Professional Documents
Culture Documents
Afzal - 2008 - Alternate Scales For Turbulent Boundary Layer On T
Afzal - 2008 - Alternate Scales For Turbulent Boundary Layer On T
Journal of Fluids Engineering Copyright © 2008 by ASME APRIL 2008, Vol. 130 / 041202-1
form of the law of the wake for rough wall flows. The method also priate Reynolds shear stress, p⬘共x兲 = −UcdUc / dx is the stream-
simultaneously determines C f and ⌬U+. Acharya et al. 关21兴 ob- wise pressure gradient, Uc共x兲 is the velocity at the edge of the
served that simultaneous selection of these parameters can yield turbulent boundary layer of thickness ␦, is the molecular kine-
an improved statistical fit of the data but may, in some cases, give matic viscosity, and is the fluid density.
an increase error in C f . Using the total stress method on rough The present work defines the roughness velocity u and rough-
wall boundary layers has its own shortcomings. This method re- ness coordinate y , in terms of axial velocity u at normal distance
lies on a plateau in the Reynolds shear stress profile, which is y, as y = y / and u = u / , where is the nondimensional
often not clearly defined in the roughness sublayer, and has fairly roughness scale defined later, such that all mean relative motions
high measurement uncertainty in this region 共Schultz and Flack and energy-containing components of the turbulent motion do not
关19,22兴兲. depend on surface roughness. The appropriate inner variables are
The theories of the transitionally rough wall turbulent boundary
layer are based on the inner layer fully smooth wall variable Z+
Zu Z+ Zu
= Zu / or the fully rough wall variable Z / h 共Schlichting 关7兴, = = , Z+ = , Z = y + ⑀r 共3兲
Raupach et al. 关10兴, Patel 关11兴, Piquet 关12兴, and Jimenez 关13兴兲.
冑 2 1
= ln Re
Cf k
冉 冑冊 Cf
2
+B+D 共30兲
4 Roughness Scale and Roughness Length Z0+
4.1 Monotonic Roughness. In commercial monotonic rough-
ness, Colebrook 关23兴 fortuitously proposed that roughness scale
The effects of the pressure gradient on the matching conditions in is simply the sum of fully smooth wall = 1 and fully rough wall
the overlap region lead to universal composite laws, described in = h+, where = exp关k共B − BF兲兴. Adopting B = 5 for fully smooth
the Appendix. For weaker pressure gradient, the matching pro- wall and BF = 8.5 for fully rough wall yields = 0.306. The com-
vides the relations 共A10兲 and 共A11兲, for strong adverse pressure posite function 关23兴 is = 1 + h+.
gradient the matching yields the half-power laws for velocity pro-
file 共A12兲 and 共A13兲 in universal variables, and in traditional vari- 4.2 Inflectional Roughness. The inflectional roughness of
ables the intercepts in two situations depend on the wall rough- Nikuradse 关25兴 has been analyzed by Loselevich and Pilipenko
ness. 共see Cebeci 关30兴兲 to suggest the following correlation:
1
⌬U+ = 共B − 8.5 + k−1 ln h+兲sin关0.4258共ln h+ − 0.811兲兴 = ln
k
3 Reduction to Traditional Picture With Shallow
共39兲
Outer Wake Layer
for 2.25艋 h+ ⬍ 90 with B = 5.2 and k = 0.42.
The universal velocity profile 共27兲, with the help of expressions In the present work, the roughness scale for the inflectional
共5兲, is expressed in traditional smooth wall variable Z+ and rough- roughness of Nikuradse 关25兴 from modification of the Colebrook
ness function ⌬U+ as monotonic relation, by introducing an exponential function that
approaches unity for h+ → ⬁, is stated below:
u 1 ⌸
= ln Z+ + B − ⌬U+ + W共Y兲
u k k
Likewise, the skin friction relation 共29兲 yields
共31兲
冉 冊
= 1 + h+ exp −
j
h+
共40兲
Uc 1 ⌸
= ln R + B − ⌬U+ + W共1兲
u k k
共32兲
⌬U+ =
1
k
冋 冉 冊册
ln 1 + h+ exp −
j
h+
共41兲
u 1 Z
u k h
⌸
= ln + BT + W共Y兲,
k
Bt +
1
k
ln h+ = BT 共34兲
Bt = B −
1
k
冋 冉 冊册
ln 1 + h+ exp −
j
h+
共44兲
respectively, and are the intercepts of the wall log laws for fully
smooth and fully rough walls. The velocity profile log law 共27兲
and the skin friction log law 共29兲 become
Uc 1
= ln R
u k
再 冒冋 冉 冊册冎
1 + h+ exp −
j
h+
+B+
⌸
k
W共1兲
共46兲
u 1
= ln Z+
u k
再 冒冋 冉 冊册冎
1 + h+ exp −
j
h+
⌸
+ B + W共Y兲
k
共45兲 The skin friction log law 共46兲 may also be expressed as
Uc
u
1
= − ln
k
1
冋
R
h
exp关− k共B + D兲兴 + exp共− k共BF + D兲兴
␦
冑 2
Cf
1
冋
= − ln exp关− k共B + D兲兴
k
再
1
R
h
+ exp −
␦
冉 冊冎册
j ␦
R h
⫻exp −冉 冊册 j
h+
共47兲
共48兲
The skin friction coefficient C f in terms of the friction Reynolds The relation 共48兲 in terms of the Reynolds number Re= Uc␦ /
number R = u␦ / becomes becomes
Fig. 3 The log law velocity distribution due to rod roughness data of Smalley et al. †34‡ in semilog plots: „a… Inner smooth
wall variables „u+ , Z+…. „b… Velocity profile shifted by the roughness function u+ + ⌬U+ against smooth wall variable Z+. „c… Our
universal variables for inner transitional rough wall „u+ , …. „d… Alternate universal inner rough wall variables „u+ , Z / Z0…. „e…
Outer velocity defect variables „„Uc − u… / u , Y….
冑 2 1
冋
= − ln exp关− k共B + D兲兴
1
再冑 2 required j = 11 and for Colebrook monotonic roughness j = 0.
冊冎册
Cf k Re Cf 4.3 k-Type and d-Type Surface Roughness. The data of
h
+ exp −
␦
冉 冑
j
Re
2 ␦
Cf h
共49兲
Osaka and Mochizuki 关31兴 for d-type roughness, Kameda et al.
关32兴 for k-type roughness, Antonia and Krogstad 关33兴 for rod
roughness, and Smalley et al. 关34兴 for rod roughness, as shown
For present predictions for inflectional roughness, fit to the data later, are predicted by the following simple relations:
冑 2 1
= ln
Cf k
␦
h
冉冊
+B+D 共52兲
ness, the k-type roughness, and the d-type roughness in the engi-
neering applications along with wide variety of the atmospheric
surfaces. The experimental data from various sources 共Osaka and the velocity profile data, shown in Fig. 1共c兲, are universal relation
Mochizuki 关31兴, Kameda et al. 关32兴, Antonia and Krogstad 关33兴, 共24兲, where all data collapse on this single line, irrespective of
Smalley et al. 关34兴, Schultz and Myers 关35兴, Schultz and Flack roughness scale h+ or ␦ / h. In the outer layer shallow defect vari-
关22,36兴, and Rahman and Webster 关37兴兲 for zero pressure gradient ables 共Ue+ − u+ , Y兲, the velocity profiles also collapse on line 共35兲,
have been considered. shown in Fig. 1共d兲, and do not depend on wall roughness. Present
The velocity profile from data of Kameda et al. 关32兴 for k-type work as well as that of Schultz and Flack 关19,22兴 and Connelly et
roughness are shown in Figs. 1共a兲–1共d兲. In traditional smooth wall al. 关5兴 support the Townsend 关38兴 similarity hypothesis in the
variables 共u+ , Z+兲, the velocity profile data are shown in Fig. 1共a兲, outer layer, which states that the turbulence outside of the rough-
which in the overlap region may be represented by lines that shift ness sublayer, a layer extending out approximately five roughness
their locations with change of roughness parameter h+ or ␦ / h. The heights from the wall, is independent of the surface condition at
velocity distribution shifted by roughness function u+ + ⌬U+ ver- sufficiently high Reynolds numbers. The velocity distribution data
sus, smooth wall variable Z+ is shown in Fig. 1共b兲 on log-log plot, of Osaka and Mochizuki 关31兴 for d-type roughness are shown in
which is the Clauser–Hama type similarity and does not depend Figs. 2共a兲–2共e兲, and the data of Smalley et al. 关34兴 in Figs.
on wall roughness. In the transitional rough wall variables 共u+ , 兲, 3共a兲–3共e兲. The velocity distribution data of Schultz and Myers
Fig. 6 The log law velocity distribution on machine honed surface roughness data of Schultz and Flack †19‡ in semilog
plots: „a… Inner smooth wall variables „u+ , Z+…. „b… Velocity profile shifted by the roughness function u+ + ⌬U+ against smooth
wall variable Z+. „c… Our universal variables for inner transitional rough wall „u+ , …. „d… Alternate universal inner rough wall
variables „u+ , Z / Z0…. „e… Outer velocity defect variables „„Uc − u… / u , Y….
关35兴, and Schultz and Flack 关36兴 are shown in Figs. 4共a兲–4共e兲, the The data of the roughness function ⌬U+ with roughness vari-
data of Schultz and Flack 关22兴 in Figs. 5共a兲–5共e兲, the machine able h+ are shown in Fig. 8共a兲. The present prediction 共38兲 for
honed surface roughness data of Schultz and Flake 关19兴 in Figs. inflectional type roughness and Colebrook 关23兴 monotonic rough-
6共a兲–6共e兲, and data of Rahman and Webster 关37兴 in Figs. ness are also shown in the same figure for appreciation of com-
7共a兲–7共e兲. The velocity distribution data in Figs. 2–7 marked a are mercial technical roughness effects. The data marked Schultz &
in traditional smooth wall variables 共u+ , Z+兲, b are in Clauser– Flack 1, Schultz & Flack 2, and Schultz & Myers 关35兴 compare
Hama variables 共u+ + ⌬U+ , Z+兲, c are in transitional rough wall well with sand grain roughness prediction 共38兲. The data of Osaka
variables 共u+ , 兲, d are in alternate rough wall variables 共u+ , Z / Zo兲, and Mochizuki 关31兴 for d-type roughness, Kameda et al. 关32兴 for
and e are in traditional outer variables 共Ue+ − u+ , Y兲 and are analo- k-type roughness, and Smalley et al. 关34兴 may be described by the
gous to Figs. 1共a兲–1共e兲. No additional comment is needed. following simple predictions 共47兲–共49兲. The roughness potential
1
冑C f = 1.76 ln冋 1
R
h
+ 0.306 exp −
␦
冉 冊册
j ␦
R h
+ 5.09 共55兲
1
冋
冑C f = 1.76 ln
1
Re 冑C f
h
+ 0.216 exp
␦
冉 −
1.414j ␦
Re 冑C f h
冊册 + 4.48
共56兲
For the data of Antonia and Krogstad 关33兴 for mesh roughness,
Schultz and Flack 关19兴 for machine honed surface roughness,
Schultz and Meyers 关35兴 for epoxy roughness, Schultz and Flack
关36兴 for sand paper roughness, Schultz and Flack 关22兴 for uniform
sphere and uniform spheres with grit and Rahman and Webster
关37兴 for river bed roughness, the skin friction is shown in Fig. 11
Fig. 10 Comparison of universal skin friction Cf predictions against Re the Reynolds number. These data are compared with
and experimental data for turbulent flow boundary layer on the predictions 共53兲 and 共54兲 with j = 11 for the inflectional type
transitional rough walls. „a… Skin friction coefficient Cf against roughness of Nikuradse, for various values of roughness param-
roughness Reynolds number R = R / data and universal pre- eter ␦ / h = 15, 30, 60, 120, 200, and 400. For each value of ␦ / h,
diction „53…, 1 / 冑Cf = 1.76 ln R + 5.09. „b… Skin friction coefficient the prediction represents a line that for large R approaches to
Cf against roughness Reynolds number Re = Re/ data and
fully rough wall limit and for small R it merges with smooth wall
universal prediction „54…, 1 / 冑Cf = 1.76 ln„Re冑Cf… + 4.48. skin friction law. For the data of Osaka and Mochizuki 关31兴 for
k-type roughness, Kameda et al. 关32兴 for d-type roughness, Anto-
nia and Krogstad 关33兴 for rod roughness, Smalley et al. 关34兴 for
Mochizuki 关31兴 for d-type roughness, Kameda et al. 关32兴 for
rod roughness, and Leonardi et al. 关14兴, the skin friction is shown
k-type roughness, and Antonia and Krogstad 关33兴, Smalley et al.
in Fig. 12 against Re the Reynolds number. These data are com-
关34兴, and Leonardi et al. 关14兴 for rod roughness.
pared with the prediction 共58兲 for k-type and d-type roughness for
In fact, the friction factor predictions 共39兲 and 共40兲 are univer-
various values of roughness parameter ␦ / h = 20, 50, 100, 200, and
sal relations, explicitly independent of surface roughness. The
400 as well as smooth walls. For each value of the roughness
skin friction, in alternate variables 共C f , Re兲 shown in Fig. 10共a兲,
parameter ␦ / h, the prediction approaches to fully rough wall lim-
also collapses on a single universal line predicted by relation 共39兲.
iting value for R and smooth wall skin friction law for small R.
Likewise, the friction factor in alternate transitional roughness
The prediction shown in Figs. 11 and 12 compares well with data
variables 共C f , Re兲 from roughness data is shown in Fig. 10共b兲,
for various values of the wall roughness parameter ␦ / h.
which collapses on a single line predicted by relation 共30兲. The
data for various types of roughness, in transitional wall roughness, 6 Conclusions
provide good support to universal skin friction, and the log laws
共29兲 or 共30兲 become The alternate new nondimensional roughness scale , transi-
tional inner coordinate = Z+ / , new parameters R = R / , the
1 roughness friction Reynolds number, and Re = Re/ , the rough-
共53兲
冑C f = 1.76 ln R + 5.09 ness Reynolds number, have been proposed. Here,
= k−1 ln ⌬U+ is roughness scale connected with traditional rough-
1
冑C f 兲 + 4.48 ness function ⌬U+.
共54兲
冑C f = 1.76 ln共Re The alternative scale is connected with traditional roughness
function, which provides the wall roughness effects in the new
The relations 共53兲 and 共54兲 demand 共B + D兲 / 冑2 = 5.09, which im- inner wall variable = Z+ / , which has opened the avenues of
ply B = 5 and D = 2.2. The universal skin friction log laws 共53兲 and new parameters such as roughness friction Reynolds number and
共54兲 are expressed in terms of traditional Reynolds numbers as roughness Reynolds numbers. These new parameters provide uni-
Acknowledgment
N.A. is thankful to the All India Council of Technical Educa-
tion, New Delhi for the support of this work.
are explicitly independent of wall roughness. It also helps to yield is also of significance. The inner limit 共 fixed, R → ⬁兲 of outer
the traditional relations and also provides insight into the rough- expansions is the outer limit 共Y fixed, R → ⬁兲 of inner expansion.
ness scale formulation for inflectionl type roughness as well as The matching condition 共A1兲, expressed in terms of outer and
k-type and d-type roughness. On the other hand, the traditional inner expansions, yield
study of rough wall is based on fully smooth wall inner scale Z+,
where effect of the wall roughness shows a shift of the velocity u+ Y F0 Y F1
= ⑀−1 共A2兲
profile log law by an amount equal to roughness function over the 冑+ 冑T Y − 冑T Y
fully smooth wall layer log law.
In the alternate rough wall variables, the velocity profile u+ as → ⬁ and Y → 0 for R → ⬁ and the perturbation parameter is
versus inner transitional roughness variable = Z+ / , the skin fric- ⑀ = u / Uc. The matching of the inner equation 共3兲 for Reynolds
tion C f versus inner transitional roughness Reynolds number shear stress + at large with the outer equation 共15兲 for Reynolds
Re= Re/ are universal, explicitly independent of wall rough- shear stress T yields
ness but implicitly involving the roughness scale . The extensive
experimental data for transitional wall roughness provide very + = 1 + K 共A3a兲
good support to present theory of universal velocity profile in the
inner layer and the skin friction. The outer velocity profiles 共28兲 T = 1 + Y 共A3b兲
are also universal, which supports Townsend’s similarity hypoth- Based on relations 共A2兲 and 共A3a , b兲, the nondimensional match-
esis. ing condition yields
The present deals with transitional inner wall variable
= Z+ / , for a given roughness level , leading to universal laws, u+ Y Fo Y F1
= ⑀−1 共A4兲
explicity independent of transitional wall roughness. On the other 冑1 + K 冑1 + Y Y − 冑1 + Y Y
hand, traditional rough wall theories are based either on fully
smooth inner layer variable Z+ = Zu / or fully rough wall inner As R → ⬁, ⑀ → 0, the matching of leading order term demands
variable Z / h and the results depend on type of roughness function
⌬U+ or roughness scale . Y F0
→0 as Y → 0 共A5兲
The roughness scale prediction of inflectional roughness 冑1 + Y Y
compares well with the data of Schultz and Flack 关22,36兴. The
data of Osaka and Mochizuki 关31兴 for d-type roughness, Kameda and the next order terms become
et al. 关32兴 for k-type roughness, and Antonia and Krogstad 关33兴
u+ Y F1
and Smalley et al. 关34兴 for rod roughness compare well with the =− 共A6兲
k-type and d-type roughness prediction = h+. The traditional log 冑1 + K 冑1 + Y Y
laws intercept BT and Bt for the appropriate roughness scale also Integration of each side of Eq. 共A6兲 leads to matching conditions
compare well with the corresponding data. for tangential velocity as
冋 册
There is no universality of scalings in traditional variables and
different expressions are needed for various types of roughness, as 冑1 + K + 1
+ 冑1 + K − 1 + B
u A
suggested, for example, with inflectional type roughness, mono- = ln + A − ln
tonic Colebrook–Moody roughness, k-type/d-type roughness, etc. u 2 2
are predicted. In traditional variables, the skin frictions are pre- 共A7兲
冋 册
dicted for inflectional roughness and k-type/d-type roughness. The
experimental data provide good support to the predictions of the 冑1 + Y + 1
+ 冑1 + Y − 1 + DS
US − u A
skin friction in terms of traditional log law for inflectional rough- = − ln Y − A − ln
ness as well as k-type and d-type roughness. u 2 2
The effects of the pressure gradient on the matching conditions 共A8兲
冋 册
The shallow outer wake layer 共US ⬅ Uc兲 under very strong adverse
= A冑K + C +
u A 1 pressure gradient was adopted by Kader and Yaglom 关42兴 to get
− 共K兲−1/2 + 共K兲−3/2 + ¯
u 2 12 the half-power law for the outer velocity profile, which is not
共A12兲 valid in the neighborhood of the separation domain 共Afzal
关41,43兴兲. From turbulent boundary layer experimental data, the
US − u
u
= − A冑Y + E +
A
2
1
冋
共Y兲−1/2 − 共Y兲−3/2 + ¯
12
册 half-power law constants A versus ⌳, C p versus ⌳R p, and E p
versus ⌳ on a fully smooth surface 共 = 1, ⌬U+ = 0兲 are also shown
in Afzal 关43兴, under strong adverse pressure gradient with negli-
共A13兲 gible wall shear stress, in the neighborhood of separation. It is
shown that the data power law slope A versus ⌳ data would not be
A inconsistent with A = 2 / k, where k = 0.4 is the universal value of
C=− ln K + A共ln 2 − 1兲 + B 共A14兲 the Karman constant.
2
A
E= ln + A共− ln 2 + 1兲 + DS 共A15兲
2 Appendix B: Nonlinear Outer Wake Layer With Eddy
The inner half-power laws 共A12兲 and 共A13兲 may also be ex- Viscosity Model
pressed as The present work dealt with the open equations of turbulent
u
up
= A 冑y p + C p +
A
2
冋 1
− ⌳ R p共y p兲−1/2 + 共⌳R p兲2共y p兲−3/2 + ¯
12
册 motion, up to a point beyond which it is not possible to proceed
any further without a closure hypothesis. Consequently, in outer
layer the Clauser’s constant eddy viscosity model has been
共A16兲 adopted for lowest order outer Eq. 共16兲 subjected to the boundary
conditions f ⬘0共X , ⬁兲 − 1 = T0共X , ⬁兲 = 0 and the matching conditions
C p = 共⌳R p兲1/2 − 冋 A
2
ln K + A共ln 2 − 1兲 + B 册 共A17兲
→ w demanding T0共X , 0兲 = 1. The eddy viscosity hypothesis
= u / y in outer layer where ␣c = / Ue␦* being Clauser con-
stant, ␦ = ␣ca1共x − xo兲 and 共1 + m兲a1 = ␦* / ␦ and the outer layer has
US − u
u␦
= − A冑Y + E p +
A
2
1
冋
⌳Y −1/2 − ⌳2Y −3/2 + ¯
12
册 been in equilibrium, f 0共X , Y兲 = f共Y兲 subjected to the power law
velocity Ue ⬀ ␦m at the edge of the boundary layer, give
E p = ⌳1/2 −冋 A
ln ⌳ + A共− ln 2 + 1兲 + DS 册 共A19兲
f共0兲 = 0, 共B2a兲
冕
2 ⬁
⑀2
Based on pressure scales, the inner variable is y p = yu p / , inner f ⬙共0兲 共1 − f ⬘兲dY = , 共B2b兲
velocity scale is u p = 共 p⬘ / 兲1/3, the outer velocity scale is u␦ o
␣c
= 共␦ p⬘ / 兲1/2 and the nondimensional parameters are ⌳ = w / 共␦ p⬘兲
and R p = u p␦ / . For a transitional rough surface, near separation, f ⬘共⬁兲 = 1 共B2c兲
the inner half-power law 共A16兲 becomes The Falkner–Skan equation 共B1兲 with the pressure gradient pa-
rameter  = m / 共1 + m兲 is subjected to the boundary conditions of
= A 冑y p + C p
u
共A20兲 prescribed wall shear stress 共B2b兲. The solution to these equations
up
predict the outer slip velocity f ⬘共0兲 = bS as a function of the pa-
The outer half-power law 共A18兲 becomes rameter ⑀2 / ␣c. For ⑀ → 0, the wake- and jetlike solutions are de-
scribed in Afzal 关27,28,41兴. The zero slip velocity 共f ⬘共0兲 = bS = 0兲
= − A冑Y + E p
US − u
共A21兲 with f ⬙共0兲 = 0 implies a point of vanishing skin friction and the
u␦ solution of the Falkner–Skan equations 共B1兲 and 共B2兲 predicts 
The additive constant C p in the inner layer half-power law 共A20兲 = −0.198838 and m = −0.0.165858. Furthermore, the outer flow is
depends on wall roughness potential or the roughness function a shallow wake 共US = Uc兲, and the lowest order outer solution
⌬U+ as becomes f ⬘共Y兲 = 1 and bS = 1.