Vous êtes sur la page 1sur 18

Computers and Structures 74 (2000) 649±666

www.elsevier.com/locate/compstruc

Structural design optimization of wind turbine towers


Hani M. Negm a, Karam Y. Maalawi b,*
a
Aerospace Engineering Department, Cairo University, Cairo, Egypt
b
Mechanical Engineering Department, National Research Center, Cairo, Egypt
Received 22 May 1998; accepted 18 February 1999

Abstract

This paper describes several optimization models for the design of a typical wind turbine tower structure. The
main tower body is considered to be built from uniform segments where the e€ective design variables are chosen to
be the cross-sectional area, radius of gyration and height of each segment. The nacelle/rotor combination is
regarded as a rigid non-rotating mass attached at the top of the tower. Five optimization strategies are developed
and tested. The last one concerning reduction of vibration level by direct maximization of the system natural
frequencies works very well and has shown excellent results for both tower alone and the combined tower/rotor
model. Extensive computer experimentation has shown that global optimality is attainable from the proposed
discretized model and a new mathematical concept is given for the exact placement of the system frequencies. The
normal mode method is applied to obtain forced response for di€erent types of excitations. The optimization
problem is formulated as a nonlinear mathematical programming problem solved by the interior penalty function
technique. Finally, the model is applied to the design of a 100-kW horizontal axis wind turbine (ERDA-NASA
MOD-0). It has succeeded in arriving at the optimum solutions showing signi®cant improvements in the overall
system performance as compared with a reference or baseline design. # 2000 Elsevier Science Ltd. All rights
reserved.

Keywords: Renewable energy; Wind turbine; Vibration; Tower structures; Dynamics; Optimization

1. Introduction be found that deals with the problem of structural de-


sign optimization. It is the main intent of the present
Many wind energy research and development pro- study to consider the basic aspects of design optimiz-
grams have been initiated since the 1973 oil crisis, and ation of the combined tower/rotor structure of a wind
di€erent con®gurations of wind turbines have been turbine. A good survey of the various theories and nu-
installed in many countries. These clean energy sources merical methods of structural design optimization can
can make a substantial and economically competitive be found in Refs. [3,4] which are usually classi®ed into
contribution to the future energy needs. Concerning two groups: the mathematical programming approach,
system design optimization, most of the published and the optimality criteria approach. The latter was
research work has mainly been focused on power-out- employed by Takewaki [5] who considered optimiz-
put economics and cost optimization [1,2]. Little may ation of a concrete tower structure with round tubular
cross section. An approximate concept was applied for
®nding the optimal bending sti€ness distribution which
* Corresponding author. minimizes the total structural weight for a speci®ed

0045-7949/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 4 5 - 7 9 4 9 ( 9 9 ) 0 0 0 7 9 - 6
650 H.M. Negm, K.Y. Maalawi / Computers and Structures 74 (2000) 649±666

fundamental natural frequency. Later on, Takewaki cantilevered to the ground, and is carrying a concen-
solved the same problem taking into consideration the trated mass at its free end approximating the inertia
¯exibility e€ect of the tower base connection [6]. He properties of the nacelle/rotor unit. This mass is
used ®nite element method and a piecewise-linear func- assumed to be rigidly attached to the tower top.
tion to approximate the sti€ness distribution along the . Material of construction is linearly elastic, isotropic
tower height. and homogeneous. The tower has a thin-walled cir-
In the present paper, several optimization strategies cular cross-section.
for the design of a wind turbine tower are developed . The Euler±Bernoulli beam theory is used for predict-
and tested, and di€erent forms of the objective func- ing de¯ections. Secondary e€ects such as axial and
tion are examined. The tower is considered to be built shear deformations, and rotary inertia are neglected.
from uniform round tubular segments (modules), . Distributed aerodynamic loads are restricted to pro-
where the e€ective design variables are the cross-sec- ®le drag forces. A two-dimensional (2D) steady ¯ow
tional area, radius of gyration and height of each seg- model is assumed.
ment. Isolated tower dynamics including the complete . Nonstructural mass will not be optimized in the de-
kinematical analysis and formulation of the applied sign process. Its distribution along the tower height
loads are treated in detail. A simpli®ed set of the gov- will be taken equal to some fraction of the structural
erning dynamical equations of motion is given in an mass distribution.
appropriate non-dimensional form, and an exact . Structural analysis is con®ned only to the case of
method for the determination of the dynamic charac- ¯apping motion (i.e. bending perpendicular to the
teristics and forced response is presented. The ®nal plane of rotor disk).
model formulation identi®es the optimum values of the
design variables which make the tower structure ful®l
its design objectives in the best possible manner while
2.2. Tower design objectives
satisfying all design constraints. The problem is formu-
lated as a nonlinear mathematical programming pro-
A wind turbine tower is the main structure which
blem. Optimum design solutions are obtained by the
supports rotor, power transmission and control sys-
interior penalty function method coupled with Powell's
tems, and elevates the rotating blades above the earth
multi-dimensional and quadratic-interpolation one-
boundary layer. A successful structural design of the
dimensional search techniques.
tower should ensure ecient, safe and economic design
Implementation of the ®nal model form to an exist-
of the whole wind turbine system. It should provide
ing wind turbine; the ERDA-NASA MOD-0 machine,
easy access for maintenance of the rotor components
is given and the optimum design trends are examined
and sub-components, and easy transportation and
and discussed. It is demonstrated that global optimal-
erection. Good designs ought to incorporate aesthetic
ity can be attained from the proposed discretized
features of the overall machine shape.
models, and exact placement of the tower frequencies
In fact, there are no simple criteria for measuring
is also achievable.
the above set of objectives. However, it should be
recognized that the success of tower structural design
is judged by the extent to which the wind turbine main
2. Optimization problem formulation
function is achieved. The di€erent optimization strat-
egies considered herein are discussed below.
In formulating an optimization problem, three prin-
cipal phases must be considered:
2.2.1. Light weight design
. De®nition and measure of the system design objec-
A minimum weight structural design is of para-
tives.
mount importance for successful and economic oper-
. Selection of the design variables and preassigned
ation of a wind turbine. The reduction in structural
parameters.
weight is advantageous from the production and cost
. De®nition of the design constraints.
points of view. For a tower composed of Ns segments
the weight (mass) function, to be minimized, can be
expressed in the non-dimensional form
2.1. Basic assumptions
X
Ns
minimize; Mˆ D k t k Hk …1†
. The basic structural model of the tower is rep- kˆ1
resented by an equivalent long, slender cantilever
beam built from segments (modules) having di€erent For the de®nition of the various parameters, refer to
but uniform cross-sectional properties. The tower is Table 1 and Table 2
H.M. Negm, K.Y. Maalawi / Computers and Structures 74 (2000) 649±666 651

2.2.2. High sti€ness criterion than maximization of the sti€ness alone or


The main tower structure must possess an adequate minimization of the structural mass alone. The related
sti€ness level. Maximization of the sti€ness is essential mathematical expression can be obtained by dividing
to enhance the overall structural stability and decrease Eq. (2) by Eq. (1).
the possibility of fatigue failure. For a cantilevered
tower, sti€ness can reasonably be measured by the
magnitude of a horizontal force applied at the free end 2.2.4. Design for minimum vibration
and producing a maximum de¯ection of unity. This Minimization of the overall vibration level is one
can be expressed mathematically as in Ref. [7]: of the most cost-e€ective solutions for a successful
 wind turbine design. It fosters other important design
X
Ns
Hk goals, such as long fatigue life, high stability and
maximize; S ˆ 1= 1 ÿ …x k‡1 ‡ x k †
kˆ1
Ik low noise level. Two di€erent criteria for measuring
…2† vibration reduction are stated in the following

1 
sections [7].
‡ x 2k‡1 ‡ x k x k‡1 ‡ x 2k
3
2.2.4.1. Frequency-placement criterion. Reduction of vi-
bration can be achieved by separating the natural fre-
2.2.3. High (sti€ness/mass)-ratio quencies of the structure from the exciting frequencies
Perhaps, maximization of the sti€ness-to-mass ratio, to avoid large amplitudes caused by resonance [8].
which is directly related to the physical realities of the This may be measured by minimizing the performance
design, is a better and more straight forward design index

Table 1
De®nition of reference beam parameters

Parameter Notation Description

Height H0 Tower reference height


Diameter D0 Mean value of the inner and outer diameters
Thickness t0 Wall thickness
Cross-sectional area (structurally e€ective) A0 A0 ˆ pD0 t0
Second moment of inertia of the cross section I0 I0  pD30 t0 =8:0 p
p
Radius of gyration r0 ˆ I0 =A0  D0 =2 2
Mass density r0 ˆ rT
Non-structural mass per unit length mns0 Not structurally e€ective in carrying stresses (e.g. stairs)
Structural mass/length ms0 r0 A0
Mass/unit length m0 ˆ ms0 ‡ mns0
Young's modulus E ET
652 H.M. Negm, K.Y. Maalawi / Computers and Structures 74 (2000) 649±666
X ÿ 2
min Wfi oi ÿ oi …3† work considers a di€erent optimization strategy by
i maximizing a weighted-sum of the system natural fre-
quencies. This is described mathematically by the fol-
where Wfi are weighting factors, oi are the actual fre- lowing linear composite form of the objective function
quencies and oi are the corresponding desired (target) [7]:
values chosen to be within close ranges (called fre- X
quency-windows) of those of a reference or baseline max Wfi Gbi …4†
tower design, which are well separated from the critical i

exciting frequencies. p


where Gbi ˆ oi are the non-dimensional frequency
parameters of bending vibrations (Eq. (14)).
2.2.4.2. Maximum frequency criterion. Another alterna-
tive for reducing vibrations is the direct maximization
of the system natural frequencies. Higher natural fre- 2.3. The preassigned parameters
quencies are favorable for reducing both of the steady-
The following tower variables will be given preas-
state and transient responses of the tower. Szyszkowski
[9] considered maximization of the fundamental fre- signed ®xed values in order to decrease the dimension-
quency for given structural weight via a multi-modal ality of the optimization problem:
optimality criterion. Another work dealing with opti- 1. tubular single-pole con®guration
mum shapes of beams for maximum fundamental fre- 2. total height, H
quency has been studied by Masad [10], who presented 3. type of cross section is thin-walled hollow circular
an ecient numerical method for computing the rate cross section
of change of the associated eigenvalues. The present 4. material of construction is chosen to be steel

Table 2
De®nition of non-dimensional quantitiesa

Quantity Notation Non-dimensional multiplier factor

Spatial coordinates (X, Y, Z )


Height H H0
De¯ections (U, V, W ) q
Time t …m0 H 40 †=…EI0 †

Spatial derivatives @ =@ X ˆ … †,x 1=H0


q
Time derivatives @ =@ t… † ˆ … †,t …EI0 †=…m0 H 40 †
Elastic rotations jx ,jy ,jz 1.0
Cross-sectional properties of the kth segment
Diameter Dk D0
Thickness tk t0
Area Ak A0
Area moment of inertia Ik I0
Mass/unit length mk m0
Mass moment of inertia/unit length Im m0 H 20
Total mass M,MR m0 H0
Mass density r m0 =H 20
Mass moment of inertia IM ,IR m0 H 30
Concentrated forces Fx ,Fy ,Fz EI0 =H 20
Concentrated moments Mx ,My ,Mz EI0 =H0
Distributed forces Px ,Py ,Pz EI0 =H 30
Distributed moments qx ,qy ,qz EI0 =H 20
Acceleration of gravity g …EI0 †=…m0 H 30 †
q
Natural frequency o …EI0 †=…m0 H 40 †
q
Damping factor gz m0 …EI0 †=…m0 H 40 †

a
E.g. Dim. (Fz)=Non-dim.(Fz)  (EI0/H20) or Fz 3 Fz  (EI0/H20).
H.M. Negm, K.Y. Maalawi / Computers and Structures 74 (2000) 649±666 653

2.4. Design variables nesses along the tower height in order to avoid local
instability.
The design variables which are subject to change
in the optimization process are chosen to be the 2.5.2. Side constraints
radius of gyration, cross-sectional area and height
of each module composing the main tower Dkl RDk RDku …10a†
structure. For a round thin-walled tubular con-
®guration, these correspond to the mean diameter, tkl Rtk Rtku , k ˆ 1,2, . . . ,Ns …10b†
wall-thickness and height of each module. In
other words, the design-variable vector Xd is de®ned where Dkl and Dku are the lower and upper limiting
as values of the kth diameter respectively, and tkl is the
 lower bounding value determined either from the mini-
Xd ˆ …D1 ,t1 ,H1 †,…D2 ,t2 ,H2 †, . . . ,…Dk ,tk ,Hk †, . . . , mum available sheet thicknesses or from considerations
…5† of local wall instability. The upper limiting value tku is
k ˆ 1,2, . . . ,Ns imposed in order that the thickness tk may not violate
the assumption of thin-walled tubular con®guration
which is of order …3Ns  1). (e.g. tku ˆ 0:1Dk ). The total tower height is assumed to
P
be equal to that of a reference design, i.e., Hk ˆ 1:
2.5. Design constraints In addition, the non-negativity constraints Hk r0 must
also be included to avoid having unrealistic negative
For the present model formulation, design con- heights.
straints are stated in the following section.

2.5.1. Behavioral constraints


2.6. Choice of the ®nal model form: functional behavior
Strength requirements
of the objective function
se,k
ÿ 1:0R0:0, k ˆ 1,2, . . . ,Ns …6†
sal Perhaps the most important part of the mathemat-
ical model is the objective function. Care must be exer-
Maximum de¯ection cised in constructing such a function because the
Wmax design recommended by the model will directly depend
ÿ 1:0R0:0 …7† on its form. In this section, we shall attempt to test the
Wal
di€erent alternatives of the objective function through
Resonance avoidance comprehensive computer implementation.
 
oi 2.6.1. Mass and sti€ness optimization analysis
1:0 ÿ Di R  R1:0 ‡ Di …8†
oi Using Eqs. 1 and 2, the level curves of the mass,
sti€ness and sti€ness-to-mass ratio can be generated
Mass limitation and plotted. Fig. 1 shows those of the sti€ness/mass
ratio of a two-segment cantilever model. In order to
…M=Mal † ÿ 1:0R0:0 …9†
minimize the mass alone below its reference value of
The various parameters and symbols of the above one the sti€ness is substantially degraded as it sinks to
inequalities are de®ned in the following: its lowest levels. The opposite trend is true, in order to
raise the sti€ness level above the reference value of 3.0,
se,k maximum e€ective axial stress within the kth unacceptable high values for the mass can be reached
module and, hence, unrealistic cost of fabrication and pro-
sal allowable tensile or compressive stress duction. Examining the well-behaved sti€ness/mass
Wmax maximum tip de¯ection function, it is seen that the middle region of the design
Wal an allowable value of de¯ection space encompasses the global optimum solutions where
oi target (or desired) frequencies that are well balanced improvement in both mass and sti€ness is
separated from the exciting ones attained. Several other con®gurations including two,
Di allowable tolerances of frequencies three and four segments have been implemented and
Mal allowable upper limiting value of the structural tested. It has been con®rmed that maximization of the
mass sti€ness-to-mass ratio is a much better and straight
forward measure of the system performance. Minimiz-
Regarding buckling stresses, the present simpli®ed ation of the mass alone or maximization of sti€ness
analysis only imposes lower bounds on the wall thick- alone requires imposition of lower limits on sti€ness or
654 H.M. Negm, K.Y. Maalawi / Computers and Structures 74 (2000) 649±666

upper limits on mass, respectively. The choice of the well behaved and continuous in the variable D1. Now
values of such limits is not easy. it is possible to choose prescribed values for Gbi and
either one of the variables r1 …ˆ D1 † or H1 and solve
2.6.2. Exact frequency optimization analysis for the remaining variable. Extensive computer analy-
Vibration reduction may be achieved either by sep- sis for numerical frequency solutions has proved the
arating the natural frequencies from the exciting ones validity of such a mathematical concept. It is thus
or by directly maximizing these frequencies. Whatever possible to freely place the natural frequencies at any
is the approach taken, it is useful to begin with study- desired values, but of course within prescribed mass
ing the behavior of the system natural frequencies and limitations. This useful conclusion has been con®rmed
see how they are changed with the selected design vari- by consideration of tower models built of more than
ables. one segment [7]. Typical results showing the associated
mass and frequency isomert curves are depicted in Fig.
2.6.2.1. Frequency behavior of tower alone: new math- 3 and Fig. 4.
ematical concept. Consider the simplest case of a uni-
form one-segment cantilever beam with non- 2.6.2.2. Frequency optimization of the combined tower/
dimensional height H1 and radius of gyration r1. The rotor structure. Two major factors should be included
associated transcendental equation of bending frequen- when considering frequency analysis of a wind turbine
cies can be shown to have the form tower [11]. The ®rst factor is the tip mass approximat-
    ing the inertia and mass properties of the nacelle/rotor
Gb H1 Gb H1 unit at the top of the tower. The computed results for
cos p cosh p ˆ ÿ1 …11†
r1 r1 a selected case is shown in Fig. 5. It is remarked that
the frequency behaves in a manner di€erent from that
Fig. 2 shows variation of the fundamental frequency of the case of the tower alone because the tip mass has
p
parameter Gb1 ˆ o1 with r1 ˆ D1 and H1 ˆ 1 for the considerable contribution to the total system mass.
round thin-walled tubular con®guration. The mass-to- The second factor is the axial compressive gravity
thickness ratio, M=t1 , is also shown in the ®gure. It is forces which may destabilize the tower in case of large
seen that both of the frequency and mass functions are wind generator systems. Fig. 6 shows variation of the

Fig. 1. Sti€ness/mass isomert curves for a two-segment tower model (H1 = H2 = 0.5, D1 = t1 = 1.0).
H.M. Negm, K.Y. Maalawi / Computers and Structures 74 (2000) 649±666 655

fundamental frequency with the non-dimensional grav- advantages of the present mathematical formulation.
ity parameter, g, for di€erent values of the tip mass, In addition, computer experimentation has shown that
MR : Variation of slenderness ratio, H0 =r0 , is also minimization of the frequency-placement index of Eq.
shown in the ®gure for di€erent tower heights. It is (3) results in a more time consuming optimization pro-
seen that a quick reduction in the frequency occurs cess and a slow rate of convergence towards the opti-
when the axial compressive stress becomes close to the mum solution. Besides, the attained optimum solutions
critical buckling stress of the tower. The calculated nu- were found to be strongly dependent on the chosen
merical values of the corresponding critical gravity fac- values of the target frequencies, which are rather arbi-
tor, gcr , are approximately equal to 1.1, 2.0, 3.35, and trarily chosen. Finally, it was decided to discard the
7.8 for MR ˆ 2:0, 1.0, 0.5, and 0.0, respectively. constraints on the module's diameters because it was
found that optimization favors increasing diameters
for maximizing frequencies.
2.6.3. Final model form
Di€erent formulations of the optimization model
have been tested on the computer using the actual data 2.6.4. Determination of the weighting factors
of the ERDA-NASA-MOD-0 wind turbine [7]. It was There are many procedures reported in the literature
found that the most signi®cant and promising optimiz- that help extract the values of the weighting factors,
ation strategy is the direct maximization of the system Wfi of Eq. (4). For instance, one may give the funda-
natural frequencies (Eq. (4)). Frequency-windows con- mental frequency the largest weight, while ignoring the
straints (Eq. (8)) were discarded from the model for- importance of other higher frequencies. A more
mulation because of the diculty in ®nding a feasible reliable procedure of adjusting the values of Wfi is
starting point in the selected design space. If it hap- given in Ref. [7] which is summarized as follows:
pened that one of the critical forcing frequencies were
. Initialize Wfi by the reciprocal of the values of the
close to any one of the natural frequencies of the opti-
frequencies of a reference uniform tower,
mized tower structure, another value for the natural
frequency can be chosen near its optimum value and Wfi ˆ 1=…Gbi †reference
the associated eigenvalue problem is resolved for any
one of the design variables instead. This is one of the . Normalize the resulting values by dividing each one

Fig. 2. Variation of frequency and mass with cross-sectional diameter (case of uniform one-segment model).
656 H.M. Negm, K.Y. Maalawi / Computers and Structures 74 (2000) 649±666

Fig. 3. Mass and frequency level curves of two-segment towers. (a) Gb1-contours, height vs diameter (D1=1, t1=1, t2=0.5).
(b) Mass contours, height vs diameter (D1=1, t1=1, t2=0.5). (c) Gb2-contours, thickness vs height (D1=1, D2=0.5, t1=1).
H.M. Negm, K.Y. Maalawi / Computers and Structures 74 (2000) 649±666 657

Fig. 3 (continued)

by their sum in order to make the ®nal sum equal to damping contributions. The ®nal approximate govern-
one, i.e. ing partial di€erential equation for the bending displa-
X cement W is cast in the following non-dimensional
Wfi 3Wfi = Wfi form:

Therefore, the ®nal appropriate values of the weight- …IWxx †xx ÿ…Fx Wx †x ÿPz ˆ 0 …12†
ing factors are given as:
* Tower alone: 61.0, 24.4, and 14.6%.
which must be satis®ed over the tower height, i.e.
* Combined tower/rotor: 61.2, 23.6, and 15.2%.
0RxR1: The non-dimensional equivalent distributed
load Pz can be calculated from [7]:

Pz ˆ ÿmT wtt ÿ gz wt ‡ Paero ‡ FZNB d…x ÿ 1 †


3. Tower analysis
‡ MYNB Z…x ÿ 1 † …13†
This section describes brie¯y the method of analysis
where the concentrated tip force FZNB and moment
of the bending motion of an isolated tower/rotor struc-
MYNB transmitted from the NB -rotating blades are
tural model. Fig. 7 shows the intermediate coordinates
transformed to distributed loads by the use of the
and degrees of freedom required to de®ne motion. The
Dirac-delta, d, and the unit doublet, Z, functions. The
associated transformation matrices and the complete
distributed aerodynamic force is given by [12]
kinematical analyses can be found in reference [7]
where the more general case of bending-bending-tor- Paero ˆ 4rair V 20 DCD …x†2ax …H0 =D0 †f…t† …14†
sional motion is given in detail. Distributed-load vec-
tors along tower height are expressed in the where f …t† is a time-dependent function accounts for
undeformed inertial coordinate system …x,y,z† and the dynamic and gusty nature of the wind, ax is called
include aerodynamic, inertial, gravitational as well as the wind shear exponent, CD is the drag coecient and
658 H.M. Negm, K.Y. Maalawi / Computers and Structures 74 (2000) 649±666

V0 is the wind speedp


at hub height H0 non-dimensio- which must be satis®ed over any module (k ) extending
nalized by the term …E=r†T : The associated boundary from x ˆ x k to x ˆ x k‡1 or 0RxRH  ˆ
k , where x
conditions are: x ÿ x k is a local coordinate system. fxk is a non-dimen-
At x = 0 (cantilevered end) sional quantity representing compressive force due to
gravity and is given by
W ˆ 0 and Wx ˆ 0:
!
At x = 1 (free end)   1 XNs
fxk ˆ gk x ÿ fk0 , fk0 ˆ MR ‡ Ai Hi ,
Ak iˆk …17†
… ÿ IWxx †x ‡Fx Wx ‡ MR Wtt ˆ 0
gk ˆ g=r2k and ok ˆ oi =rk
… ÿ IWxx † ÿ IR …Wx †tt ˆ 0 …15†
oi and gi are the non-dimensional eigenvalues (natural
frequencies) and eigenfunctions (mode shapes), respect-
3.1. Natural vibration analysis: the eigenvalue problem ively. For thin-hollow circular cross section, the non-
dimensional properties are:
The ®rst step in the solution procedure is the deter-
mination of the free natural vibration characteristics
by removing all of the forcing functions and consider- Ak ˆ Dk tk , Ik ˆ D3k tk , and rk ˆ D k : …18†
ing only the homogeneous equation. The resulting
eigenvalue problem is described by the 4th order ordin- The above di€erential equation has been shown to
ary di€erential equation: have an exact solution by the method of Frobenius in
ÿ 0 which the spatial functions gi …x† is represented by a
gi0000 …x † ÿ fxk gi0 …x † ÿo2k gi …x † ˆ 0 …16† four-term power series [7]:

Fig. 4. Fundamental frequency for towers built of four segments.


H.M. Negm, K.Y. Maalawi / Computers and Structures 74 (2000) 649±666 659

X
4
anm are the unknown coecients given by
gi …x† ˆ Cn ln …x†,
nˆ1
…19† gk fk0 an,mÿ2 gk …m ÿ 4 †an,mÿ3
X
1 anm ˆ ÿ ‡
ln …x† ˆ anm …x† mÿ1
, mrn …m ÿ 1 †…m ÿ 2 † …m ÿ 1 †…m ÿ 2 †…m ÿ 3 †
mˆn
o2k an,mÿ4
‡ mrn …20†
where ln are four linearly independent solutions and …m ÿ 1 †…m ÿ 2 †…m ÿ 3 †…m ÿ 4 †

Fig. 5. Fundamental frequency and mass isomert curves for the combined tower/rotor structural models. (a) One-segment model.
(b) Three-segment model.
660 H.M. Negm, K.Y. Maalawi / Computers and Structures 74 (2000) 649±666

Since an exact solution is available for one uniform erning equation of motion and the associated bound-
segment, then a non-uniform tower structure built ary conditions (Eq. (12)) results in N-independent set
from Ns -uniform segments must have an exact solution of uncoupled di€erential equations of the form
for its natural frequencies and mode shapes. The trans-
fer matrix method has been found to be suitable for x i …t† ‡ 2Zi oi x_ i …t† ‡ o2i xi …t† ˆ Qi =Mi ,
the present discretized model formulation. Application …23†
of the associated boundary conditions and consider- i ˆ 1,2, . . . ,N
ation of the non-trivial solution yields the following
exact frequency equation: where Qi is called the ith generalized load given by
…1
a1 o4i ‡ a2 o2i ‡ a3 ˆ 0 …21† Qi …t† ˆ gi …x† Paero …x,t† dx ‡ gi …1 †FZNB
0 …24†
where the coecients ai are de®ned as
ÿ gi0 …1 † MYNB
a1 ˆ IR MR …T23 T14 ÿ T24 T13 †,
Assuming zero initial conditions, the response is given
by:
a2 ˆ IR …T24 T43 ÿ T23 T44 † ‡ MR …T34 T13 ÿ T33 T14 †,
…t
 
xi …t† ˆ …1=Mi oid † Qi …t†exp ÿ Zi oi …t ÿ t†
a3 ˆ …T33 T44 ÿ T34 T43 † …22† 0

Tij are the elements of an overall transfer matrix  sin oid …t ÿ t† dt: …25†
formed by taking the products of all the intermediate p
transfer matrices of the di€erent segments. Detailed where oid ˆ oi 1 ÿ Z2i is the ith damped natural fre-
derivations can be found in Ref. [7]. Once the natural quency and Mi is the ith generalized mass de®ned as:
frequencies oi , i ˆ 1,2, . . . ,N, have been determined, …1
the associated mode shapes gi …x† can be obtained from Mi ˆ mT g2i dx ‡ MR g2i …1 † ‡ IR gi0 2 …1 † …26†
Eqs. (19) and (20). 0

3.2. Forced response 3.3. Internal loads and stress analysis

Application of the normal mode method to the gov- After the response has been found, displacements,

Fig. 6. Variation of fundamental frequency with gravity (case of one-segment tower with tip mass).
H.M. Negm, K.Y. Maalawi / Computers and Structures 74 (2000) 649±666 661

velocities and accelerations can be substituted for the phase of the supporting tower structure were investi-
calculation of the applied distributed loads. The in- gated in detail in Ref. [7]. They include the cases of
ternal forces and moments can then be determined at suddenly applied severe wind, transient loads induced
any location x along the tower height by integrating by tower base motion and vortex-shedding induced vi-
the equilibrium equations of motion. Stresses resulting brations.
from axial, bending, torsion and shearing e€ects can
be systematically calculated at any point within the
structure. For the present structural model of round 4. Optimization analysis by mathematical programming
tubular con®guration the most signi®cant stress com-
ponents are sxx and txy : Other components are usually The tower optimization model developed in the pre-
neglected. Stresses are calculated within the limitations vious sections is a non-linear mathematical program-
of the engineering beam theory of bending and torsion, ming model. The model uses the interior penalty
and the combined stress is determined on the basis of function technique [13] in ®nding the constrained opti-
Von Mises equivalent stress theory. mum solution. In contrast to the exterior technique,
the interior technique presents a series of improving
3.4. Other critical loading conditions feasible designs which gradually approach the ®nal op-
timum design. For multi-dimensional search the model
The most important critical loading conditions uses Powell's technique. The quadratic interpolation
which should be taken into consideration in the design method has been chosen for the determination of the

Fig. 7. Isolated tower structural model.


662 H.M. Negm, K.Y. Maalawi / Computers and Structures 74 (2000) 649±666

Table 3 Table 5
Parameters of the reference cantilever beam (thin-walled Baseline design variables. (Tubular construction)
round tubular con®guration)
Segment …Dk ,tk ,Hk †, k ˆ 1,2, . . . Ns
Parameter Notation (units) Numerical values
1 (1.4460, 1.25, 0.233)
Height H0 (m) 30.0 2 (1.1360, 1.05, 0.183)
Mean diameter D0 (m) 2.421 3 (0.9290, 1.00, 0.167)
Wall-thickness t0 (m) 0.01 4 (0.8260, 0.85, 0.150)
Cross-sectional area A0 (m2) 76:058  10ÿ3 5 (0.6196, 0.75, 0.133)
Second moment of inertia I0 (m4) 557:243  10ÿ4 6 (0.5163, 0.55, 0.134)
Radius of gyration r0 (m) 0.856 Rotor mass, MR 0.5
Material of construction Rotor inertia, IR 0.01
Type: Steel Tower mass, M 1.0
Mass density r0 (kg/m3) 8:7  103 Gravity, g 0.019
Young's modulus E (N/m2) 21:0  1010
Allowable normal stress sa (N/m2) 1:6  108
Allowable shear stress ta (N/m2) 0:8  108
Poisson's ratio n 0.3 5.1. The reference cantilever beam and baseline design
Structural mass/length ms0 (kg/m) 661.667
data
Non-structural mass/length mns0 (kg/m) 181.667
Total mass/length m0 (kg/m) 843.333
Total structural mass Ms0 (kg) 19850.0
Table 3 presents the pertinent data of the reference
Total non-structural mass Mns0 (kg) 5450.0 cantilever beam which has a round tubular cross-sec-
Total mass M0 (kg) 25300.0 tion type. Its total mass is chosen to be the same as
that of the original MOD-0 con®guration. Table 4
gives the associated values of the non-dimensional mul-
tiplier factors of the various quantities. The equivalent
step size along the search direction. A computer pro- discretized model of the baseline tower design is deter-
gram was developed to automate the tower analysis mined on the basis of having the same total mass and
and design procedures. Detailed description of the pro- sti€ness and mass distributions as that of the actual
gram can be found in Ref. [7]. MOD-0 design. It is selected to be composed of six
segments with the non-dimensional design variables
given in Table 5.

5. Model implementation: a case study 5.2. Results of structural analysis

As a case study, the proposed optimization model Fig. 8 depicts the calculated frequencies and mode
was implemented on an experimental, horizontal axis shapes of the reference cantilever beam and baseline
wind turbine; namely, the ERDA-NASA MOD-0 [11]. designs. It is seen that the ®rst and second frequencies
Its rotor has two blades with 38.1 m diameter located of the baseline design are higher than those of the
downwind from the tower at 30 m height above the reference cantilever indicating a good tower structural
ground. The original tower con®guration is a steel±lat- con®guration. The third frequency, however, is slightly
tice truss structure having four legs connected together less than its reference value. The fundamental fre-
by chord members. quency of the combined tower/rotor structure is about

Table 4
Non-dimensional multiplier factors

Non-dimensional quantity Notation Multiplier factor Units

Concentrated moments Mx ,My ,Mz 3:90  108 Nm


Concentrated forces Fx ,Fy ,Fz
Distributed moments qx ,qy ,qz 1:30  107 N
Distributed forces px ,py ,pz 4:333  105 N/m
Natural frequency o 4.1389 sÿ1
H.M. Negm, K.Y. Maalawi / Computers and Structures 74 (2000) 649±666 663

1.82 Hz …ˆ 4:1389  1:660782 =2p), which is higher value of 70.0 m/s. Two di€erent optimum designs were
than the primary forcing frequency 2/rev showing a obtained by considering two types of the objective
sti€ baseline design. The peak values of the concen- function. The ®rst type, type-1, considered maximiza-
trated loads transmitted from the rotor at the top of tion of the ®rst three bending frequencies, Gbi , for the
the tower were calculated at a rated wind speed of 8.0 combined tower/rotor structure, whereas the second
m/s. type, type-2, considered maximization of the frequen-
The associated maximum stress at the tower base cies of the tower alone. Fig. 9 presents the attained op-
and de¯ection at the top were found to be 0:268  108 timum tower designs for the ERDA-NASA MOD-0
N/m2 and 0.02844 m, respectively, which are much less wind turbine by implementing each of the above two
than the allowable values. On the other hand, consid- formulations. As seen, the optimum tower con®gur-
ering critical loading conditions discussed above in ation produced from the ®rst type consists of six-mod-
Section 3.4, the maximum stress at tower base has ules while that from the second type consists of eight-
reached a value of 0:6523  108 N/m2 for the overload modules, although the starting design was the same for
design condition with wind velocity V0 ˆ 27 m/s. The both cases. It was found that the optimization algor-
tip de¯ection increased to 0.0658 m. ithm treated the number of modules as an additional
implicit variable. Sometimes the computer discarded
5.3. Optimum tower solutions one or more segments by letting their heights sink
towards their lower limits (i.e. zeros), and sometimes
In this section we shall consider, as a case study, made two consecutive modules identical, i.e. having
tower optimization when subjected to a suddenly the same diameter and thickness. It is also remarked
applied wind with a velocity equal to the survival that type-1 formulation results in a higher frequency

Fig. 8. Natural frequencies and mode shapes of reference and baseline designs.
664 H.M. Negm, K.Y. Maalawi / Computers and Structures 74 (2000) 649±666

level than type-2, with the penalty of increasing struc- hand, type-2 formulation produces a slightly softer de-
tural mass. Both types produce substantial improve- sign where mass saving has reached a value of 12.4%
ments in mass and sti€ness as compared with those of but with increased tip de¯ection to 0.0563 m.
the baseline design described in Table 5. Mass saving The optimum values of the second mode frequencies
achieved from type-1 is shown to be about 7.6% and obtained from type-1 were too close to 29/rev and 49/
the calculated maximum de¯ection at tower top is rev for the combined system and tower alone, respect-
0.0393 m, which is much less than the corresponding ively. Even though they are far away from the primary
value of 0.097 m of the baseline design. On the other forcing frequencies (1/rev, 2/rev, 3/rev), they have been

Fig. 9. Optimum tower con®gurations of MOD-0 wind turbine (round tubular sections).
H.M. Negm, K.Y. Maalawi / Computers and Structures 74 (2000) 649±666 665

Table 6
Comparisons of MOD-0 tower designs

Tower design Fundamental frequency (Hz) Mass saving Maximum de¯ection m, (V0 = 70 m/s)

Tower alone Combined system

Original space truss 5.1 1.59, 1.76, 2.1 (di€erent rotor positions) ± ±
Baseline cantilever 5.06 1.82 0.0 0.097
Optimum cantilever 7.92 3.06 7.60% 0.039

separated by applying the previously suggested math- natural and exciting frequencies are normalized with
ematical concept. That is, instead of solving the eigen- respect to the rotor design speed, O0 ˆ 4:2 rad/s.
value problem in the natural frequencies, we choose
the desired frequency and solve the eigenvalue problem
in one of the design variables. The resulting modi®ed
6. Conclusions
optimum frequencies were given by:
Combined system Gbi ˆ 2:14344,5:44208,9:17620: Several optimization strategies for the structural de-
sign of wind turbine towers are developed and investi-
n=rev ˆ 4:53961,29:26349,83:19975: gated through extensive computer implementations.
The developed models have been successfully applied
Tower alone Gbi ˆ 3:43703,7:01730,11:08690:
to an existing 100 kW wind turbine (ERDA-NASA,
MOD-0), where optimum design trends were obtained
n=rev ˆ 11:67249,48:65001,121:45533: through the use of the interior penalty function tech-
nique. It has been proved that maximization of a
weighted-sum of the system natural frequencies is the
which are seen to be well separated from the exciting most representative objective function which directly
frequencies. re¯ects the major design goals and ensure a balanced
Comparison between the present optimum solution improvement in both mass and sti€ness. Extensive
and the original design of MOD-0 tower is given in computer experimentation has proved that the natural
Table 6. The frequency interference diagram (Camp- frequencies, even though implicit functions in the de-
bell-diagram) is presented in Fig. 10, in which both the sign variables, are well-behaved, monotonic and

Fig. 10. Campbell frequency-diagram of combined tower/rotor. ERDA-NASA MOD-0 wind turbine …O0 ˆ 40 rpm).
666 H.M. Negm, K.Y. Maalawi / Computers and Structures 74 (2000) 649±666

de®ned every where in the design space. Global optim- [3] Levy R, Lev OE. Recent developments in structural op-
ality can be attained from the proposed models and timization. J Structural Engineering 1987;ASCE-
exact placement of the frequencies can be obtained by 113:1939±62.
freely selecting their desired values. Appropriate non- [4] Grandhi R. Structural optimization with frequency con-
straints: a review. AIAA Journal 1993;31(12):2296±303.
dimensionalization of the various parameters and vari-
[5] Takewaki I. Optimal frequency design of tower struc-
ables throughout the problem formulation has led to a tures via an approximation concept. J. Computers and
naturally scaled optimization model, which eliminates Structures 1996;58(3):445-52.
the need for scaling the design variables as usually [6] Takewaki I. Ecient optimal frequency design of elasti-
suggested by similar optimization procedures. Another cally supported distributed-parameter cantilevers. J.
useful conclusion is the possibility of selection of the Computers and Structures 1997;62(1):107-17.
module height as a main design variable. This import- [7] Maalawi KY. Structural design optimization of wind tur-
ant variable is always missed by most of the previous bines. Ph.D. Dissertation, Faculty of Engineering, Cairo
research work dealing with structural optimization University, 1997.
where the cross-sectional parameters are considered the [8] Thresher RW. Structural dynamic analysis of wind tur-
only e€ective design variables. Finally, the present bine systems. J. Solar Energy Engineering 1982;104:89±
95.
exact optimization analysis saves much of the compu-
[9] Szyszkowski W. Multimodal optimality criterion for
ter time required by the ®nite-element and other discre- maximum fundamental frequency of free vibrations. J
tized approximate methods. Computers and Structures 1991;41(5):909±16.
[10] Masad JA. Optimization methods with structural
dynamic applications. J Computers and Structures
References 1997;62(3):521±5.
[11] Shamis CC, Sullivan TL. Free vibrations of the ERDA-
[1] Bae HM, Devine MD. Optimization models for the econ- NASA 100 kW wind turbine, NASA TMX-71879, 1976.
omic design of wind power systems. J Solar Energy [12] Simiu E, Scanlan RH. Wind e€ects on structures. New
1978;20:469±81. York: Wiley, 1978.
[2] Cromack DE, Oscar D. Design optimization of small [13] Gallagher, Zienkiewicz OC. Optimum structural design:
wind turbines for low wind regimes. J. Solar Energy theory and applications. New York, Toronto: Wiley,
Engineering 1984;106:347±50. 1973.

Vous aimerez peut-être aussi