks+ 6 155; and 28 6 δ99 /ks+ 6 199). For the rough-wall
measurements, the mean wall shear stress is determined using a floating element drag
balance. All smooth- and rough-wall data exhibit, over an inertial sublayer, regions
of logarithmic dependence in the mean velocity and streamwise velocity variance.
These logarithmic slopes are apparently the same between smooth and rough walls,
indicating similar dynamics are present in this region. The streamwise mean velocity
defect and skewness profiles each show convincing collapse in the outer region of the
flow, suggesting that Townsend’s (The Structure of Turbulent Shear Flow, vol. 1, 1956,
Cambridge University Press.) wall-similarity hypothesis is a good approximation for
these statistics even at these finite friction Reynolds numbers. Outer-layer collapse is
also observed in the rough-wall streamwise velocity variance, but only for flows with
+
δ99 & 14 000. At Reynolds numbers lower than this, profile invariance is only apparent
+
when the flow is fully rough. In transitionally rough flows at low δ99 , the outer region
of the inner-normalised streamwise velocity variance indicates a dependence on ks+
for the present rough surface.
Key words: turbulent boundary layers, turbulent flows

1. Introduction
Wall-bounded turbulence, particularly at high Reynolds number, is important in
a wide range of practical flows. Three examples (commonly called the canonical
flows) are pipe, channel and zero streamwise pressure gradient boundary layer
flows above smooth walls. While the canonical flows are common in a wide range
of applications, for the majority of practical flows the bounding wall has surface

† Email address for correspondence: [email protected]

 Smooth- and rough-wall boundary layers 211
topography that exerts a dynamical effect on the flow. Herein and in the literature,
surfaces with dynamically significant perturbations are referred to as rough walls.
Rough-wall-bounded flows include, but are by no means limited to, geophysical flows
such as the atmospheric and benthic boundary layer, boundary layers developing
on marine and terrestrial vehicles and flows in piping networks. A characteristic
feature of these applications is the large span of Reynolds numbers they cover.
Therefore, understanding the Reynolds number scaling of rough-wall flows is central
to their accurate prediction. Even in the canonical flow cases, however, sufficient
understanding of the flow dynamics has been elusive despite decades of research.
There is much to be learnt generally about wall-bounded flows from examination of
boundary layers subjected to perturbations such as wall roughness (for example, in
understanding the relative importance, dynamics and interactions of the inner and
outer scales – see Abe, Kawamura & Choi 2004; Toh & Itano 2005; Hutchins &
Marusic 2007a,b).

1.1. Scaling and similarity hypotheses

Roughness manifests over a range of geometries and sizes. It is therefore useful
to have a means for comparing different roughnesses. For most roughness types
– ‘k-type’ roughness (see Perry, Schofield & Joubert 1969) – this is commonly
provided in the form of an equivalent sand grain roughness Reynolds number, ks+ ,
where ks+ = ks Uτ /ν (Uτ is the friction velocity and ν is the kinematic viscosity). As
in smooth-wall flows, the inner-normalised mean streamwise velocity over a rough
wall exhibits a region of log-linear dependence, with the same slope but with the
additive constant being shifted vertically downward by 1U + , the roughness function.
This shift is due to the increased drag of the rough surface; in rough-wall flows there
is a drag contribution from both viscous and pressure forces. Because ks+ depends
on 1U + , it can be used to characterize the rough-wall drag increment relative to the
smooth wall, relating flows of geometrically different roughnesses through this net
effect. At high roughness Reynolds number the viscous component of the surface
drag is negligible compared with the pressure drag of the roughness elements and
the flow is conventionally termed ‘fully rough’. For fully rough flows there is a
log-linear relationship between ks+ and 1U + (Nikuradse 1933). Conversely, at very
low roughness Reynolds numbers, the wall roughness has negligible effect on the
viscous sublayer (1U + → 0), and the flow is ‘dynamically smooth’. At intermediate
roughness Reynolds numbers where the near-wall flow is influenced by both pressure
and viscous drag, the flow is referred to as ‘transitionally rough’. Flows are commonly
considered to be hydraulically smooth for ks+ . 4, and fully rough above ks+ ≈ 70,
although these bounds are only approximate (Ligrani & Moffat 1986; Jiménez 2004).
Experimental data exist for a range of roughness geometries across the transitionally
and fully rough regimes. Reviews of the rough-wall turbulent flow literature are
given by Raupach, Antonia & Rajagopalan (1991) and more recently by Jiménez
(2004). One conclusion of both reviews is that the majority of experimental and
computational evidence provides support for what is commonly referred to as
Townsend’s wall-similarity hypothesis. The hypothesis, termed Reynolds number
similarity by Townsend (1956), implies that at high Reynolds numbers, a ‘fully
turbulent’ flow is unaffected by viscosity, except through the boundary conditions
(a region where the direct influence of viscosity is negligible is congruous with an
inertial region of the flow). On rough-wall-bounded turbulence, Townsend writes:
212 D. T. Squire and others
At distances from the wall large compared with the extent of the flow
patterns set up by individual roughness elements, the turbulent flow is
unlikely to be affected by the exact nature of the roughness and, as with
the smooth wall, it will be determined by the averaged wall stresses, the
channel width [or in boundary layers, δ, the boundary layer thickness] and
the fluid viscosity.
In the literature, numerous roughness studies lend support for Townsend’s
hypothesis in boundary layer (e.g. Raupach 1981; Flack, Schultz & Shapiro 2005;
Volino, Schultz & Flack 2007), pipe (e.g. Shockling, Allen & Smits 2006; Allen
et al. 2007; Hultmark et al. 2013) and channel flows (e.g. Flores & Jimenez 2006;
Hong, Katz & Schultz 2011). There are, however, a number of studies that report a
lack of similarity between smooth- and rough-wall outer-layer flows (e.g. Krogstad,
Antonia & Browne 1992; Tachie, Bergstrom & Balachandar 2000; Keirsbulck et al.
2002; Leonardi et al. 2003; Bhaganagar, Kim & Coleman 2004; Lee & Sung 2007).
Jiménez (2004) asserts that the conflicting findings with regard to Townsend’s
hypothesis may simply be the result of the large relative roughness height, k/δ,
that has been employed in a number of studies. When k/δ is large, the region of the
flow directly influenced by the roughness may occupy a significant fraction of the
boundary layer. This notionally defined region is called the roughness sublayer and
has been suggested to extend a few roughness heights above the roughness (Raupach
et al. 1991; Flack, Schultz & Connelly 2007). Jiménez (2004) concludes that a set of
well-characterized experiments in the fully rough flow regime is required to clarify
the validity of Townsend’s hypothesis. He contends that these experiments should be
carried out in flows in which the equivalent sand grain roughness is high (ks+ > 100),
and the relative roughness height is small (δ/k & 40), suggesting that the friction
Reynolds number, δ + = δUτ /ν, must be at least 4000, for both requirements to
be simultaneously satisfied. Some studies, however, imply that the relative roughness
height must be at least δ/k & 130 for wall similarity to be observed (Efros & Krogstad
2011), shifting the limiting δ + to over 10 000. Note that Jiménez (2004) also compiles
transitionally rough data from previous studies and from this concludes that there is
a similar need for measurements in transitionally rough flows that have a low relative
roughness height. In this paper, data are presented across a wide range of roughness
and friction Reynolds numbers, targeting these regimes identified by Jiménez (2004)
as being sparsely populated by pre-existing data.

1.2. Measuring wall shear stress

A pertinent issue confounding the empirical analysis of rough-wall flows is the
difficulty in accurately estimating the mean wall shear stress, τw . Aside from its
obvious importance in assessing the drag due to the √ rough wall, the wall shear stress
is required to compute the friction velocity, Uτ = τw /ρ, which is a fundamental
scaling parameter in wall-bounded turbulent flows (note that ρ is the fluid density).
The ubiquitous challenges associated with determining τw have been discussed by a
number of researchers (e.g. Perry et al. 1969; Acharya, Bornstein & Escudier 1986;
Brzek et al. 2007; Walker 2014). The vast majority of rough-wall boundary layer
studies rely on similarity in either the inner or outer layer to determine τw . This is
obviously not desirable in studies aiming to assess wall similarity, and is, to a certain
degree, a circular argument. There are only a few notable examples of rough-wall
boundary layer experiments that independently determine τw . Mulhearn & Finnigan
(1978), Acharya et al. (1986) and Krogstad & Efros (2010), for example, use a
 Smooth- and rough-wall boundary layers 213
force balance to directly measure the average wall shear stress on a small sensing
element. Brzek et al. (2007) evaluated the momentum integral equation to determine
the shear stress on a rough wall. However, the use of the momentum integral equation
requires the streamwise evolution of the boundary layer to be documented, which
is uncommon in the literature. In the following we present experimental results
based on Uτ measurements obtained from a purpose built drag balance. As described
recently by Baars et al. (2016), these are capable of higher accuracy than previous
measurements, with an estimated uncertainty of <2.5 %.

1.3. Outline of paper

Streamwise velocity statistics and spectral intensities are presented for rough-wall
+
turbulent boundary layers over the range 3000 . δ99 . 30 000. Herein, δ99 denotes
the wall-normal location at which the mean streamwise velocity is 99 % of the free
stream velocity, U∞ . Only results from hot-wire anemometry (HWA) sensors are
presented in this paper, although particle image velocimetry measurements were also
acquired (to be detailed in a later paper). The measurements are unique in that
they span nearly a decade in Reynolds number while maintaining good spatial and
temporal resolution. Two sets of experiments are presented. The first was taken at
two streamwise locations at which the free stream velocity was varied to generate a
wide Reynolds number range. A subset of these experiments used the force balance
to directly measure τw . These measurements span the transitional and fully rough
regimes, all with a very small relative roughness (δ99 /ks > 120). The second set
of measurements documents the spatial development of the fully rough, rough-wall
boundary layer from low to high Reynolds number over a fetch of nearly 22 m.
Spatially developing rough-wall boundary layer data are obtained at four free stream
velocities. At the highest velocity, the data are all fully rough (ks+ > 100) and
cover a wide range of relative roughness heights (28 < δ99 /ks < 198). Smooth-wall
data (present; Marusic et al. 2015; Morrill-Winter et al. 2015) measured in the
same facility at Reynolds numbers that approximately match those in the present
rough-wall experiments, provide comparisons. Throughout this paper, x, y and z
denote the streamwise, spanwise and wall-normal directions, respectively, with z = 0
located at the roughness crest. Uppercase variables are mean quantities and lowercase
variables denote fluctuations. A ‘+’ superscript indicates inner normalisation (i.e.
using ν and Uτ ).

2. Experimental conditions and procedures

2.1. Facility
The experiments were performed in the High Reynolds Number Boundary Layer
Wind Tunnel (HRNBLWT) at the University of Melbourne. In this open-return
blower facility, a turbulent boundary layer develops on the floor of the 0.92 m ×
1.89 m × 27 m working section. Details of the facility are available in Nickels et al.
(2005, 2007). Measurements were made above a smooth and rough wall, between
1.6 m and 21.7 m downstream of the laminar/turbulent trip. For the smooth-wall
measurements, the tunnel floor was constructed from five polished aluminium plates
with a root-mean-squared surface roughness less than 15 µm (at most 1.2 wall units
for the measurements presented here). For all measurements above the rough wall,
the floor of the working section was covered with sandpaper. Details of the sandpaper
roughness and its installation are provided in § 2.2. The free stream velocity variation
214 D. T. Squire and others

(a) (b)
3
P.d.f.

2
25.4 mm
1
25.4 mm

0
–0.6 –0.3 0 0.3 0.6
–0.4 –0.2 0 0.2 0.4

F IGURE 1. (Colour online) Details of the sand grain surface roughness obtained by
scanning a 25.4 mm × 25.4 mm section of the sandpaper. In (a), the solid black line
shows a p.d.f. of the surface elevation about the mean sand grain height, while the dashed
purple line shows a normal distribution. (b) Shows the scan of the sandpaper section.

along the entire working section was checked at 20, 30 and 40 m s−1 above the
smooth wall, and at 10 and 20 m s−1 above the rough wall and was less than
±0.50 % for all cases. The floating element drag balance is a large flat plate of
dimensions 3 m × 1 m, mounted between 19.5 and 22.5 m downstream of the
laminar/turbulent trip. Four air bearings allow for almost frictionless streamwise
movement of the plate. Because the surface area of the drag balance is large, the
shear force measurement has a uniquely high signal-to-noise ratio (>25). See (Baars
et al. 2016) for further details of the floating element drag balance.

2.2. Surface roughness

For the rough-wall measurements, the entire floor of the wind tunnel working section
was covered with P36 grit sandpaper (SP40F, Awuko Abrasives). With the exception
of the drag balance measurements, all rough-wall measurements were performed
above a single 1.82 m × 28 m sheet of sandpaper. The sheet was affixed to the
tunnel floor at the trip using four spanwise strips of double-sided tape. At the tunnel
exit, the sheet passed over a pulley that spanned the full tunnel width, and was
weighted using two 20 kg free-hanging masses. As such, the sheet was in tension for
the duration of the measurements. This removed undulations in the sandpaper sheet
that formed when the sheet was not in tension. During drag balance measurements,
multiple sandpaper sheets were used to allow for unrestricted movement of the floating
element (see Baars et al. 2016). A Veeco Wyco NT9100 optical profilometer was
used to quantify the surface parameters of the sandpaper over a 25.4 mm × 25.4 mm
area (results given in figure 1b). This apparatus utilizes white light interferometry and
has submicron vertical accuracy. Figure 1(a) shows the probability density function
(p.d.f.) of roughness surface elevation. Note that the distribution of roughness heights
for the sandpaper surface is approximately normally distributed. Here, the physical
roughness height of the sandpaper surface is defined as k = 6σ = 0.902 mm. Key
roughness surface parameters are presented in table 1.
Figure 2 presents the roughness function dependence on the equivalent sand grain
roughness for all of the rough-wall data. The data outlined in black were taken
 Smooth- and rough-wall boundary layers 215

0
101 102

F IGURE 2. (Colour online) The roughness function, 1U + , as a function of the

inner-normalised roughness height, ks+ . Symbols with black outlines show the Case 1
rough-wall data at 21.7 m, for which Uτ is determined directly from the drag balance
measurements; the solid grey line is a smoothing spline fit to a cubic interpolation of the
Case 1 rough-wall data at 21.7 m. The dashed black line shows the fully rough asymptote
of Nikuradse (1933). All symbols are defined in table 2.

Roughness parameter Value Units Formula

p
k 0.902 mm 6 h02
ka 0.119 mm |h0 |
0
kp 1.219 mm max hp − min h0
krms 0.150 mm h02
ksk 0.093 — h /h3rms
03

kku 3.128 — h04 /h4rms

 0
 dh 
ESx 0.482 —  
 dx 
TABLE 1. Key surface parameters from the scanned surface data, where h0 is the surface
deviation about the mean height (h0 = h − h).

above the drag balance, enabling Uτ to be determined directly. The method used to
determine Uτ for the remaining rough-wall data is described in § 2.4. The roughness
function is chosen for each dataset to minimise the least-squares error between the
inner-normalised streamwise velocity profile and the rough-wall logarithmic law,
1
U+ = log (z + )+ + A − 1U + , (2.1)
κ
across the inertial sublayer. The wall-normal bounds of the inertial sublayer and the
wall correction parameter  are discussed in § 2.4. Note, however, that 1U + is largely
insensitive to changes in the start and end locations of the inertial sublayer; changing
either location by ±15 % causes 1U + to change by at most 0.4 %. Nikuradse’s (1933)
equivalent sand grain roughness, ks , is calculated from
1 0
1U + = log ks+ + A − AFR , (2.2)
κ
216 D. T. Squire and others

Case 2
250

200

150 Case 1

100

50

0 50 100 150 200

F IGURE 3. (Colour online) The two rough-wall Cases. The arrows show the direction of
increasing Reynolds number for each Case. All symbols are defined in table 2.

0
with AFR = 8.5 (Nikuradse 1933), and is found to be ks = 1.96 mm. Only data taken
above the drag balance and for which 1U + > 8.0 were used to compute ks . Hence,
ks+ is a roughness Reynolds number which forces the apparently fully rough data in
figure 2 to lie on Nikuradse’s (1933) empirically determined fully rough asymptote
for sand grain roughness. Note that this definition of ks does not represent a physical
roughness height. It is used here as a relative means to indicate the effect of the
roughness on the viscous sublayer.

2.3. Experiments
The rough-wall HWA measurements fall into two regimes: profiles taken over a
range of free stream velocities, but at a fixed streamwise location (15 and 21.7
m) – hereafter referred to as Case 1 measurements – and profiles taken at a fixed
streamwise velocity, but at varying distances from the trip – hereafter referred to
as Case 2 measurements. All experimental conditions for Case 1 and Case 2 are
summarised in table 2. While both the Case 1 and Case 2 data span a similar friction
Reynolds number range, the evolution of the roughness Reynolds number, ks+ , differs
greatly between the two types of development. This is demonstrated in figure 3. Case
1 developments span a range of ks+ values at approximately constant δ99 /ks . Conversely,
+
Case 2 data have approximately constant ks+ with increasing δ99 , but cover a range of
+
δ99 /ks . The arrows in figure 3 show the direction of increasing δ99 for each Case. Note
that all rough-wall data in table 2 are coloured according to their ks+ value; light data
points indicate a low ks+ , while dark data points indicate a high ks+ . A similar summary
of the smooth-wall measurements is provided in table 3. Across all measurements,
data denoted by triangular symbols were obtained using a multi-wire HWA probe,
with all other data gathered using single-wire HWA probes. The multi-wire HWA
measurements employ a probe arrangement similar to that of Foss & Haw (1990),
but the present probe occupies approximately 10 % of their probe’s volume. The
arrangement has four wires and is capable of measuring streamwise and wall-normal
velocity fluctuations, and spanwise vorticity fluctuations. In the present study, only
one wall-parallel wire on the multi-wire probe was employed to obtain streamwise
velocities. For identification purposes, however, these data are denoted ‘multi-wire’
data, whereas those obtained using a single-wire probe are referred to as ‘single-wire’
data.
 Smooth- and rough-wall boundary layers 217
Case 1

Case 2

TABLE 2. Details of the rough-wall experimental data. δ99 is the wall-normal location at
which the mean streamwise velocity is 99 % of U∞ . T̃ = TU∞ /δ99 × 10−3 and Plat. and
Tung. are abbreviations for Platinum and Platinum-coated Tungsten, respectively.

Details of the multi-wire probe geometry, operating conditions and calibration

procedure are provided in Morrill-Winter et al. (2015). All single-wire rough-wall
data were obtained using probes that were operated in constant temperature mode
using a Melbourne University Constant Temperature Anemometer (MUCTA II) of
in-house design. The overheat ratio was chosen to maintain the probe temperature
at approximately 200 ◦ C above ambient temperature. The lowest frequency response
of the system to an external square wave was approximately 18 kHz, occurring at
zero free stream velocity. The single element probes were constructed using Dantec
55P15 boundary layer type prongs with a tip spacing of 1.5 mm. Wollaston or
218 D. T. Squire and others

+
x U∞ Sym. Uτ δ99 δ99 l+ t+ T̃ Mat. Reference
(m) (m s−1 ) (m s−1 ) (mm)
4.5 10.1 A 0.37 87 2 020 11.6 0.29 83.3 Tung. MW15
7 10.1 0.36 120 2 680 11.1 0.27 25.3 Tung. MW15
17.78 4.7 0.17 244 2 690 11.0 0.09 11.6 Plat. Present
4.5 15.2 A 0.54 88 2 990 17.0 0.62 103.8 Tung. MW15
3.75 20.0 E 0.71 71 3 370 25.5 0.51 22.5 Plat. M15
7 15.1 A 0.52 120 3 950 16.4 0.57 37.8 Tung. MW15
6.3 20.0 E 0.69 104 4 760 24.6 0.48 19.2 Plat. M15
18 10.0 A 0.34 265 5 580 10.5 0.24 22.7 Tung. MW15
10 20.2 E 0.69 143 6 450 24.2 0.47 16.9 Plat. M15
18 15.3 A 0.50 261 8 220 15.7 0.52 52.6 Tung. MW15
17.5 20.1 E 0.66 228 9 830 23.3 0.44 15.9 Plat. M15
21.7 30.0 @ 0.93 278 16 960 61.1 2.85 22.7 Plat. Present
21.7 40.8 @ 1.23 270 21 430 79.5 4.90 27.3 Plat. Present
TABLE 3. Details of the smooth-wall experimental data. The abbreviations MW15 and
M15 refer to Morrill-Winter et al. (2015) and Marusic et al. (2015), respectively. Other
abbreviations and units are as given in table 2.

copper-coated platinum-coated tungsten wires were soldered to the prong tips and
etched to expose a 2.5 µm diameter platinum or platinum-coated tungsten filament
of length, l = 0.5 mm. The hot-wire signals were filtered using a low-pass analogue
filter (Frequency Devices 9002), and were sampled using a Data Translation DT9836
Series true 16 bit data-acquisition board. Details of the sampling frequencies and
filter settings for each rough-wall measurement are presented in table 2.
The smooth-wall measurements of Marusic et al. (2015) employed 2.5 µm
diameter platinum filaments. For the three present smooth-wall datasets, 5 µm
diameter filaments with l = 1 mm were operated under the same conditions as
the rough-wall single-wire HWA measurements described above. All single-wire
hot wires were calibrated before and after each boundary layer traverse, and the
atmospheric conditions were monitored throughout. Additionally, the intermediate
single point recalibration technique was used to account for calibration drift during
all measurements (Talluru et al. 2014). A stationary Pitot-static tube pair located in
the free stream constituted the calibration standard for all single-wire measurements.
A third-order polynomial fit to the calibration data was used to convert the hot-wire
voltages to velocities.
The location of the probe relative to the smooth wall was determined using a wall-
normal traversing microscope. The microscope was equipped with a digital micrometer
(CDI BG3600) with a resolution of 1 µm. The distance between the probe and the
wall was determined as the wall-normal distance between the location at which the
probe was in focus and the location at which the wall was in focus. Repeatability
tests suggest that the maximum wall-normal error inherent to this process is ±25 µm.
For the rough-wall measurements, the same procedure was not possible due to the
local heterogeneity of the sandpaper. Instead, a 250 mm × 250 mm aluminium tooling
plate was placed on the surface of the roughness. The location of the probe relative
to the top surface of the plate was determined using the procedure described above.
The plate was subsequently removed and the probe was moved towards the wall by
a distance equal to the thickness of the plate (9.52 mm ± 0.02 mm). Thus, the wall-
 Smooth- and rough-wall boundary layers 219
normal location of the probe above the rough wall was determined relative to the plane
defined by the local roughness peaks (that is, z = 0 is located at the roughness crest).
For the drag balance measurements, the friction force on the floating element was
registered by a load cell and digitally sampled. Data were acquired across the full
range of free stream velocities explored herein. The load cell was calibrated against
free-hanging weights of accurately known mass before and after each measurement,
and a linear curve was fitted to the calibration data. As is demonstrated by Baars et al.
(2016), the variation in the wall drag across the floating element is approximately
linear, meaning that the integrated drag across the element is equal to the local value
at the centre of the element to within 0.024 %. The estimated experimental uncertainty
of the drag balance measurements is ±2.3 % at U∞ ≈ 5 m s−1 , with less error at
higher U∞ . The present rough-wall profiles obtained above the drag balance were
taken at 21.7 m rather than at the streamwise centre of the drag balance, introducing
an additional error at U∞ ≈ 5 m s−1 of 0.2 % (again, with less error at higher U∞ ).
Further details of the drag balance and its operation are given in Baars et al. (2016).

2.4. Friction velocity

The results from the drag balance measurements only directly apply to the rough-wall
data taken at x = 21.7 m. For the remaining data, an alternative approach is required.
One popular method to determine Uτ is to equate the total stress – viscous stress and
Reynolds shear stress – and wall shear stress in the inner part of the boundary layer.
However, this approach demands an accurate measure of a velocity cross-correlation,
uw, and a mean velocity gradient, ∂U/∂z. Aside from the fact that the former is
not obtained for the majority of the measurements presented here (which employ
only a single wire), both statistics are difficult to accurately obtain empirically. It
is therefore common in both smooth- and rough-wall flows, to find Uτ indirectly
using some variation of the Clauser (1956) approach. This involves forcing the mean
velocity to adhere to a predefined logarithmic law. A caveat of this method is that it
requires knowledge of the region over which the logarithmic law is valid (classically,
synonymous with an inertial sublayer).
Wei et al. (2005) formally define the start of the inertial region, zI , by quantifying
the wall-normal location at which the viscous force loses dominant order. By
examining the balance of the terms in the mean momentum equation across the
smooth-wall boundary layer, Wei et al. (2005) show z+ I occurs just exterior to the
peak in the Reynolds shear stress, z+ m (the zero crossing in the Reynolds shear stress
gradient). It is well documented√that above smooth walls the Reynolds shear stress
attains a maximum at z+ m ≈ 1.9 δ
+ (Afzal 1982; Sreenivasan & Sahay 1997; Wei

et al. 2005). Thus, (Wei√et al. 2005) provide analytical evidence that in smooth-wall
flows z+ I scales with δ + . Empirical evidence for this scaling is provided by
Marusic et al. (2013), who demonstrate across a range of high Reynolds number
smooth-wall facilities that log-linearity is observed as the leading-order function in
both the streamwise mean velocity and turbulence √ intensity when the beginning of
the logarithmic region is chosen to scale with δ + (note that log-linearity of the
streamwise turbulence intensity on the inertial subrange is predicted by the attached
eddy hypothesis of Townsend (1976)). Therefore, in the present study the smooth-wall
friction velocity is determined using a Clauser (1956) fit approach with κ = 0.39 and
A = 4.3 across the inertial region limits employed by Marusic et al. (2013). However,
Marusic et al. (2013) use the composite fit of Chauhan, Monkewitz & Nagib (2009) to
determine the boundary layer thickness, whereas here δ99 is defined as the wall-normal
220 D. T. Squire and others
location at which the mean streamwise velocity is 99 % of its free stream value. The
ratio between these two definitions of the boundary layer thickness has only a weak
+
trend with friction Reynolds number; i.e. Reτ ≈ 1.26 × δ99 , where Reτ is computed
+
using composite value, and δ99 uses the present definition. Thus, the inner and outer
bounds of the inertial region used by Marusic p + et al. (2013) are adapted here from
√
z+
I = 3 Re τ and z+
/Re τ = 0.15, to z+
I = 3.4 δ99 and z+
/δ +
99 = 0.19, respectively.
As emphasized by Perry & Joubert (1963), there are two additional unknowns in
the rough-wall formulation of the logarithmic law, namely  and 1U (see (2.1)). The
wall correction parameter, , accounts for the fact that the roughness itself displaces
the entire flow away from the wall. It is therefore dependent on both the flow and
the roughness. Thom (1971) found experimentally, and Jackson (1981) provided
theoretical arguments, that the difference between the roughness height and  (for
them z = 0 is located at the roughness trough) physically represents the mean height
of momentum absorption by the surface. Thus, to correctly measure  the centroid of
the drag profile in the roughness must be calculated; an exercise which is relatively
straightforward when the nature of the flow within the roughness canopy is known
(either via empirical observation – see Jackson (1981) – or via numerical simulation
– see Chan et al. (2015)) but is currently impractical for complex roughnesses. In
the rough-wall literature, it is common to employ the modified Clauser technique
of Perry & Li (1990) which chooses  to maximise the extent of the logarithmic
region of the mean streamwise velocity profile. Such an approach is only appropriate
when there exists no data interior to the onset of the inertial region. However, there
are few studies that examine where inertial dynamics begin to dominate rough-wall
flows. The present study focusses primarily on outer-region rough-wall flow physics
with large δ99 /ks . Thus, the definition of  employed here has negligible effect on the
results, as long as the physical constraint that 0 <  < k is satisfied (Raupach et al.
1991). For simplicity,  = k/2 is used here. However, it is noted that this definition
has no physical justification.
In rough-wall flows, a prevalent assumption is that the flow is inertial starting from
the tops of the roughness crests (for example Perry & Joubert (1963), Perry & Li
(1990) and Schultz & Flack (2003)). As noted by Mehdi, Klewicki & White (2013),
this implies that the positive contribution of the Reynolds shear stress gradient (the
integral of which across the boundary layer is equal to zero) to the mean momentum
equation is confined to the roughness canopy regardless of friction Reynolds number.
Mehdi et al. (2013) examined the balance of terms in the mean momentum equation
for a series of experiments, and concluded that a similar structure existed for both
smooth and rough walls, where for the rough-wall flows the ratio of scales of ν/Uτ ,
δ and k influenced the wall-normal location at which inertial dynamics begin to
dominate. Based on this, Mehdi et al. (2013) proposed z+ m as a surrogate scale for
the onset of the inertial layer, where, since the transition to inertial mean dynamics
is qualitatively the same as for smooth walls, z+ +
I = CI zm . The constant CI is O(1),
+
and zm depends on the three length scales present in rough-wall flows:

zm = C(ν/Uτ )a ksb δ c . (2.3)

For the data examined by Mehdi et al. (2013), three regimes are identified. These
correspond to the ratio of ks to zm being less than, equal to or greater than O(1), with
the constants C, a, b and c empirically estimated for each regime. Visual inspection
across all of the present rough-wall data reveals that the wall-normal location of
the beginning of logarithmic decay in the streamwise turbulence intensity profiles is
 Smooth- and rough-wall boundary layers 221
always significantly exterior to ks . That is, zm /ks > O(1), so the constants C, a, b and
c recommended by Mehdi et al. (2013) in this regime are employed in this paper:

zI = 2.5 × 0.89(ν/Uτ )0.36 ks0 δ99

0.64
. (2.4)

Here CI = 2.5 is assumed, and justification of this value is provided in § 3.3. Note that
C, a, b and c in (2.4) were obtained by Mehdi et al. (2013) by fitting (2.3) to only
a few rough-wall datapoints. Further investigation of these constants is required using
more data that span a wider range of boundary layer parameters (all with zm /ks >
O(1)). As for the outer location of the rough-wall inertial sublayer, this is relatively
inconsequential in the present study and is taken, as for the smooth wall, to be z+ =
+
0.19δ99 (recall that 0.19δ99 ≈ 0.15δc , where δc is computed using the composite fit of
Chauhan et al. (2009)).
For the rough-wall data above the drag balance (Case 1 at 21.7 m), Uτ is
determined from a linear fit of the form Uτ /ν = C1 U∞ /ν + C2 applied to the drag
balance data. To determine Uτ for the remaining rough-wall data, the characteristic
relationship between roughness function and equivalent sand grain roughness (see
figure 2) is employed. This relationship is determined using only data taken above
the drag balance, for which Uτ can be determined directly. Here, a spline fit to a
cubic interpolation of these data (shown by the solid line in figure 2) characterises
the dependence of 1U + on ks+ . The relationship is assumed to be a property of
the roughness and is hence taken to be applicable for all of the present rough-wall
measurements. The wall friction velocities of the remaining rough-wall data are
then determined by forcing these data to lie on the fitted curve, and minimizing
the least-squares error between the inner-normalised streamwise velocity profile and
the rough-wall logarithmic law (2.1) across the inertial sublayer using κ = 0.39 and
A = 4.3. Note that using  = 0 mm and  = k causes the determined Uτ to change by
at most −2.0 % and 2.2 %, respectively, across all rough-wall profiles. However,  has
the most affect when x is small, where the boundary layer is thinnest. Downstream
of x = 4.75 m, Uτ changes by less than 0.8 % when  = 0 mm or  = k mm.

3. Results
3.1. Inner-normalised turbulence statistics
Figure 4 shows the inner-normalised mean streamwise velocity profiles above the
rough wall. Figure 4(a) presents Case 1 profiles at 21.7 m (for which Uτ is determined
directly from the drag balance data), while figure 4(b) shows single-wire Case 2 data
+
at U∞ ≈ 20 m s−1 . In both plots, smooth-wall data at δ99 = 9830 are included for
comparison, and the location of zI (calculated using (2.4)) for a particular profile is
indicated by the symbol with a thick black outline. The latter is true for all remaining
figures in this paper. Recall that the Case 1 data are taken at approximately matched
δ99 /ks , with ks+ ranging from the transitionally rough (ks+ = 22) to fully rough regime
(ks+ = 150) with increasing Reynolds number. Conversely, all Case 2 data are at
approximately matched ks+ , but δ99 /ks ranges with streamwise development from
relatively small (δ99 /ks = 28) to very large (δ99 /ks = 198).
The results in figure 4 show a convincing log-linear region for all profiles, with a
slope very similar to the logarithmic slope of the smooth-wall data; the dashed lines
in figure 4 show a log-linear slope of 1/κ. (Note that the approach used to determine
Uτ for the data shown in figure 4(a) does not assume that there is a logarithmic
profile in the mean streamwise velocity.) The results also show the well-known
222 D. T. Squire and others

30

25

20

15

10

5
(a) (b)
0
101 102 103 101 102 103 104

F IGURE 4. (Colour online) Comparison of streamwise mean velocity between the smooth
and rough walls. (a) Shows the rough-wall data for Case 1 at 21.7 m, (b) shows the
single-wire rough-wall data for Case 2 at U∞ ≈ 20 m s−1 with the data at every second
+
streamwise location removed for clarity. Smooth-wall data at δ99 = 9830 is included in
both plots. The symbols with thick black outlines show the location of the onset of inertial
dynamics according to (2.4). Dashed black lines have a slope of 1/κ = 1/0.39.

result that the rough wall causes a vertical shift, 1U + , in the inner-normalised mean
streamwise velocity due to the increase in drag above the rough wall (relative to a
smooth wall) and the resulting increase in momentum flux towards the wall (Raupach
et al. 1991). In smooth-wall flows the no-slip condition requires that at z = 0, U = 0,
while for rough walls the effective location of the wall varies. Consequently, for the
rough wall flows there is no constraint that U = 0 at z +  = 0. Nonetheless, there is a
location somewhere within the roughness canopy where U = 0. Thus, because of the
downward shift in the rough-wall profiles (1U + ), the near-wall characteristics of the
mean streamwise velocity profile must differ between the smooth and rough walls.
Due to this near-wall modification, most profiles in figure 4 for which 7 . 1U + . 10
appear to be approximately log-linear down to the lowest measured wall-normal
position. Of course, these profiles may contain error near to the wall due to the
choice of  described in § 2.4. However, it seems that such behaviour should be
observed at some 1U + for any sensible choice of . Based upon its classically
defined attributes, the logarithmic region is herein taken to coincide with the inertial
sublayer. Approximate log-linearity of the mean streamwise velocity is thus not
sufficient to define the inertial sublayer. That is, although the mean streamwise
velocity may exhibit log-linearity external to the roughness crest, the wall-normal
location of the beginning of the logarithmic region (or inertial sublayer) may be
further out in the boundary layer. Such behaviour occurs in smooth-wall flow, where
U + is approximately logarithmic interior to zI (see Marusic et al. 2013).
The Reynolds number trends of the streamwise velocity variance are presented in
figure 5. Smooth-wall data are shown in figure 5(a), Case 1 data at 21.7 m are shown
in figure 5(b) and single-wire Case 2 data at U∞ ≈ 20 are presented in figure 5(c).
All data show consistent development trends with increasing friction Reynolds number.
However, the two highest Reynolds number smooth-wall profiles have significantly
poorer spatial resolution (l+ = 61 and 80, respectively) than the other smooth-wall
profiles (l+ < 26). Hutchins et al. (2009) show that in smooth-wall turbulent flows
an increase in l+ causes attenuation of the streamwise variance that is largest near to
the wall, where the contributing scales of motion are relatively small. In this paper,
 Smooth- and rough-wall boundary layers 223

8
(a) (b)

6 (c)

0
101 102 103 104 102 103 104 102 103 104

F IGURE 5. (Colour online) The friction Reynolds number trends of the streamwise
velocity variance above the smooth and rough wall. In (a), the smooth-wall data are
presented with some intermediate Reynolds number data removed for clarity. The thin
grey line shows the boundary layer DNS of Sillero, Jiménez & Moser (2013) and the
thick grey lines show the two highest Reynolds number smooth-wall profiles, corrected
for spatial attenuation using the approach introduced by Smits et al. (2011). (b) Shows
the rough-wall data for Case 1 at 21.7 m, and (c) shows the single-wire rough-wall data
for Case 2 at U∞ ≈ 20 m s−1 with every second streamwise location removed for clarity.
The symbols with thick black outlines show the location of the onset of inertial dynamics
according to (2.4).

the smooth-wall data are employed to make comparisons with rough-wall data in the
outer region of the flow. Such comparisons can lead to specious interpretations when
the effect of spatial resolution is not carefully considered. The thick grey lines in
figure 5(a) show the two highest Reynolds number smooth-wall profiles, corrected for
spatial attenuation affects using the approach of Smits et al. (2011). The corrected
+
data indicate that for the smooth-wall data there is little effect on u2 due to spatial
averaging external to z ≈ zI (the region of interest in this paper). Although it has not
been validated, it is prudent to assume that the correction proposed by Smits et al.
(2011) provides a reasonable estimate of the effect of spatial attenuation in rough-wall
flows, at least away from the near-wall region. It is therefore unlikely that any of the
+
rough-wall u2 profiles presented herein (all with l+ < 40) are discernibly influenced
by spatial attenuation beyond (z + ) ≈ zI .
Comparing figure 5(a–c), the smooth and fully rough variances exhibit very similar
development trends with friction Reynolds number. In accord with Townsend’s (1976)
attached eddy hypothesis, both wall conditions appear to exhibit a region of log-linear
slope (see § 3.3), that lengthens (in wall units) with increasing Reynolds number. At
the lowest Reynolds number in figure 5(b), the rough wall appears to have little
effect on the general shape of the streamwise variance profile, with a near-wall
peak clearly visible at z+ ≈ 15. With increasing friction and roughness Reynolds
+
numbers, the magnitude of the near-wall peak diminishes, such that at δ99 = 29 900
+
and ks = 150, no near-wall peak is apparent. Note however, that the spatial resolution
varies substantially in figure 5(b): from l+ = 5.7 at the lowest friction Reynolds
number to l+ = 38.5 at the highest. Therefore, spatial resolution effects are also
partially responsible for the reduction in the near-wall peak intensity observed with
+
increasing δ99 . However, these effects are not sufficient to account for the complete
+
absence of a near-wall peak at δ99 = 29 900; Hutchins et al. (2009) suggest from
224 D. T. Squire and others

(b) 0.05
(c) 0.050
E
E 0.025
0
(a) 0
25 (d) 0.050

E 0.025
20

0
15 (e) 0.050

10 E 0.025

5 0
10–1 100

0
10–3 10–2 10–1 100

F IGURE 6. (Colour online) (a) The velocity defect (U∞ − U)/Uτ for all smooth- and
+
rough-wall data. The deviation E about the mean defect D as defined in (3.1) is presented
for all of the data in (b), and for the smooth wall, rough-wall Case 1 and rough-wall Case
2 data in (c–e), respectively.

their data that for l+ < 20 the error in the inner-normalised streamwise variance
+
due to spatial averaging should be less than 10 % for δ99 & 3500. The lack of a
near-wall variance peak is well documented in fully rough flows and is associated
with a destruction/disturbance of the near-wall cycle in the immediate vicinity of
the roughness elements (Grass 1971; Schultz & Flack 2007). This disturbance of the
near-wall cycle appears to be a gradual process with increasing ks+ and is likely tied to
the corresponding increase in pressure drag at the wall. While a clear near-wall peak
is observed in the transitionally rough data in figure 5(b), no such peak is observed in
the fully rough data in figure 5(c), even at approximately matched friction Reynolds
number. Interestingly, there appears to be little affect in changing δ99 /ks (which varies
from 28 to 198 in figure 5c) on the general shape of the variance profiles. Note,
however, that for the majority of the rough-wall data considered here, the height of
the roughness relative to the boundary layer thickness would typically be considered
small. This is especially true when compared to some pre-existing measurements in
which outer-layer similarity was not observed. For example, Krogstadt & Antonia
(1999) studied a case with ks /δ99 = 1/7, whereas the largest relative roughness
presented here is ks /δ99 = 1/28. It is worth noting that such empirical studies are
susceptable to errors in determining the virtual origin (that is, determining ) that
can influence the outer region.

3.2. Outer-normalised turbulence statistics

Figure 6 compares the smooth- and rough-wall mean velocity profiles in defect
form. Examination of such plots is a common approach for assessing the validity
of Townsend’s (1956) wall-similarity hypothesis. If the hypothesis is correct, high
 Smooth- and rough-wall boundary layers 225
Reynolds number data normalised in this way should define a single profile in
the outer region regardless of the wall condition. Figure 6(a) shows that, with the
exception of a few low Reynolds number smooth-wall profiles, the velocity defect data
(smooth and rough wall) follow a single outer-layer profile. To probe these results in
greater detail, the deviation of the individual profiles from a mean defect profile is
calculated. This is presented in figure 6(c–e) where the deviation, E, is plotted for
the smooth-wall data, and Case 1 and Case 2 rough-wall data, respectively. Here,
 
 (U − U)+ − D+ 
 ∞ 
E= , (3.1)
 U − 1U
+ + 

+
where D is the mean defect of all of the smooth- and rough-wall profiles. It can be
seen that the maximum deviation observed in any of the profiles in the outer layer is
<5 %. Numerous studies have reported an increase in the wake strength, Π, of rough-
wall flows relative to that for smooth-wall flows (Krogstad et al. 1992; Keirsbulck
et al. 2002; Akinlade et al. 2004; Bergstrom, Akinlade & Tachie 2005; Castro 2007).
Castro, Segalini & Alfredsson (2013) suggests that Π may be a particularly sensitive
measure of whether the outer-region boundary layer structure is truly universal, noting
rough-wall studies in which the outer-layer stress profiles collapse well with those for
smooth walls, but where Π differs. Based on an examination of extant rough-wall
data, Castro et al. (2013) suggest that to ensure outer flow similarity the flow must be
fully rough with δ99 /ks & 11 (note Castro et al. (2013) actually pose their approximate
criterion on δ99 /z0 , where z0 is an alternative measure to ks+ : using κ = 0.39, ks =
27.5z0 ). In figure 6(a) all data satisfy δ99 /ks & 11. Evidence of invariance, however, is
observed even for data that are transitionally rough.
It is important to note that perfect invariance of the mean defect velocity is
not expected even for smooth-wall flows. For example, numerous studies (Erm &
Joubert 1991; Schlatter & Örlü 2012; Marusic et al. 2015) have shown that Π can
be a strong function of x immediately downstream of the trip, with the tripping
strength also having an effect. Only at some distance downstream of the trip does Π
appear to become asymptotically constant (see Marusic et al. 2015). There are also
questions regarding whether Π depends on Reynolds number (Coles 1962; De Graaff
& Eaton 2000; Perry, Marusic & Jones 2002). Therefore, the observed agreement
between the smooth- and rough-wall defect data provides only ostensible support for
Townsend’s wall-similarity hypothesis. Certainly, the deviation E is generally higher
for the smooth-wall data (figure 6c) than for the rough-wall data, particularly at lower
speeds. This may be indicative of some dependence on upstream conditions. It is,
however, encouraging that the profiles of E are very similar between the smooth and
rough walls; suggesting that Townsend’s hypothesis is, at worst, a good approximation
for the mean streamwise velocity defect.
In figure 7 all smooth- and rough-wall turbulence intensity profiles are presented
on an outer-normalised abscissa (linearly and logarithmically spaced in figure 7(a,b),
respectively). All rough- and smooth-wall data apparently merge beyond (z + )/δ99 ≈
+ +
0.5. However, even at this location the data spreads over a range, u2 max − u2 min ≈ 0.7. In
figure 7(c) the highest Reynolds number smooth- and rough-wall data are compared.
Here, excellent agreement between these two profiles is observed external to the
beginning of the inertial sublayer, providing convincing support for Townsend’s
hypothesis, at least for fully rough flows at high friction Reynolds numbers.
Comparisons of smooth- and rough-wall streamwise variance profiles across a wide
226 D. T. Squire and others

6
(a) 9 (b)
4
8
2
7
0 –3
10 10–2 10–1 100
6

0 0.2 0.4 0.6 0.8 1.0 1.2 10–2 10–1 100

F IGURE 7. (Colour online) The streamwise velocity variance for all smooth- and rough-
wall data. A linear abscissa is used in (a) and a logarithmic abscissa is used in (b). In the
+
inset of (b), the highest Reynolds number smooth-wall profile (δ99 = 21 430) is compared
+
to the highest Reynolds number rough-wall profile (δ99 = 29 900).

range of friction and roughness Reynolds numbers (such as in figure 7b) are common
in the literature for assessing the efficacy of Townsend’s hypothesis. However, such
comparisons can be difficult to interpret since the variance has a strong dependence
+
on δ99 . See, for example, figure 8(a) where all inner-normalised smooth-wall variance
profiles are plotted on an outer-normalised abscissa. Disregarding any experimental
error in the data, and acknowledging that any effects of poor spatial resolution are
negligible beyond z/δ99 ≈ 0.02 (see § 3.1), the spread of the profiles internal to
+
z/δ99 ≈ 0.2 in figure 8(a) is evidently due to the inherent trends with δ99 . Returning
+
to figure 7, it appears that data with ks . 70 (lighter shaded symbols) are attenuated
below the fully rough and smooth-wall data. Additionally, the degree of attenuation
apparently depends on the magnitude of ks+ ; with increasing ks+ (darkening symbols)
the transitionally rough data lie increasingly closer to the cluster of fully rough
and smooth profiles. However, for the present experiments, data with a low ks+
+
typically also have a low δ99 . Therefore, in figure 7 it is difficult to isolate trends
+ +
with ks from trends with δ99 . Note that comparison of rough-wall variance profiles
at matched friction Reynolds number but different roughness Reynolds number (such
as the lowest friction Reynolds number profiles in figure 8(b,c), respectively) suggest
+
that the trends described above are related to ks+ , rather than δ99 . This is discussed in
further detail in §§ 3.3 and 3.4.
The streamwise velocity skewness profiles for the Case 1 data at 21.7 m and
single-wire Case 2 data at U∞ ≈ 20 m s−1 are presented in figure 9(a,b), respectively.
For comparison, the highest and lowest friction Reynolds number smooth-wall
skewness profiles are also included in both figures. Note that the skewness is
normalised by the standard deviation cubed and is therefore not subject to errors
in Uτ . Interestingly, the smooth- and rough-wall Case 1 data in figure 9(a) apparently
 Smooth- and rough-wall boundary layers 227

8 (a)
(b)

6 (c)

0
10–3 10–2 10–1 100 10–2 10–1 100 10–2 10–1 100

F IGURE 8. (Colour online) The friction Reynolds number trends of the streamwise
velocity variance above the smooth and rough wall. (a) Shows the smooth-wall data with
some intermediate Reynolds number data removed for clarity, (b) shows the rough-wall
data for Case 1 at 21.7 m, (c) shows the single-wire rough-wall data for Case 2 at
U∞ ≈ 20 m s−1 with every second streamwise location removed for clarity. The symbols
with thick black outlines show the location of the onset of inertial dynamics according to
(2.4).

0 (a) (b)

–1

–2

–3

10–3 10–2 10–1 100 10–2 10–1 100

F IGURE 9. (Colour online) Streamwise skewness profiles for the smooth and rough walls.
(a) Shows the rough-wall data for Case 1 at 21.7 m, (b) shows the single-wire rough-wall
data for Case 2 at U∞ ≈ 20 m s−1 . The highest and lowest Reynolds number smooth-wall
profiles are included in each plot for comparison. Note that data beyond (z + )/δ99 = 1.5
are removed for clarity.

merge beyond (z + )/δ99 ≈ 0.02, suggesting that ks+ has little effect on skewness.
Conversely, in figure 7(b) there is a clear trend in the rough-wall data. Inside of
(z + )/δ99 ≈ 0.2 the rough-wall skewness is larger than that of the smooth wall, with
a greater difference nearer to the wall. The discrepancy between the smooth- and
rough-wall data in this region also appears to decrease with increasing distance from
+
the trip, x (that is, increasing δ99 or increasing δ99 /ks ). Since the range of Reynolds
numbers spanned in figure 9(a,b) is approximately the same, it appears that the
rough-wall skewness trends observed in figure 9(b) result from the changes in δ99 /ks
+
rather than the changes in δ99 . However, this result is opposite to that observed in
figures 5 and 8. Here, changing ks+ appeared to influence the general shape of the
228 D. T. Squire and others
streamwise velocity variance, whereas changing δ99 /ks had seemingly little influence
(at least over the range of values considered).

3.3. The inertial sublayer in rough-wall flows

In wall-bounded flows, there is now considerable support for the existence of a
logarithmic region (or inertial sublayer in physical space). Here, the streamwise
velocity is independent of first-order viscous effects and scales with Uτ regardless
of the characteristic length scale for z (an alternative viewpoint is that z is the
characteristic length scale). In smooth-wall flows, there is theoretical and experimental
evidence
√ that the wall-normal location of the beginning of the inertial sublayer scales
with δ + , the geometric mean of the inner and outer length scales of the flow
(Klewicki, Fife & Wei 2009; Chin et al. 2014). In rough-wall flows, Mehdi et al.
(2013) hypothesise and show evidence that, as in smooth-wall flows, the location of
the onset of inertial dynamics scales with a geometric ratio of the scales present in
the flow, including the length scales imposed by the roughness (see (2.3)).
In what is commonly referred to as the attached eddy hypothesis, Townsend
(1976) theorized that the logarithmic scaling of the wall can be associated with a
three-dimensional distribution of self-similar eddying motions whose sizes scale with
their distance from the wall. The attached eddy hypothesis is independent of viscosity
and consistent with the leading-order mean dynamics in the boundary layer inertial
sublayer. Townsend showed that this description leads to logarithmic profiles in the
streamwise mean velocities, and in the spanwise and streamwise velocity variances.
The latter is shown to have the form:
 
+ z+
u = B1 − A1 log
2 . (3.2)
δ99
Note that the constant A1 is expected to be universal, but B1 is expected to change
with external flow geometry. Marusic et al. (2013) provide empirical evidence for
(3.2) in smooth-wall flows over approximately the same wall-normal domain as the
logarithmic region of the mean velocity profile.
In figure 10 the streamwise mean velocity and turbulence intensity are presented
for all rough-wall profiles. On the abscissa, the wall-normal location, z, is normalised
by the predicted location of the onset of inertial dynamics, zI , aligning (z + ) = zI
for each profile at (z + )/zI = 1. For clarity, the streamwise mean velocities and
turbulence intensities in figure 10 are shifted by their respective values at the location
of the beginning of the inertial sublayer. Using (3.2), it is easy to show
 
+ + z+
u − u |(z+)=zI = A1 log
2 2 . (3.3)
zI
That is, figure 10 is independent of the constant B1 across the inertial sublayer.
The darker points in each profile indicate the region of data contained within the
inertial sublayer (defined here for rough-wall flows as z+ I < z
+ +
< 0.19δ99 ). The
data all show a distinct logarithmic profile in both the streamwise mean velocity and
turbulence intensity, with uniformly good agreement across the full range of Reynolds
numbers and roughness conditions. A similar observation was also made by Hultmark
et al. (2013) in high Reynolds number rough-wall pipe flows, but using a different
normalisation for z. The dashed black lines that bound the data in figure 10(b) have
+ +
a slope of A = 1.26 (Marusic et al. 2013) and show a range in u2 − u2 |z=zI of ±0.2.
 Smooth- and rough-wall boundary layers 229

12

(a)
6

–2

(b)
–4

–6
10–1 100 101

F IGURE 10. (Colour online) The streamwise mean velocity and turbulence intensity with
the abscissa shifted by the estimated location of the onset of inertial dynamics zI , and the
ordinate shifted by its respective value at this location. All rough-wall data are presented.
The dark points in each profile show the data contained within the inertial sublayer
according to the definitions in § 2.4. The dashed black lines that bound the data in 10(b)
+ +
have a slope of A = 1.26 (Marusic et al. 2013) and show a range u2 − u2 |z=zI = ±0.2.

Note that in order to obtain zI from the formulation of Mehdi et al. (2013) a value
for the constant, CI , is required. Here, CI = 2.5 is used, since this yields good collapse
of the data in figure 10. This paper, however, is not concerned with determining an
exact value for CI . The important result here is that good collapse can be obtained
using a fixed value for CI that is O(1).
The results in figure 10 suggest that the formulation proposed by Mehdi et al.
(2013) provides, for the current rough surface conditions (zm > ks+ ), a good estimate
of the wall-normal scale at which inertial dynamics become dominant. Additionally,
the slope of the logarithmic regions of the rough-wall streamwise mean velocities
and turbulence intensities agree within experimental uncertainty to those found above
smooth walls (Marusic et al. 2013). This suggests similarity between the inertial
region dynamics in smooth- and rough-wall flows. The poor collapse of the wake
region of the streamwise velocity variance in figure 10 is, to some extent, due to
the inherent friction Reynolds number trends associated with this statistic. However,
figures 7 and 8 indicate that the outer region of the flow also has a dependence on
+
ks+ at least for transitionally rough surfaces at low δ99 . In figure 11, the constant B1
from (3.2) is presented as a function of friction and roughness Reynolds number. For
each profile, B1 is determined by minimizing the root-mean-square error between that
profile and (3.2) across the inertial region with A1 = 1.26. For the smooth-wall data,
+
B1 appears to become approximately constant for δ99 & 4000. The hatched regions in
figure 11(a,b) show B1 = 2.17 ± 0.2, where B1 = 2.17 was determined for the highest
Reynolds number smooth-wall dataset and the range was chosen to match that shown
230 D. T. Squire and others

(a) (b) (c)

3
2.5

2 2.0

1.5
1
1.0
3
2 140
1 100
0 1 2 3 0 50 100 150 60
0 20

+
F IGURE 11. (Colour online) B1 (see (3.2)) as a function of δ99 and ks+ . The shaded regions
in (a) and (b) show B1 = 2.17 ± 0.2, where B1 = 2.17 was determined for the highest
Reynolds number smooth-wall dataset. A two-dimensional exponential surface is fitted to
the data in (c) to demonstrate its general trend.

by the dashed lines in figure 10(b). That is, for any two datasets with B1 values that
+
agree to within 0.4, u2 (when plotted against (z + )/δ99 ) will exhibit equivalent
collapse across the entire inertial sublayer to that observed in figure 10(b). For the
+
rough-wall data, B1 appears to be a weak function of δ99 and a strong function of
ks+ ; the purple surface in figure 11(c) shows the general trend of the data. At high ks+
+
and δ99 , B1 appears to plateau to the same approximate value as for the smooth wall.
For all flows with B1 ' 2.17, similarity of the inner-normalised streamwise variance
is observed beyond zI regardless of wall condition (accepting, of course, that there
is universality in the wake region: see figure 7). Inertial and outer-layer collapse
of the inner-normalised streamwise velocity variance is ostensibly observed for the
+ +
present roughness geometry for δ99 & 14 000, or independent of δ99 if ks+ & 100. At
low/intermediate friction and roughness Reynolds numbers, however, it appears that
for the present roughness Uτ and δ are not sufficient to scale the streamwise velocity
in the outer region of the flow, and that some representation of the roughness strength
is necessary (e.g. B1 in figures 7 and 8). This is discussed in further detail in the
following section.

3.4. Matched friction Reynolds number comparisons

Figure 12 compares the smooth- and rough-wall streamwise variance and skewness
+
at eight approximately matched friction Reynolds numbers in the range 2900 . δ99 .
22 000. The figures are ordered from top to bottom according to the magnitude of
ks+ . Matched Reynolds number comparisons such as these may avert, to some extent,
the inherent friction Reynolds number trends of smooth- and rough-wall statistics
that can cloud assessments of outer-layer similarity. However, it is difficult to apply
any physical significance to such comparisons, other than that in each case the range
of scales (between viscous and outer) present in the flow is approximately the same
between the smooth and rough wall. Certainly, from a practical standpoint it may
be more informative to compare smooth- and rough-wall boundary layers at matched
Rex (Rex = xU∞ /ν, where x is the distance from the laminar/turbulent trip) since such
comparisons would demonstrate the effect of roughness in developing engineering
flows such as on the hull of a ship or on the wing/fuselage of an aeroplane.
 Smooth- and rough-wall boundary layers 231

(a) (b)
2
(41, 4)
1 (22, 3)

0
8
6
4
2
0
(c) (d)
(68, 6)
2 (49, 5)
1
0
8
6
4
2
0
(e) (89, 7) (f) (97, 7)
2
1
0
8
6
4
2
0
(g) (121, 8) (h) (155, 9)
2
1
0
8
6
4
2
0
10–3 10–2 10–1 100 10–3 10–2 10–1 100

F IGURE 12. (Colour online) Smooth- and rough-wall streamwise variance and skewness
+ +
comparisons at approximately matched friction Reynolds numbers ((a) δ99 ≈ 4000. (b) δ99 ≈
+ + + + +
2900. (c) δ99 ≈ 9800. (d) δ99 ≈ 5400. (e) δ99 ≈ 17 100. ( f ) δ99 ≈ 7900. (g) δ99 ≈ 4700. (h)
+
δ99 ≈ 22 000). The inset in each figure shows the ks+ and 1U + values of the rough-wall
data in that figure. The figures are ordered from low ks+ to high ks+ . Dashed black lines
mark the wall-normal locations of 3ks /δ99 . The symbols with thick black outlines show
the location of the onset of inertial dynamics (zI ) according to (2.4).
232 D. T. Squire and others
The comparisons in figure 12 are presented simply because they reveal interesting
trends in rough-wall-bounded flows. The aim here is to reveal any influence of
roughness Reynolds number on low order rough-wall streamwise statistics in the
absence of known friction Reynolds number trends. Primarily, the focus is on the
outer region of the flow, with emphasis on the accuracy of Townsend’s wall-similarity
hypothesis.
For all the transitionally rough flows (ks+ < 70) in figure 12, apparent differences
in the variance of the streamwise velocity fluctuations exist well into the outer layer
when compared to the smooth wall at a matched Reynolds number. Note that this is
consistent with the observed variations in B1 for these transitionally rough surfaces.
Of course, Townsend’s hypothesis is posed for high Reynolds number flows, and all
+
the transitionally rough cases have δ99 . 10 000. The results, however, indicate that,
as the flow becomes fully rough, similarity in the variance of the streamwise velocity
fluctuations in the outer layer emerges even for cases in which the friction Reynolds
+
number is relatively low. For example, the δ99 ≈ 4700, ks+ = 121 case (figure 12g)
displays similarity in the outer layer with the smooth-wall case at matching Reynolds
number. To the authors’ knowledge, Townsend (1956) hypothesised wall similarity
across all flow conditions, including transitionally rough flows. Certainly, there are
numerous instances in the existing literature where comparisons between smooth and
transitionally rough boundary layers are made and outer-layer similarity is empirically
found to hold (Schultz & Flack 2003, 2007; Wu & Christensen 2007; Allen et al.
2007). The present results tentatively suggest, however, that the outer region of
the inner-normalised streamwise velocity variance profile has a dependence on ks+
across the range of boundary layer and roughness parameters considered here. It
is interesting to note that for the rough-wall data in figure 12(a) the drag balance
measurements indicate an increase in the wall drag coefficient relative to the smooth
wall at matched δ + (presumably due to the pressure drag caused by the roughness
2
elements). If, however, the ordinate of figure 12(a) is normalised by U∞ instead of Uτ2 ,
excellent collapse between the smooth- and rough-wall variance profiles is observed,
indicating that for these data the strength of the streamwise turbulent fluctuations
relative to the free stream velocity appears to be relatively unaffected by the rough
surface. It is possible that this observation, and indeed the general trends with ks+
observed in figure 12, are caused by (or at least augmented by) experimental errors.
For reasons discussed in detail in appendix A, it is not believed that this is the case.
Another observation from figure 12 is that the distance from the wall at which
similarity begins to be observed – the edge of the roughness sublayer – in flows
at high ks+ does not seem to scale in any straightforward manner on k, ks or δ99 .
This result stands in contrast to the findings of Raupach, Thom & Edwards (1980),
Flack et al. (2005) and Krogstad & Efros (2012) who assert that the region directly
influenced by the roughness is confined to within approximately 5k of the wall.
It should be noted, however, that the conclusions of these studies are based on
lower Reynolds number flows in which the scale separation between k and δ99 was
much smaller than in the present case. Furthermore, other studies have documented
rough-wall modifications to the Reynolds stresses that extend deep into the outer layer
(Krogstadt & Antonia 1999; Tachie et al. 2000; Keirsbulck et al. 2002). Based on the
present results, the location at which inertial dynamics emerge, zI , for the rough-wall
flows appears to be a better indicator of where similarity in the variance of the
streamwise velocity fluctuations is observed, but only when ks+ is sufficiently large.
The comparisons of figure 12 are further scrutinised by considering the inner-
normalised premultiplied streamwise turbulent energy spectrograms of selected
 Smooth- and rough-wall boundary layers 233

0 1 2 0 1 2 0 0.2 0.4 0.6

(a) 106
105
104 − =
103
102

(b) 106
105
104 − =
103
102

(c) 106
105
104 − =
103
102

(d) 106
105
104 − =
103
102

(e) 106
105
104 − =
103
102

( f ) 106
105
104 − =
103
102
101 102 103 104 101 102 103 104 101 102 103 104

F IGURE 13. (Colour online) Smooth- and rough-wall premultiplied streamwise energy
+
spectrograms at approximately matched δ99 , and the difference between them. Column 1,
the smooth-wall spectrograms; Column 2, the rough-wall spectrograms; Column 3,
subtraction of the rough-wall spectrograms from the smooth-wall spectrograms. Rows
+
(a–f ) present the spectrograms for most of the comparisons in figure 12 (δ99 ≈
4000, 2900, 17 100, 7900, 4700, 22 000, respectively). Therefore ks+ increases from the
top to the bottom of the figure. In the third column, contour lines are plotted at
kx Φuu /Uτ2 |SW−RW = ±[0.15, 0.3, 0.45, 0.6]. White lines represent positive contours and black
lines represent negative contours. The dashed black vertical lines show the wall-normal
location of ks+ and the solid black vertical lines show zI .
234 D. T. Squire and others
smooth- and rough-wall datasets in figure 13. The same matched Reynolds number
comparisons as those given in figure 12(a,b), (e–h) are presented. The first column
in figure 13 shows the smooth-wall spectrograms; the second column shows the
rough-wall spectrograms and the third column shows the difference between the
smooth- and rough-wall spectrograms. In all comparisons, the Reynolds number
is sufficient to reveal in the smooth-wall data wall-normal separation between the
near-wall (predominantly small scale) energy peak and the logarithmic/outer-region
+
(large scale) energy peak (Hutchins & Marusic 2007b). In comparison (a) (δ99 ≈ 4000
+
and ks = 22) the smooth- and rough-wall spectrograms are qualitatively similar, but
the magnitude of the rough-wall energy at all scales across the boundary layer is
reduced relative to that of the smooth wall. At higher roughness Reynolds numbers,
previous studies have shown that the surface roughness causes a reduction in the
near-wall energy peak relative to that of the smooth wall (Grass 1971; Schultz &
Flack 2007). This observation, however, is not made here due to the lack of resolved
energy in this region. Farther into the boundary layer, figure 13(b–f ) indicates
the differences between the smooth- and rough-wall boundary layers appear to be
primarily in the large scales of motion, i.e. motions that are several δ99 in length (note
that a similar observation was also made by Monty et al. (2011) above a braille-type
wall roughness). The magnitude of the difference between the smooth- and rough-wall
large-scale near-wall motions seems to be related to the friction Reynolds number,
with larger differences observed for flows with higher ks+ . It also appears that the
effect of the rough-wall on the large-scale motions extends farther into the boundary
layer (in wall units) with increasing Reynolds number (recall that figure 13 is ordered
+
vertically according to ks+ , not δ99 ). However, this trend (if real) is weak, and thus
difficult to differentiate from experimental uncertainties. Note that in the third column
of figure 13, it is apparent that zI (black lines) provides a reliable indicator of the
wall-normal location beyond which there is little difference between the smooth- and
+ 0.64
rough-wall spectrograms. It then follows that this location scales with δ99 (as per
+
(2.4)), perhaps explaining the apparent trend with δ99 .

4. Conclusions
A new and unique set of wind tunnel measurements have been used to empirically
examine the streamwise velocity statistics and spectral intensities of turbulent
boundary layers above a randomly distributed rough surface. Additionally, existing and
newly acquired smooth-wall turbulent boundary layer data were employed to provide
comparisons to the rough-wall results. The rough-wall measurements are unique
+
in their range of roughness and boundary layer parameters (2890 6 δ99 6 29 900,
+
26 6 ks 6 155 and 28 6 δ99 /ks 6 199), and in that they incorporate, at one streamwise
location, direct measurements of the mean wall shear stress using a floating element
drag balance. A new approach was introduced to determine Uτ at streamwise locations
where drag balance data was not available. This approach uses the direct wall shear
force measurements to construct the characteristic relationship between ks+ and 1U + .
It is assumed that this relationship is universal for a particular roughness, allowing
Uτ to be estimated for data taken at any streamwise location and any free stream
velocity. All data show self-consistent trends with friction and roughness Reynolds
number.
The formulation proposed by Mehdi et al. (2013) was employed to estimate the
wall-normal location of the onset of inertial mean dynamics, zI , in rough-wall flows.
Aligning all rough-wall mean streamwise velocity profiles at this location revealed
 Smooth- and rough-wall boundary layers 235
a distinct logarithmic region with constant slope (κ) across all rough-wall profiles,
extending from zI to (z + )/δ99 ≈ 0.19. Such a region is consistent with an inertial
sublayer of the flow, where the streamwise velocity is independent of first-order
viscous effects. A similar log-linear region was revealed in the streamwise velocity
variance across the same approximate bounds, as predicted by the attached eddy
hypothesis of Townsend (1976) (which is consistent with an inertial layer description).
As was observed for the mean streamwise velocity, the slope (A1 ) of this region was
approximately constant across all measurements. Thus, we provide evidence that the
formulation of Mehdi et al. (2013) well predicts the wall-normal location of the
onset of inertial dynamics in rough-wall flows. Additionally, the rough wall κ and
A1 values agree to within experimental uncertainty with those recently determined for
smooth-wall flows using high friction Reynolds number data from a range of facilities
(Marusic et al. 2013), suggesting similarity between the inertial region dynamics in
smooth- and rough-wall flows.
Examination of the rough-wall mean streamwise velocity defect suggests that
Townsend’s (1956) wall-similarity hypothesis is an excellent approximation for this
statistic across the full range of roughness and boundary layer parameters examined.
Similarly, in the streamwise velocity skewness, good collapse is observed across all
rough- and smooth-wall data beyond (z + )/δ99 ≈ 0.2. In the streamwise velocity
+
variance outer-layer collapse is observed for all flows with δ99 & 14 000. At low ks+
+
and intermediate δ99 , however, the outer region of the inner-normalised streamwise
velocity variance seems to have some dependence on ks+ . At very low ks+ (ks+ . 30), the
rough-wall streamwise velocity variance is qualitatively similar to that in smooth-wall
+
flow at matched δ99 , but is attenuated across the entire boundary layer. Comparison
of the rough- and smooth-wall spectrograms indicate that this attenuation acts equally
at all wavelengths and wall-normal locations. The level of attenuation appears to
decrease with increasing ks+ , such that, for fully rough flow, outer-layer collapse is
+
observed for all available δ99 . That is, Townsend’s hypothesis also appears to be valid
+
across all δ99 so long as the flow is fully rough. In this regime, zI appears to predict
the edge of the roughness sublayer, indicating similarity across the entire inertial
region of the flow. At moderate and high ks+ , the energy associated with near-wall
streamwise motions is different between rough and smooth walls at matched friction
Reynolds numbers. Much of the difference appears to reside in the large scales of
motion (λ+ = O(δ99 )), with smaller scales (even those of the order of the roughness
height, k) apparently less affected. The magnitude of the difference appears to be
related to ks+ , and the inner-normalised distance that this difference extends into the
+
boundary layer seems to be a function of δ99 . A physical explanation for the trends
described above is not yet known. The current measurements do not obtain data with
+
low ks+ and high δ99 , which would be very valuable in elucidating the present results.

Acknowledgements
The authors wish to thank the Australian Research Council for the financial support
of this research. M.P.S. would like to thank the U.S. Office of Naval Research for
supporting his sabbatical visit to the University of Melbourne.

Appendix A. Are the trends with ks+ real?

The present results indicate that ks+ may play a larger role in outer-region rough-wall
boundary layer dynamics than previous studies have indicated. Schultz & Flack (2007),
for example, report outer-layer collapse of the streamwise velocity variance for flows
236 D. T. Squire and others

Case 1 Case 2

5
0
–5

2
0
–2

Measurement

F IGURE 14. (Colour online) The percentage difference between Uτ determined using
the technique in § 2.4 (i.e. using the drag balance measurements), and that determined
using an independent method. ‘MC’ refers to the modified Clauser (1956) technique,
and ‘TSS’ refers to the total shear stress method. The latter is only possible for the
multi-wire measurements, for which uw profiles are available. Recall that the circular
symbols represent data taken above the drag balance, for which Uτ is determined directly.
The data are listed along the abscissa in the same order that they are presented in table 2.
The hatched regions show 1Uτ ,MC = ±5 % and 1Uτ ,TSS = ±2 %.

ranging from hydraulically smooth to fully rough. It is possible that the apparent
importance of ks+ observed in the present study results from errors in determining
wall friction velocity, or from changes in the tripping condition or streamwise pressure
gradient effects at low speeds (which were required in the present study to generate
flows with low ks+ ). The following sections explain why it is difficult to use these
explanations to refute the poor outer-layer collapse observed in the inner-normalised
streamwise velocity variance at low ks+ .

A.1. Wall friction velocity

The rough-wall data with ks+ = 22show the largest difference in the outer region of the
+
inner-normalised streamwise velocity variance relative to a matched δ99 smooth-wall
profile (see figure 12a). These data are normalised by a wall friction velocity that
was determined using direct drag balance measurements. While the uncertainty in the
drag balance measurements certainly increases at the lower speeds required to generate
low roughness Reynolds numbers (the rough-wall data in figure 12(a) is obtained at
U∞ = 4.7 m s−1 ), at U∞ ≈ 5 m s−1 the estimated error in Uτ is only ±2.3 % (for a
detailed discussion of the errors associated with the drag balance measurements see
Baars et al. (2016)). It is possible to rescale the rough-wall profile in figure 12(a)
+
to obtain outer-layer collapse with a matched δ99 smooth-wall profile. However, this
+
requires a change in Uτ of approximately 9 % (note that changing Uτ changes δ99 ,
which in turn requires comparison to a different smooth-wall dataset). Additionally,
collapse of the streamwise mean velocity to the rough-wall log-law (2.1) using the
adjusted value for Uτ is then compromised unless κ is changed by approximately
12 %. Furthermore, figure 14 shows the percentage difference between Uτ determined
using the process in § 2.4, and that determined from two independent approaches. The
first approach – hereafter called the modified Clauser (MC) method – determines Uτ
and 1U + to maximise agreement between the inner-normalised streamwise velocity
and (2.1) across the inertial subrange. In the second approach – hereafter called the
 Smooth- and rough-wall boundary layers 237
total shear stress (TSS) method – it is assumed that a nominally constant shear stress
region exists in the inner part of the boundary layer which is equal to the wall shear
stress. Therefore, Uτ can be calculated in this region using
r
∂U
Uτ ' ν − uw. (A 1)
∂z
Note that both the MC and TSS methods are common in the literature. Of course, the
TSS method is only possible when uw profiles are available and hence is only used
here for the multi-wire measurements. The Uτ values determined using the present
approach (i.e. using the drag balance measurements – see § 2.4) agree with the MC
method to within 5.3 %, and with the TSS method to within 1.9 % across all available
rough-wall measurements. The agreement conveys little information regarding the error
associated with each method (all three approaches have inherent errors). However, it
is unlikely that three completely disassociated methods should produce such similar
estimates of Uτ if those estimates are incorrect.

A.2. Low U∞
A favourable streamwise pressure gradient or poorly stimulated boundary layer can
exhibit inner-normalised streamwise velocity fluctuations that are attenuated relative to
the zero pressure gradient equilibrium case (Harun et al. 2013; Marusic et al. 2015).
In such flows the wake strength of the mean streamwise velocity is also significantly
affected. In the present study, the streamwise pressure gradient in the wind tunnel
was not assessed at the lowest operating speed (U∞ ≈ 5). Additionally, it is possible
that the smooth-wall trip (which generates canonical boundary layer evolution at
U∞ = 10–20 m s−1 ) understimulates the boundary layer at low speeds. However, in
figure 6, excellent collapse of the streamwise velocity defect across all smooth- and
rough-wall datasets is observed, indicating that there is little difference in the wake
strength across all data presented, including the rough-wall profiles at low ks+ (and
U∞ ). Additionally, smooth-wall data taken at U∞ = 4.7 m s−1 are included in the
comparisons in the present paper. These data agree well with smooth-wall data taken
+
approximately matched δ99 but higher U∞ (see figures 5 and 6), suggesting that
equilibrium flow is obtained in the wind tunnel even at these low operating speeds.

REFERENCES

A BE , H., K AWAMURA , H. & C HOI , H. 2004 Very large-scale structures and their effects on the wall
shear-stress fluctuations in a turbulent channel flow up to Reτ = 640. Trans. ASME J. Fluids
Engng 126 (5), 835–843.
ACHARYA , M., B ORNSTEIN , J. & E SCUDIER , M. P. 1986 Turbulent boundary layers on rough
surfaces. Exp. Fluids 4 (1), 33–47.
A FZAL , N. 1982 Fully developed turbulent flow in a pipe: an intermediate layer. Ing.-Arch. 52 (6),
355–377.
A KINLADE , O. G., B ERGSTROM , D. J., TACHIE , M. F. & C ASTILLO , L. 2004 Outer flow scaling
of smooth and rough wall turbulent boundary layers. Exp. Fluids 37, 604–612.
A LLEN , J. J., S HOCKLING , M. A., K UNKEL , G. J. & S MITS , A. J. 2007 Turbulent flow in smooth
and rough pipes. Phil. Trans. R. Soc. Lond. A 365 (1852), 699–714.
BAARS , W. J., S QUIRE , D. T., TALLURU , K. M., A BBASSI , M. R., H UTCHINS , N. & M ARUSIC , I.
2016 Wall-drag measurements of smooth- and rough-wall turbulent boundary layers using a
floating element. Exp. Fluids (under review).
238 D. T. Squire and others
B ERGSTROM , D. J., A KINLADE , O. G. & TACHIE , M. F. 2005 Skin friction correlation for smooth
and rough wall turbulent boundary layers. Trans. ASME J. Fluids Engng 127 (6), 1146–1153.
B HAGANAGAR , K., K IM , J. & C OLEMAN , G. 2004 Effect of roughness on wall-bounded turbulence.
Flow Turbul. Combust. 72 (2–4), 463–492.
B RZEK , B., C AL , R. B., J OHANSSON , G. & C ASTILLO , L. 2007 Inner and outer scalings in rough
surface zero pressure gradient turbulent boundary layers. Phys. Fluids 19 (6), 065101.
C ASTRO , I. P. 2007 Rough-wall boundary layers: mean flow universality. J. Fluid Mech. 585,
469–485.
C ASTRO , I. P., S EGALINI , A. & A LFREDSSON , P. H. 2013 Outer-layer turbulence intensities in
smooth- and rough-wall boundary layers. J. Fluid Mech. 727, 119–131.
C HAN , L., M AC D ONALD , M., C HUNG , D., H UTCHINS , N. & O OI , A. 2015 A systematic investigation
of roughness height and wavelength in turbulent pipe flow in the transitionally rough regime.
J. Fluid Mech. 771, 743–777.
C HAUHAN , K. A., M ONKEWITZ , P. A. & NAGIB , H. M. 2009 Criteria for assessing experiments in
zero pressure gradient boundary layers. Fluid Dyn. Res. 41 (2), 021404.
C HIN , C., P HILIP, J., K LEWICKI , J. C., O OI , A. & M ARUSIC , I. 2014 Reynolds-number-dependent
turbulent inertia and onset of log region in pipe flows. J. Fluid Mech. 757, 747–769.
C LAUSER , F. H. 1956 The turbulent boundary layer. Adv. Appl. Mech. 4, 1–51.
C OLES , D. E. 1962 A manual of experimental boundary-layer practice for low-speed flow. RAND
Corp. Rep. R-403-PR. The Rand Corp, Santa Monica, CA, USA.
D E G RAAFF , D. B. & E ATON , J. K. 2000 Reynolds-number scaling of the flat-plate turbulent
boundary layer. J. Fluid Mech. 422, 319–346.
E FROS , V. & K ROGSTAD , P.-Å. 2011 Development of a turbulent boundary layer after a step from
smooth to rough surface. Exp. Fluids 51 (6), 1563–1575.
E RM , L. P. & J OUBERT, P. N. 1991 Low-Reynolds-number turbulent boundary layers. J. Fluid Mech.
230, 1–44.
F LACK , K. A., S CHULTZ , M. P. & C ONNELLY, J. S. 2007 Examination of a critical roughness
height for outer layer similarity. Phys. Fluids 19 (9), 095104.
F LACK , K. A., S CHULTZ , M. P. & S HAPIRO , T. A. 2005 Experimental support for Townsend‘s
Reynolds number similarity hypothesis on rough walls. Phys. Fluids 17 (3), 035102.
F LORES , O. & J IMENEZ , J. 2006 Effect of wall-boundary disturbances on turbulent channel flows.
J. Fluid Mech. 566, 357–376.
F OSS , J. & H AW, R. 1990 Transverse vorticity measurements using a compact array of four sensors.
T. Heuris. Therm. Anemom. 97, 71–76.
G RASS , A. J. 1971 Structural features of turbulent flow over smooth and rough boundaries. J. Fluid
Mech. 50 (02), 233–255.
H ARUN , Z., M ONTY, J. P., M ATHIS , R. & M ARUSIC , I. 2013 Pressure gradient effects on the
large-scale structure of turbulent boundary layers. J. Fluid Mech. 715, 477–498.
H ONG , J., K ATZ , J. & S CHULTZ , M. P. 2011 Near-wall turbulence statistics and flow structures over
three-dimensional roughness in a turbulent channel flow. J. Fluid Mech. 667, 1–37.
H ULTMARK , M., VALLIKIVI , M., BAILEY, S. C. C. & S MITS , A. J. 2013 Logarithmic scaling of
turbulence in smooth- and rough-wall pipe flow. J. Fluid Mech. 728, 376–395.
H UTCHINS , N. & M ARUSIC , I. 2007a Evidence of very long meandering features in the logarithmic
region of turbulent boundary layers. J. Fluid Mech. 579, 1–28.
H UTCHINS , N. & M ARUSIC , I. 2007b Large-scale influences in near-wall turbulence. Phil. Trans. R.
Soc. Lond. A 365 (1852), 647–664.
H UTCHINS , N., N ICKELS , T. B., M ARUSIC , I. & C HONG , M. S. 2009 Hot-wire spatial resolution
issues in wall-bounded turbulence. J. Fluid Mech. 635, 103.
JACKSON , P. S. 1981 On the displacement height in the logarithmic velocity profile. J. Fluid Mech.
111, 15–25.
J IMÉNEZ , J. 2004 Turbulent flows over rough walls. Annu. Rev. Fluid Mech. 36, 173–196.
K EIRSBULCK , L., L ABRAGA , L., M AZOUZ , A. & T OURNIER , C. 2002 Surface roughness effects on
turbulent boundary layer structures. Trans. ASME J. Fluids Engng 124 (1), 127–135.
 Smooth- and rough-wall boundary layers 239
K LEWICKI , J. C., F IFE , P. & W EI , T. 2009 On the logarithmic mean profile. J. Fluid Mech. 638,
73–93.
K ROGSTAD , P.-Å., A NTONIA , R. A. & B ROWNE , L. W. B. 1992 Comparison between rough- and
smooth-wall turbulent boundary layers. J. Fluid Mech. 245, 599–617.
K ROGSTAD , P.-Å. & E FROS , V. 2010 Rough wall skin friction measurements using a high resolution
surface balance. Intl J. Heat Fluid Flow 31 (3), 429–433.
K ROGSTAD , P.-Å. & E FROS , V. 2012 About turbulence statistics in the outer part of a boundary
layer developing over two-dimensional surface roughness. Phys. Fluids 24 (7), 075112.
K ROGSTADT, P.-Å. & A NTONIA , R. A. 1999 Surface roughness effects in turbulent boundary layers.
Exp. Fluids 27 (5), 450–460.
L EE , S.-H. & S UNG , H. J. 2007 Direct numerical simulation of the turbulent boundary layer over a
rod-roughened wall. J. Fluid Mech. 584, 125–146.
L EONARDI , S., O RLANDI , P., S MALLEY, R. J., D JENIDI , L. & A NTONIA , R. A. 2003 Direct
numerical simulations of turbulent channel flow with transverse square bars on one wall.
J. Fluid Mech. 491, 229–238.
L IGRANI , P. M. & M OFFAT, R. J. 1986 Structure of transitionally rough and fully rough turbulent
boundary layers. J. Fluid Mech. 162, 69–98.
M ARUSIC , I., C HAUHAN , K. A., K ULANDAIVELU , V. & H UTCHINS , N. 2015 Evolution of
zero-pressure-gradient boundary layers from different tripping conditions. J. Fluid Mech. 783,
379–411.
M ARUSIC , I., M ONTY, J. P., H ULTMARK , M. & S MITS , A. J. 2013 On the logarithmic region in
wall turbulence. J. Fluid Mech. 716, R3.
M EHDI , F., K LEWICKI , J. C. & W HITE , C. M. 2013 Mean force structure and its scaling in
rough-wall turbulent boundary layers. J. Fluid Mech. 731, 682–712.
M ONTY, J. P., A LLEN , J. J., L IEN , K. & C HONG , M. S. 2011 Modification of the large-scale
features of high Reynolds number wall turbulence by passive surface obtrusions. Exp. Fluids
51 (6), 1755–1763.
M ORRILL -W INTER , C., K LEWICKI , J. C., BAIDYA , R. & M ARUSIC , I. 2015 Temporally optimized
spanwise vorticity sensor measurements in turbulent boundary layers. Exp. Fluids 56, 216.
M ULHEARN , P. J. & F INNIGAN , J. J. 1978 Turbulent flow over a very rough, random surface.
Boundary-Layer Meteorol. 15 (1), 109–132.
N ICKELS , T. B., M ARUSIC , I., H AFEZ , S. & C HONG , M. S. 2005 Evidence of the k-1 law in a
high-Reynolds-number turbulent boundary layer. Phys. Rev. Lett. 95 (7), 074501.
N ICKELS , T. B., M ARUSIC , I., H AFEZ , S., H UTCHINS , N. & C HONG , M. S. 2007 Some predictions
of the attached eddy model for a high Reynolds number boundary layer. Phil. Trans. R. Soc.
Lond. A 365 (1852), 807–822.
N IKURADSE , J. 1933 Laws of flow in rough pipes. NASA Tech. Memo. 1292.
P ERRY, A. E. & J OUBERT, P. N. 1963 Rough-wall boundary layers in adverse pressure gradients.
J. Fluid Mech. 17 (02), 193–211.
P ERRY, A. E. & L I , J. D. 1990 Experimental support for the attached-eddy hypothesis in zero-
pressure-gradient turbulent boundary layers. J. Fluid Mech. 218, 405–438.
P ERRY, A. E., M ARUSIC , I. & J ONES , M. B. 2002 On the streamwise evolution of turbulent boundary
layers in arbitrary pressure gradients. J. Fluid Mech. 461, 61–91.
P ERRY, A. E., S CHOFIELD , W. H. & J OUBERT, P. N. 1969 Rough wall turbulent boundary layers.
J. Fluid Mech. 37 (02), 383–413.
R AUPACH , M. R. 1981 Conditional statistics of Reynolds stress in rough-wall and smooth-wall
turbulent boundary layers. J. Fluid Mech. 108, 363–382.
R AUPACH , M. R., A NTONIA , R. A. & R AJAGOPALAN , S. 1991 Rough-wall turbulent boundary layers.
Appl. Mech. Rev. 44 (1), 1–25.
R AUPACH , M. R., T HOM , A. S. & E DWARDS , I. 1980 A wind-tunnel study of turbulent flow close
to regularly arrayed rough surfaces. Boundary-Layer Meteorol. 18 (4), 373–397.
S CHLATTER , P. & Ö RLÜ , R. 2012 Turbulent boundary layers at moderate Reynolds numbers: inflow
length and tripping effects. J. Fluid Mech. 710, 5–34.
240 D. T. Squire and others
S CHULTZ , M. P. & F LACK , K. A. 2003 Turbulent boundary layers over surfaces smoothed by sanding.
Trans. ASME J. Fluids Engng 125 (5), 863–870.
S CHULTZ , M. P. & F LACK , K. A. 2007 The rough-wall turbulent boundary layer from the
hydraulically smooth to the fully rough regime. J. Fluid Mech. 580, 381–405.
S HOCKLING , M. A., A LLEN , J. J. & S MITS , A. J. 2006 Roughness effects in turbulent pipe flow.
J. Fluid Mech. 564, 267–285.
S ILLERO , J. A., J IMÉNEZ , J. & M OSER , R. D. 2013 One-point statistics for turbulent wall-bounded
flows at Reynolds numbers up to δ+ ≈ 2000. Phys. Fluids 25 (10), 105102.
S MITS , A. J., M ONTY, J., H ULTMARK , M., BAILEY, S. C. C., H UTCHINS , N. & M ARUSIC , I. 2011
Spatial resolution correction for wall-bounded turbulence measurements. J. Fluid Mech. 676,
41–53.
S REENIVASAN , K. R. & S AHAY, A. 1997 The persistence of viscous effects in the overlap region,
and the mean velocity in turbulent pipe and channel flows. arXiv Physics (9708016).
TACHIE , M. F., B ERGSTROM , D. J. & B ALACHANDAR , R. 2000 Rough wall turbulent boundary
layers in shallow open channel flow. Trans. ASME J. Fluids Engng 122 (3), 533–541.
TALLURU , K. M., K ULANDAIVELU , V., H UTCHINS , N. & M ARUSIC , I. 2014 A calibration technique
to correct sensor drift issues in hot-wire anemometry. Meas. Sci. Technol. 25 (10), 105304.
T HOM , A. S. 1971 Momentum absorption by vegetation. Q. J. R. Meteorol. Soc. 97 (414), 414–428.
T OH , S. & I TANO , T. 2005 Interaction between a large-scale structure and near-wall structures in
channel flow. J. Fluid Mech. 524 (1), 249–262.
T OWNSEND , A. A. 1956 The Structure of Turbulent Shear Flow, vol. 1. Cambridge University Press.
T OWNSEND , A. A. 1976 The Structure of Turbulent Shear Flow, vol. 2. Cambridge University Press.
VOLINO , R. J., S CHULTZ , M. P. & F LACK , K. A. 2007 Turbulence structure in rough- and smooth-
wall boundary layers. J. Fluid Mech. 592, 263–293.
WALKER , J. M. 2014 The application of wall similarity techniques to determine wall shear velocity
in smooth and rough wall turbulent boundary layers. Trans. ASME J. Fluids Engng 136 (5),
051204.
W EI , T., F IFE , P., K LEWICKI , J. C. & M C M URTRY, P. 2005 Properties of the mean momentum
balance in turbulent boundary layer, pipe and channel flows. J. Fluid Mech. 522, 303–327.
W U , Y. & C HRISTENSEN , K. T. 2007 Outer-layer similarity in the presence of a practical rough-wall
topography. Phys. Fluids 19 (8), 085108.