Energy Conversion & Management 41 (2000) 281298

www.elsevier.com/locate/enconman

Local design, testing and manufacturing of small mixed

airfoil wind turbine blades of glass ber reinforced plastics
Part II: Manufacturing of the blade and rotor
S.M. Habali a,*, I.A. Saleh b
a

Mechanical Engineering Department, Faculty of Engineering and Technology, University of Jordan, Amman, Jordan
b
Renewable Energy Center, Royal Scientic Society, Amman, Jordan
Received 14 November 1998; accepted 16 May 1999

Abstract
Wind energy has attracted a great deal of attention in recent years in Jordan as one of the possible
alternative renewable energy resources. Almost all of the local research and development activities in
this eld were directed to explore, develop and optimally utilize wind energy systems. The time has come
to establish a link between the local scientic (academic) work and local industries to produce a usable
technology which will increase the local share in an inevitably emerging wind energy industry in Jordan.
To achieve this goal, a well founded manufacturing base is required. In part I of this research, two
airfoil sections were chosen and the rotor blade was designed as a combination of these airfoils. In this
paper (part II), the blade and the blade rotor will be manufactured, tested and its performance
examined. The rotor was manufactured of GRP composite material and then installed on a 15 kW grid
connected, pitch controlled wind machine. The blade has also passed successfully all required tests and,
when installed on a functional wind turbine, performed satisfactorily with a favorable power coecient
of 41.2%. # 1999 Elsevier Science Ltd. All rights reserved.
Keywords: Mixed airfoil blades of CM; Mixed airfoil blades of GRP; A mixed-airfoil-blades rotor of GRP; Manufacturing of a mixed-airfoil-blades rotor of GRP

* Corresponding author.
E-mail address: habali@fet.ju.edu.jo (S.M. Habali)
0196-8904/00/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 9 6 - 8 9 0 4 ( 9 9 ) 0 0 1 0 4 - 1

282

S.M. Habali, I.A. Saleh / Energy Conversion & Management 41 (2000) 281298

Nomenclature
WEC
RSS
RNL
GRP
Cp
R
JD
O
l

wind energy converter

Royal Scientic Society
Risoe National Laboratory
glass ber reinforced plastic
power coecient
total rotor radius
Jordanian Dinar
angular speed of rotor
tip-speed ratio

1. Introduction
To establish a wind energy industry in Jordan, a founded manufacturing base is required,
which starts with rotor blade manufacturing. One Jordanian company has succeeded in
adapting this new wind energy technology and has produced a high quality 5 m long GRP
blade.
The method of manufacturing the proposed prototype blade will be described in three stages,
manufacturing the model, manufacturing the mold and manufacturing the prototype. Each
stage is independent from the other two, but in this prototype environment, the rst stage is
the most critical one from the strength point of view. This method is appropriate for blades of
this size category because the blade will be relatively light weight and can be easily handled.
For large blades, however, (when weight becomes a problem) other methods are used such as
winding on mandrels [1].
Manufacturing of the rotor blade is one of the prime objectives of this paper and is divided
into three integrated parts, the model, the mold and the prototype. The model is made of soft
white wood to simplify shaping the exact geometry, taper and twist. The model will also
facilitate verication and modication to certain parts of the blade especially ensuring
streamlining.
A mold having the exact imprint of the model is then produced from GRP material. The
mold will have two activities, upper and lower, to facilitate ease of extracting the molded
blade. The interior surfaces of the mold cavities are carefully treated and highly polished to
yield a ne smoothness of the blade surface. A method that is simple and practical is
developed for the local manufacturer to apply, which produces a strong and exact geometry
blade. The method also includes a proof load-test procedure. When applied and passed, it will
guarantee fatigue strength capabilities.
Four blades were manufactured. One was proof load tested and three were installed on a
grid connected wind turbine having rated power of 15 kW with pitch regulation and
electrohydraulic control. A special measuring system was installed between the turbine and the
grid that measures the power, the wind speed, the voltage and the frequency of the generator.
These values will be used to evaluate the performance of the rotor blade after proper

S.M. Habali, I.A. Saleh / Energy Conversion & Management 41 (2000) 281298

283

accounting for the various eciencies of the whole turbine system. The treatment of these data
can ultimately produce a power curve for the turbine, which leads to calculation of the overall
performance and nally leads to the performance of the single rotor blade.

2. Manufacturing of the rotor blade

The procedure of manufacturing the rotor blade, followed in this paper, consists of three
stages.
2.1. Manufacturing of the model
A full scale exact geometry blade model will be made from wood. This model will enable the
designer to investigate and check some parameters of the of the blade, especially continuity of
the twist and smoothness of the surfaces.
The rst step in making the model is to divide the blade length into a number of intervals,
keeping in mind that the data available is for the outer 4.2 m part of the blade (divided into 11
intervals with 12 positions) and the 0.8 m root part, as shown in Fig. 1. For these 12 airfoil
positions, the 12 airfoil sections given in Table 1 will be produced from 20 mm thick white
wood. The perimeter of each section will be reduced by 10 mm to allow for installation of
15  15 mm2 cross section wood sticks as shown in Fig. 2. These sticks, when stacked next to
each other and bonded to the 12 airfoil sections using bending resin and hidden nails, will
form the skin of the blade, which can be later shaped and smoothed and nally sanded to a
very ne nish.
The twist distribution given in Table 1 is incorporated in the blade model by xing the
sections at their respective stations as shown in Fig. 3, keeping in mind that the aligning datum
is the aerodynamic centerline of the blade which shall also pass through the center of the root
ange.
After all segments are installed and all nishing has been done, a completed full scale 5 m
rotor blade wooden model is obtained.

Fig. 1. Distribution of the chord length over the blade (in mm).

284

S.M. Habali, I.A. Saleh / Energy Conversion & Management 41 (2000) 281298

Table 1
Complete blade data and prole mixing
Station (m)

0.8
1.2
1.6
2.0
2.4
2.8
3.2
3.6
4.0
4.4
4.8
5.0
a

Chord (mm)

Thickness (mm)

600
562
524
486
448
410
371
333
295
257
219
200

178
163
149
134
120
105
91
76
61
47
32
25

Thickness (%)

Twist (8)

Prole
NACA 63-621 = N21
FXS 66-196 = F196

29.7
29.0
28.4
27.6
26.8
25.6
24.5
22.8
20.7
18.3
14.6
12.5

16.0
12.8
10.1
7.8
5.8
4.2
2.8
1.8
0.9
0.4
0.2
0.0

1.41
1.38
1.35
1.31
1.28
1.22
0.85
0.25
1.06
0.93
0.74
0.64

(N21)
(N21)
(N21)
(N21)
(N21)
(N21)
(N21) + 0.34 (F196)
(N21) + 0.90 (F196)
(F196)
(F196)
(F196)
(F196)

Same as in Table 1 of part I.

2.2. Manufacturing the mold

The mold will be made from GRP and consists of two parts: upper half and lower half
which are then strengthened using lateral rip stieners. Figs. 4 and 5 show the technique used
by the mold manufacturing.
It is of great importance to locate the line where the two mold halves meet to avoid
overlapping, which because of the blade twist will result in complicated surface. However, the
best guide is the chord line, which passes through the leading edge to the trailing edge of the
airfoil section.

Fig. 2. Airfoil section cut from wood and reduced to allow for skin segments builtup.

S.M. Habali, I.A. Saleh / Energy Conversion & Management 41 (2000) 281298

285

Fig. 3. Method of xing the airfoil sections in their respective chord and twist distribution prior to installing the
skin.

Fig. 4. Fiberglass and polyester resin laminates layed on the wooden model to get its shape.

286

S.M. Habali, I.A. Saleh / Energy Conversion & Management 41 (2000) 281298

Fig. 5. The two-mold halves after nishing are strengthened by lateral rips and steel frame.

2.3. Manufacturing the prototype blade

The blade is made of two shell halves and a central axial spar, whereby each shell is
fabricated in its respective mold half (upper or lower). The prototype blade is manufactured
rst. The two blade shells will be joined and glued together at the leading and the trailing
edges. Therefore, some means must be provided to allow enough area for adhesions as shown
in Fig. 6. For this purpose, the spar is used.
After curing, the blade is taken out of the mold and cleaned from any extra materials at the
edges. The root section will be prepared to be tted with the steel ange assembly. This ange

Fig. 6. Cross-section of upper and lower shell construction and area for adhesion allowance.

S.M. Habali, I.A. Saleh / Energy Conversion & Management 41 (2000) 281298

287

(proposed and designed in part I of this research) will be an integral part of the blade that
facilitates installation on the wind turbine system. A three-dimensional view of this steel ange
is shown in Fig. 7, where the central ange will be installed into the blade, allowing the spar to
go inside its bore and is completely glued. The outer two halves of this ange are then tted on
the outside surface of the root and also glued with great care to the ber glass and pressed and
tack welded together. Fig. 8 shows the completed 5 m GRP blade. It weighs 120 kg.
3. Proof load test
Four blades were manufactured. One was taken at random to be proof load tested using
load case 1 considered in part I of this research (load case 1: Py 9:33 kN/blade) acting at the
root section. A linear load distribution over the blade length is assumed, whereby q 0 kN/m
at the rotor center and qmax 3:424 kN/m at the rotor tip. This test is part of the
manufacturing process, and a condition was set forth to the manufacturer that the blade must
pass this test successfully, otherwise it will be rejected. However, the technique used for the
proof load test is shown in Fig. 9, whereby the blade was bolted in the apwise position from
the root ange to one of the vertical structural columns at the factory building and loaded
around 2/3 of its length from the root. The tip position was recorded rst under the blade's
own weight and then with the platform which will carry the load. The load was then applied in

Fig. 7. Steel ange conguration for the blade root.

288

S.M. Habali, I.A. Saleh / Energy Conversion & Management 41 (2000) 281298

Fig. 8. Photograph of the rst prototype GRP blade.

the y ' direction (perpendicular to the chord direction) in increments of 27 kg until the 640 kg
load was reached making a total load of 740 kg with platform and ropes. The blade passed
this test successfully with the load deection diagram shown in Fig. 10.
4. Performance testing
The performance of a rotor is measured by the amount of power it delivers during rotation
compared to the theoretical power available in the wind passing through the same area swept
by the blade. Obviously, the main parameters to be measured for this purpose are power and
wind speed, which will produce the power wind curve (PV curve). This blade is installed on
the rst 15 kW Wind Energy Converter (WEC) Prototype built by the RSS in 1992. A
photograph of this WEC is shown in Fig. 11.
Some of the specications for this WEC will be veried by performance tests, such as cut in
speed, rated wind speed, rated power and power coecient. Other parameters, like voltage,
current and frequency, must be measured and analyzed.
The acquisition of the PV curve will ultimately lead to production of a power coecient
curve (Cp curve) for this WEC, which is usually calculated in terms of the dimensionless tip
speed ratio (l) of the blade. The net performance of the wind turbine blade will be obtained
after deducting the eciencies of the various mechanical and electrical components in the
WEC.

S.M. Habali, I.A. Saleh / Energy Conversion & Management 41 (2000) 281298

289

Fig. 9. Proof-load testing of the prototype blade after manufacturing.

4.1. Testing strategy

The test methodology was established in a special study done by the RSS in 1985 during a
time when there was no agreement upon international standards for testing wind turbines [2].
The most recent reports [3,4], from the testing station for wind turbines at RISO, Denmark,
support the method used by the RSS, and therefore, they will be applied here for testing and
evaluating the RSS wind and, hence, the rotor blades.
The main objective of the test will be to obtain a set of performance curves for the WEC
which will be indicative of the overall performance of the individual blade. In this test, reduced
10 min averaged data sets are used in the construction and derivation of these curves. The
physical parameters measured for this purpose are: wind speed V, voltage U, frequency f and
power P. The independent variable for these measurements is the wind speed V. However, time
dependent curves can also be produced in order to see the variation in response and the
simultaneous eects among the variables. According to the European wind turbine standard on
performance determination [3], the data must be corrected to standard air density of 1.225 kg/
m3 and averaged for power values below 70% of the nominal power. Above 70% of the
nominal power, no correction is needed.

290

S.M. Habali, I.A. Saleh / Energy Conversion & Management 41 (2000) 281298

Fig. 10. Blade-load tip-deection curve.

4.2. Description of the measuring system

The measuring system used in this test is a data acquisition system utilizing electric
measuring transducers and a BCS-16 computer furnished with a terminal and a printer. The
measurements taken and the ow of data within the system are shown in Fig. 12 as given by
Table 2
Ranges for measured values
Item

Measured quantity

Measured value (volt DC)

Transducing range

1
2
3
4
5
6
7
8
9

Wind velocity
Total active power
Frequency
Current phase 1
Current phase 2
Current phase 3
Phase voltage 1
Phase voltage 2
Phase voltage 3

010
010
010
010
010
010
010
010
010

020 m/s
040 kW
0100 Hz
060 A
060 A
060 A
0300 V
0300 V
0300 V

S.M. Habali, I.A. Saleh / Energy Conversion & Management 41 (2000) 281298

291

Fig. 11. Photograph of the RSS 15 kW wind turbine prototype using the GRP blade manufactured here.

Amr et al. [3]. Fig. 13 shows the data acquisition system components and interactions, adapted
from RNL [5].
All the measured values are compiled in the form of a DC signal with values range from 0
to +10 V for further processing by the computer, see Table 2. The measuring system scans
these listed parameters every second for 600 s and then takes the statistical average and records

292

S.M. Habali, I.A. Saleh / Energy Conversion & Management 41 (2000) 281298

Fig. 12. Block diagram of the measuring system, adapted from Ref. [6].

it on a mass storage media as a single 10 min averaged data point for each measured
parameter. This function is continued around the clock, and at the end of a 24 h period, the
compiled 144 (10 min) averaged data points are recorded on a data disk.
4.3. Results of the measuring program
The data is treated by eliminating erroneous records, which usually result from the behavior
of the control system of the WEC. For example, when the power output production decreases
from the WEC below the acceptable limit set by the control strategy, the control system shuts
the WEC down for a pre-assigned time interval of 15 min. During this time the wind speed
might increase, and it is, of course, being recorded but with zero power from the turbine. This
data point will be eliminated and only wind speed records with power values are kept. The
data is then classied according to the wind speed classes up to 20 m/s, which is usually a
satisfactory range for performance purposes. The maximum cumulative error in the
measurements is 3% according to the manufacturer catalog.
Fig. 14 shows a true (virgin) power curve obtained from the original data, with no averaging
within each class. The reason that no curve tting was done is that as mentioned above, one
can see the great deviation in the data which can be traced back to the control system.

S.M. Habali, I.A. Saleh / Energy Conversion & Management 41 (2000) 281298

293

Fig. 13. Block diagram of the data acquisition system, adapted from Ref. [6].

However, the trend is clear, and a single line curve cannot be a true representation of the
WEC's behavior and ultimately the blade's, especially at low wind speeds.
This note came from long experience and observations of many WECs under test where
power production can have too many dierent values for the same wind speed class at the
unregulated part of the power curve below the rated power. Over the rated wind speed (which
is the regulated part of the power curve), the power values are seen to be closer within the
same wind speed class because the control system simply uses the blade pitch angle to lower

Fig. 14. PV curve of the RSS WEC shown in Fig. 11.

294

S.M. Habali, I.A. Saleh / Energy Conversion & Management 41 (2000) 281298

the lift force in order to keep the power level from exceeding a preset value which is usually
the rated value of the installed generator. With the availability of high wind speeds at this part
of the curve, power values tend to remain almost constant. The cut in speed and the rated
wind speed can be seen at 5 and 10 m/s, respectively. The rated power has exceeded the
installed generator capacity to a little over 16 kW. This is due to the nature of the
asynchronous motors where they tend to produce 10% more power when used as generators,
as is the case here, which is why these machines are favored by most commercial wind turbine
builders.
The line voltage and frequency behavior as a function of wind speed are also shown in Figs.
15 and 16, respectively. It is clear that below rated wind speed, the voltage and frequency can
have any value, including the normal values of 220 V and 50 Hz. However, it should be
reiterated that these are 10 min average values from 600 scans, where some scans could be
taken during normal operation, and the rest is taken during shut down because of below cut in
wind speed as explained earlier. This situation will result in a data point having values below
the nominal for above cut in wind speed classes. Nevertheless, nominal voltage and frequency
can be seen to start from cut in, which is also a condition for grid connection machines.
The eciency of a wind turbine is usually characterized by its power coecient Cp. But pitch
regulated wind turbines have variable Cp and, therefore, their eciency is best represented by a
Cp l curve, where
l

OR
V

R and O are constant, but V is variable according to the wind speed class. The Cp values are
calculated according to Eq. (2):

Fig. 15. Line voltage versus wind speed for the RSS WEC shown in Fig. 11.

S.M. Habali, I.A. Saleh / Energy Conversion & Management 41 (2000) 281298

295

Fig. 16. Line frequency versus wind speed for the RSS WEC shown in Fig. 11.

Cp

P
0:5rV 3 A

from the measured values. The results are shown in Fig. 17. From this gure, the WEC
reached a maximum Cp,max 33% at l 5. Again, the data points are more coherent at high
wind speeds (i.e., at low tip speed ratios) and more spread apart at low wind speed s (i.e., at
high tip speed ratios), but the theoretical shape of the curve is evident. The performance of the
rotor alone can be obtained by eliminating the eciencies of the following components from
the WEC:

Fig. 17. Cp l curve for the RSS WEC shown in Fig. 11.

296

S.M. Habali, I.A. Saleh / Energy Conversion & Management 41 (2000) 281298

drive train assembly: ZD:T: 0:95

generator: ZG 0:86
power line and electronics: ZE 0:98
Therefore, the isolated rotor performance, which is the blades' performance, will be:
Cp,blade

Cp
ZD:T: ZG ZE

:033=0:95  0:86  0:98 0:412

This value is comparable with the Bitz limit of 0.59 and is considered acceptable.
The performance data is presented on a time scale as shown in Fig. 18, which tells the
history of the power production and the time lag of the control. In this gure, a continuous

Fig. 18. Instantaneous measurements of the RSS WEC shown in Fig. 11.

S.M. Habali, I.A. Saleh / Energy Conversion & Management 41 (2000) 281298

297

string of data for 24 h is plotted without classication, where the variations of the wind speed
and where also the eect of the control can be noticed.
5. Conclusions and recommendations
The results of the measurements have produced the necessary data for evaluation of the
rotor blade performance. The single most important piece of information is the power curve.
Being installed on a pitch regulated wind turbine, the performance of this design of GRP
blades can only be seen in the part of the curve up to rated power (Prat 15 kW at 10 m/s).
Beyond this point, the performance of the blade is overshadowed by the control system which
literally dumps all excess power (over Prat), thereby indicating a decrease in the eciency as
indicated by the Cp l curve.
The power production characteristics of the blade are best seen in the part of the PV curve
from Vcut-in to Vrat, where the theoretical cubic behavior is obvious. From this part of the
curve where no pitch regulation occurs, the following characteristics can be seen:
1. The rotor blade can produce a wide range of power outputs within the same wind speed
class which is largely due to the production history. After cut in, the rotor will be `trying' to
produce power while following an increasing wind speed. This will produce power less than
expected for that given wind speed class. However, after a while in operation during highspeed winds, the rotor will (if the wind starts to calm down) be producing more power than
expected for that wind speed class. This explains the large scatter of data from Vcut-in to Vrat
values.
2. The scattering of power production is referred to a variety of reasons:
* Rotor inertia: during start up before cut in, the rotor requires higher torque to start
turning and production, but much less torque is needed to produce the same `unit' power
after it has run up to full speed.
* Yaw error: sometimes the rotor is not facing the wind stream as it should because the
wind tracking device cannot instantaneously follow the wind direction changes. This
situation will alter the wind vector conguration at the blade from the design settings to a
big drop in power output.
* Wind shear: the eect of this phenomenon is more clear with large rotors but still has a
remarkable inuence on small turbines, whether this shear is vertical or horizontal.
The performance of the rotor blade (Cp,max 0:412) is considered extraordinary, being the rst
prototype designed and manufactured in Jordan. This leads us to conclude that the design
procedure is valid, and all assumptions put forth are justied. This also leads us to conclude
that the manufacturing method is acceptable despite the fact that this was a one time
undertaking by the manufacturing company. The extreme proof load test has proven the
ability of the blade to sustain severe conditions during operation, and moreover, no vibrations
from imbalance or fabricating errors were detected throughout the testing program of the
WEC.In light of these ndings, the following recommendations are due:
. Some basic research has to be done on locally manufactured GRP materials to aid future
designers and encourage them to use this extraordinary composite in many daily uses.

298

S.M. Habali, I.A. Saleh / Energy Conversion & Management 41 (2000) 281298

. The cost of this 5 m long blade is JD 1250 (1 JD = $1.42), and according to the
manufacturer, this cost will be reduced by 3040% for mass production. The cost the mass
produced 7.5 m long Aeroman blade made in Germany by M.A.N. company is DM 10000
(= $6000 = JD 4200). The price length ratios give a rst judge of the economy of the two
blades. For the RSS blade, this ratio is 1250 JD/5 m = 250 JD/m, while for the Aeroman
blade the ratio is 4200 JD/7.5 m = 560 JD/m. Thus, similar blades can be made in Jordan
using this technique with only about a third of the cost. Therefore, it is recommended that
this blade design and manufacturing procedure be considered for future improvements.
. The real transfer of this technology resides in developing a local industry concerned with
wind energy in all its aspects. Therefore, any future development should be directed toward
producing blades of 10 m in length and above, or WECs of over 100 kW capacities in order
to be cost eective and more meaningful to investors.

References
[1] Boeing Engineering and construction. MOD-2 wind turbine system concept And preliminary design report. US
Department of energy/NASA 002.80/2, Seattle, Washington, 1979.
[2] Royal Scientic Society. Development of testing procedure, WP Report. RSS, Amman, Jordan, 1986.
[3] Amr et al. Testing and evaluation of the small stand alone wind farm for water pumping. WP Report, RSS,
Amman, Jordan, 1988.
[4] Risoe National Laboratory. Wind turbine test vestas V27-225 kW. Risoe-M- 2861, Roskilde, Denmark, 1990.
[5] Risoe National Laboratory. Contributions from the Department of Meteorology and Wind Energy to the
ECWEC'93 Conference in Travemuende, Germany, Risoe-R-683(EN), Roskilde, Denmark, 1993.
[6] Royal Scientic Society. Testing and evaluation of the small stand alone wind farm for water pumping. WP
report. RSS, Amman, Jordan, 1988.