DOE/NASA/1059-79/4

NASA TM-79275
{NASA-TM-79275) MODIFIED POWER LAW N80-13623
EQUATIONS Fu_ VERTICAL WIND PBOFILES (NASA)
: : 13 p EC A02/MF A01 CSCL 10A
Unclas
' " ,, G3/4q _6296

, ;. MODIFIED POWER LAW EQUATIONS

FOR VERTICAL WIND PROFILES

D. A. Spera and T. R. Richards

National ,a.eronaut,csand Space Administration
Lewis Recearch Center

Work performed for

U.S. DEPARTMENT OF ENERGY
Energy Technology
Distributed Solar Technology Division

Prepared for
Wind Characteri,,':icsand WandEnergy Siting Conference
sponsored by the U.S. Department of Energy, the American
Meteorological Society, and the Pacific Northwest Laboratory
Portland, Oregon, ,June 19-21, 1979

i
.,, ' __ :, -
 i

DOEINA
SAii 0.59-i914
NASATM-79275

MODIFIED
POWER
LAWEQUATIONS
': FORVERTICAL
WINDPROFILES

D. A. SperaandT. R. Richards
NationalAeronauticsand SpaceAdministration
LewisResearchCenter
Cleveland.Ohio 4413.5

Workperformedfor
U.S. DEPARTMENT OFENERGY
EnergyTechnology
DistributedSolarTechnologyDivision
Washington,D.C. 20545
UnderinteragencyAgreementE(49-26)-1059
P

WindCharacteristicsandWindEnergySiting CoDference
sponsoredby the U.S. Departmentof Energy,the American
Meteorologica]Society,andthe PacificNorthwestLabo,'atory
Portland, Oregon,June 19-21,1979

/
 MODIFIED POWER LAW EQUATIONS
FOR VERTICAL WIND PROFILES

D. A. Spera and T. R. Richards

National Aeronautics and Space Administration

Lewis Research Center
Cleveland, Ohio, 44135

ABSTRA_

Equations are presented for calculating power law exponents from wind
speed and surface roughness data. Results axe evaluated by comparison with
wind profile date measured at a variety of sites.

INTRODUCTION

The power law equation is a simple yet useful model of the vertical
wind profile which was first proposed by Hellman (1916), according to Simiu
and Scanlan (1978). The general from of this equation is

V2 = Vl{Z2/Zl )_ (1)

_Tjwhich _i and V2 are simultaneous steady wind speeds over level ter-
rain at elevations zI and z2, respectively. The exponent _ is deter-
_ined experimentally. For example, early work by Von Xarman (1921} showed
that under certain conditions _. is equal to I/7, indicating a correspon-
dence between wind profiles and fluid flow over flat plates (Schlicting,
1968). In the general case, however, 0L is a highly variable quantity.
Sisterson and Frenzen (1978) measured wind profile exponents which changed
from 1/7 during the day t_ 1/2 at night, over the same terrain. Golding
(1955) describes _ as an exponent which varies with height, time of day,
season of the year, nature o£ the terrain, wind speed, and temperature.
:bst investigators agree that a constant value of _ is an oversimplifi-
cation, and that 0_ must be treated as a statistical parameter.

In spite of the variable nature of the exponent _, Equation (1) is

simple enough in form to permit adding some of the factors listed by Golding
without losing the engineering convenience of the power law. Two of these
factors will be considered here: (1) the nature of the terrain, tn terms of
its surface roughness, and (2) the wind speed.

Up until now, the effects of wind speed and surface roughness have only
been considered separately. As shown in Figure I, several functional rela-
*.ionships have been proposed for the variation of the exponent (_ with wind
speed. These functions are the constant 117 law, a step function (Fales,
1967), a linear functim_ for win_ speeds exceeding 27 meters per second
(ASCE, 1961), a pvwer fu_ction (Fichtl _ Smith, 1977), and a logarithmic
function (Justu: _ Mikhs_l, 1976). Ali of the variable functions show a
 decrease in the exponent with increasing wind speed, but none includes the
effect of surface roughness. Frost et al (1978) and Justus (1978) give
equivalent values of the exponent g and the well-known surface roughness
length Zo, but without specifying the wind speed.
I

Thus, information on the separate effects of wind speed and surface

roughness is available, but a wind profile model which combines these two
effocts is lacking. The objective of this paper, then, is to present equa-
tions relating the mean value of /_. to both the surface roughness and the
steady wind speed. Other factors are assumed to be responsible for statis-
tical variations about this mean value.

The equations presented in this paper are exploratory in nature and

i
: should be reevaluated with more data, when such data are available.

SUMMARYOF EQUATIONS

The Justus-Mikhail equation shown graphically in Figure 1 was selected

as a starting point for this study because of its relative simplicity and
the fact that it has been shoan to be compatible with wind distribution
data. This equation for the wind profile exponent _ in Equation (1) can
be written as follows:

= 0.37 - 0.203 log Vr (2)

in which Vr is the wind speed at a reference evaluation of 10 meters.

Equation (2) can be generalized to include a surface roughness para-

=eter and statistical variations. Assume _. is a random variable which is
;;ormally distribt_ted about a mean value _ with standard deviation _'.
Assume further that both _ and I/" are functions of the steady (non-
gusting) wind speed and _. is also dependent on the surface roughness.
iiaen, a proposed equation for _ , based on Equation (2) is

1 - log Vl/lOg Vh (3a}

1 - 0(o log CZl/Zr}/log gh
and

_'o = (Zo/Zr)0"2 (3b)

in which

_o surfac e roughness exponent

Vh homogeneous wind speed (_ _ 0), m/s

V1 steady wind speed at elevation Zl, m/s

Zr reference elevation, 10m

Zo surface roughness length, m

 Equation (3b) is _ e_irical relationship _ich will be discussed
later. _e standard deviation _(Vl) must be determined from site data.

Equations (3) are s_o_ gr_hicaJly in Fibres 2 _d 3, The te_ain

! descriptions in Figure 3 follow Frost et al (1978} and Justus (1978).

Weibull distribution par_eters C _d k and the wind profile para-

meters _o _d Yh are r_lated, as sho_ by the following equations:

• in _ich

P(VI_V) probability that the steady wind speed _ elevation

z 1 will exceed V

C1 Weibull scale factor at elevation z 1, m/s

k1 Weibull shoe factor at elevation z1

Weibull distribution patterers at an elevation z 2 cm be calculated

from those at elevation z 1 by means of the following equations:

_C,1
C2 = Cl(Z2/Zl) (Sa_

I - log el/lOg Vh
(Sb)
-c,l --o 1 - _o log (Zl/Zr)/log Vh

1 - _o log (Zl/Zr)/log Vh
(Sc)
k2 = kl 1 - _o log (z21Zr)/log Vh

The Justus-t:ikhail equation, (2), is a special case of Equation (3a)

with 0Co equal to 0.37 and Vh equal to 67 meters per second. Equations
equivalent to Equations (3a) and (S) with these specific v_lues of b_o and
Vh were also used by Justus et al (1976) to calculate reference wind speeds.
CALCULATED RND OBSERVED WIND PROFILES

Mean Wind Speed Profiles

Wind profile data (WSSI, 1976 to 1978) from wind turbine sites or poten-
tial sites selected by the Department of Energy were used for a preli_inary
vvaluation of Equations (3). Table I lists calculatea roughness exponents
_ for five such sites and also for the Justus-Mikhail reference site.
A homogeneous wind speed of 67 meters per second was assumed for all sites
in this prelininary evaluation. Figure 4 illustrates the procedure for
calculating go, using data for Clayton, N_4. Exponents calculated from
monthly average wind speeds at several elevations are plotted versus the
logarithm of the reference elevation wind speed, Vr. A straight line
through the centroid of these data points and the homogeneous point
(Vr = Vh = 67 m/s and _. = O) defines _o, at a wind speed of one meter per
second.

The roughness exponents in Table I vary from 0.1 to almost O.S, in

qualitative agreement with the site characteristics. Higher do values
occur at rougher sites and lower values at smoother sites. The Justus-
Mikhail constant of 0.57 appears to be somewhat too large to be used as an
average roughness exponent. However, more profile data would have to be
analyzed before a better reference value can be established. Based on these
preliminaz7 calculations, an average _o of about 0.50 is indicated.

Equation 5[b) is an empirical relationship between 0_o and Zo, which

is the conventional measure o£ surface roughness. As shown in Figure 5, the
exponent of 0.2 was obtained by curve-fitting equivalent values of 0_ and
Zo from the literature (Frost, et al, 1978, and Justus, 1978). Calculated
values of 0_o from Table I then established the coefficient as (zr) -0-2.

Peak _ind Speed Profiles

As shown in Figure 2, the proposed profile model is based on the assump-

*.ion that wind profiles become uniform or "homogeneous" at high wind speeds,
_¢or all values of surface roughness. This concept represents an extrapola-
tion of the Justus-Mikhail equation as originally proposed, and additional
v.orification is required. To do this, profile exponents were calculated for
monthly peak winds measured at 11 DOE sites (WSSI, 1976 to 1978), to deter-
mine if there was a correlation between high winds and low profile exponents.

Typical results of the peak wind analysis are shown in Figure 4, for
the Clayton NM, site. Values of O_ calculated from monthly peak winds
which averaged 23 meters per second were found to be significantly lower
than exponents calculated from monthly average winds of about 6 meters per
.'ccond. The assumed homogeneous wind speed of 67 meters per second is con-
sistent with the Clayton peak wind data.

Table II summarizes the analysis of peak wind profile data. This table
lists mean peak wind speeds, mean profile exponents, and standard deviations,
for each of II DOE sites. A co._posite average is also given which indicates
that mean profile exponents less than 0.07 can be expected at wind speeds
higher than 22 meters per second. To predict extreme wind loads, the "mean
plus 3_'" value of _ would be used, which is 0.27 for the composite average
site. II_is compares closely wJ th the ASCE recommended value of 0.50 shown in
Figure 1_ for a wind speed of 22 meters per second. Thus, equations [3a) and
(3b) are representative of the mean vatucs of the wind profile exponent_ and
should not be used in extreme winds ana]ysis. More conservative values should
be used.

4
 Mean profile exponents in Table II are shown graphically in Figure 6.
Also shown in this figure are mean profile exponents calculated from monthly
average wind speeds measured at five DOE sites (see Table I). A band repre-
senting predicted values of _ for flat, open terrain is also shown for
comparison. At low wind speeds surface roughness is an important considera-
tion in determining _(. Exponents for the two rough sites (Pltun Brook and
Block Island) fall abcve the band, and that for a smooth hillside site
(Culebra) falls well below. At high wind speeds surface roughness is no
longer a significant variable and mean exponents for all sites are low.

CONCLUDING REMARKS

Equations have been presented and evaluated for a wind profile model
which incorporates both roughness and wind speed effects while retaining the
basic simplicity of the Hellman power law. Moreover, these equations recog-
nize the statistical nature of wind profiles and are compatible with exist-
ing anal)rticalmodels and recent wind profile data.

Predictions of energ; output based on the proposed profile equa-

tions are l0 percent to 20 percent higher than those made with the 1/7 power
law. In addition, correlation between calculated and observed blade loads
is significantly better at hSgher wind speeds when the proposed wind profile
model is used instead of a constant power model.

It is recommended that statistical analysis of site wind data include a

bivariate distribution of profile exponent versus wind speed. This would
permit more accurate calculation of the parameters in the proposed model anJ
permit estimates of standard deviation as a function of wind speed. Statis-
tical information of this type would be extremely useful to wind turbine
desiglers for predicting wind load spectra, fatigue life o£ components, and
energy output.

REFERENCES

American Society of Civil Engineers (1961): "Wind Forces on Structures,"

Trans., 126, Part II, 1124-1198.

Fales, E. N. (1967): "Windmills," Standard Handbook for Mechanical Engin-

, eers, 7th Ed., Baumeister and Marks, Eds. McGraw Hill (New Yerk),
9-13.

Fichtl, G. H., and Smith, O. E. (1977): "Wind," Terrestrial Environment

_Climatic) Criteria Guidelines for Use in Aerospace Vehicle Develop-
ment, 1977 Revision, J. W. Kaufman, Ed., NASA TM-;'8118, 8.15-8.17.

Frost, W., Long, B. H., and Turner, R. E. (1978): "Engineering Handbook

on the Atmospheric Environmental Guidelines for Use in Wind Turbine
Generator Development," NASA TP-i359, 3.15.

golding, E. W. (1955): The Generation of Electricit 7 by Wind Power

(London), 80.
 Hellman, 6. (1916): "Uber die Bewegung der Luft in den untersten Schichten
der Atmosphere," Meteorol. Z., 34, 273.

Justus, C. G., and Mxkhail, A. (1976): "Height Variation of Wind Speed

and Wind _,stribution Statistics," Gcophys. Res. Letters, 3, 261-264.

Justus, C. G., Hargraves, W. R., and Mikhail, A. (1976): "Reference

Wind Speed Distributions and Height Profiles for Wind Turbine Design
and Performance Evaluation Applications," ERDA Tech. Report ORO/5!08-
76/4, A-21 to A-Z3.

Justus, C. G. (1978): Winds and Wind System Performance, Franklin Inst.

Press (Philadelphia), 56-65.
\,
Schlicting, H. {1968): Boundary La_er.The0ry, 6yh Ed., McGraw-}lill
(New York), 564.

Simiu, E., and Scanlan, R. H. (1978): Wind Effects on Structures, Wiley

(New York), 47.

Sisterson, D L., and Frenzen, P. (1978): "Nocturnal Boundary-Layer

Wind Maxima and the Problem of Wind Power Assessment," Env. Sci.
Tech., 12, 218-21.

Von Karman, T. (1921): Z. Angew. Math. Mech., 1, 239.

Western Scientific Services, Inc. (1976 to 1978): "Candidate Wind

Turbine Generator Site Meteorological Monitoring Program Monthly
Reports," (unpublished).

il
 Table I. - Surface roughness exponents for five candidate wind
turbine sites,

Surface _roughness a
i Site Type of Terrain exponent, [_o

Plum Brook, OH flat; high woods 0.48

Block Island, RI hilly; medium woods 0.44
Reference (Justus and relatively flat 0.37
Mikhail, 1976)
Clayton, NN flat; open 0.30
Russell, KS flat; open 0.30
' Culebra, PR shoreline hillside; 0.I0
bushes

E = _o(l - log Vr/log Vh), with Vh = 67 m/s

Table II. - Wind profile exponents for monthly peak winds at

candidate wind turbine sites.

Mean peak Wind profile exponent,_

No. of speed at

Site readings 10m, m/s , Hkan' Std.0 _dev''

Amarillo, TX 26 22.3 0.053 0,045
Block Island, RI 30 19.5 .I14 .052
Clayton, NH 24 22.8 .054 .044
Cold Bay, AK 9 24.1 .042 .044
Culebra, PR 30 17.4 .0]0 .078
Huron, SD 28 20.6 .094 .056
Kingsley Dam, NA 29 24.6 .061 .O73
. Medicine Bow, WY 5 29.0 .077 .027
Point Arena, CA 23 20.1 .072 .144
Russell, KS 29 23.4 .059 .040
San Gorgonio, CA 33 23.5 .088 .053

Composite Site 266 22.2 0.067 0.068

 0
I 2 q 6 8 I0 20 qO 60 80 100

WINDSPEEDATI0M,V_,M/S

Figurel - Various recommendationsfor the relationshipbetween

wind profileexponent and wind speed.

0.6 OL-Olo(1-
_5L06
v,)
0,5

MEAN
WIND O.q
PROFILE
EXPONENT,
E 0.3

0
I 2 q 6 8 ]D 20 qO 60 80 100

WINDSPEEDAT 10M,VR,MIS
Figure 2 - Graph of proposedequation for wind proflleexponent
in terms of surface roughnessand wind speed.

8
E
 1,0 D

0.8- /
SUBURBIA-._/,
// "'-CITY
0.6-
SURFACE PALMETTO-,.,y/,(.
ROUGHNESS _HIGN gOODS
HIGHGRASS-_ _'"Y//v"
EXPONENT, OA-- LogGRASS,STEPP_
0_, i _WN GRASS
-_, i._w_._"-- LOWW_DS
,L SAND_ _"_,_(_ --FALLOWFIELD

10"5 10"4 10-3 10.2 I0"1 1 10

SURFACE
ROUGHNESS
_NGTH.Zo,M

Figure 3 - Graph of relationshipbetween surface roughnessexponent _o'

surface roughnesslength, zo and terrain description.

• - MONTHLY
AVERAGE
WINDDATA
0.4-- • MONTHLY
PEAKWINDDATA j" WSSI,
CENTROID
O_MONTHLY
AVENGES 1976-78

0.)

0.2-
_• /r _ - 0.30(I-0.55
LOGV_)
/
WINS . •
EXPONENT, 0.i "

PROFILE ,,"
0 I I I I 1 'l _" I .-_',L I I
2 _ 6 8 10 20• • _0 60 80100

W;NDSPEE])
AT 10 M,VR,WS
-0.1

Figure4 - Comparisonof observed and calculatedwind profile

exponents for Clayton, New Mexico site.
 [_0 V - S-lOWS
} NEUTRALSTABILITY

: '_Jl: A _oN,_ j
:

WIND
.qO
I Zn

,0_ I ! I ] I
10.q 10-3" 10-2 10"1 1 10

SURFACE
ROUGHNESS
LENGTH,
Zo: M
u.

Figure 5 - Curve-fit of data relating wind profile exponent _(.

to surface roughness length.

O,S i
A- BLOCK
ISLAND,
I_I
•- PLUM
BROOK,OH
e- CLAYTON,
N_
0,_ -- _'- RUSSELL,
KS
D- CULEBRA,
PR
.. MONTHLY
AVERAGE G-OTHER(TABLE
2)

EXPONENI,
WIND
0,2 '.i_ MONTHLY
PEAK
FLAT,OPEN--//__i!i:_!_'_WI_S--

l I I I I I_1 . I _'ll_ i 1
0
1 2 q f) 8 10 20 qO r_ 80 10o

AT 10' _, VM,WS
W'NOSPEED

Figure 6 - Comparison of observed and calculated wtnd p_oftle

exponents for vartous sttes.

10