Vous êtes sur la page 1sur 10

# Engineering Structures 60 (2014) 189–198

## Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

## Analytical formulation and solution of arches deﬁned in global

coordinates
F.N. Gimena, P. Gonzaga, L. Gimena ⇑
Department of Engineering Projects, Public University of Navarre, Campus Arrosadia C.P. 31006, Pamplona, Navarre, Spain

a r t i c l e i n f o a b s t r a c t

Article history: This work deals with the arch deﬁned in global coordinates. It shows the procedure followed to obtain its
Received 13 May 2013 formulation from the differential system of a curved beam. This procedure consists mainly in applying in
Revised 26 November 2013 the differential system, besides the habitual assumptions of the Strength of Materials, the simpliﬁcations
Accepted 9 December 2013
of geometric conditions of the planar curves. The resulting model includes the analysis of arches of var-
Available online 23 January 2014
iable section under any action force, moment, rotation or displacement in its plane. The successive inte-
gration of the equations of the system permits obtaining the solution that represents the structural
Keywords:
behavior of the arch under any type of support. Applying the boundary conditions of each problem it
Structural analysis
Curved beam
is obtained directly the exact analytical solution. Examples of calculus of parabolic arches are given to
Arches show the practical viability of the procedure followed. Results presented in graphs and tables are com-
Global coordinates parable with those obtained in the literature. The method is suitable for educational purposes.

1. Introduction frame and with independent variable the arc length [30] or with
another variable [31], taken into account shearing deformations,
The mechanical behavior of a curved beam, applying the theo- variable section, asymmetrical section, generalized loads and any
ries of Euler–Bernuolli or Timoshenko, is expressed usually support condition. Later, this general formulation was expressed
through the equations of equilibrium, constitutive relationships under a system of global coordinates and with the arc length as
and compatibility equations [1–4] or through compact equations independent variable [32], with the objective of representing and
of energy [5–12]. These two ways to annotate the structural behav- interpreting more efﬁciently, results of internal forces and dis-
ior have permitted offering results, analytical and/or numerical, placements. In this later document there were shown the advanta-
only for certain type of pieces. The planar pieces [13–16] more ges of this new formulation compared with the formerly. The
widely studied are with circular axis-line [17,18], parabolic [19– principal advantage is the lower triangular nature of the system
21] and elliptical [22]. The twisted beam more studied is with cir- annotated, that permits the obtaining of analytical results through
cular helix axis-line [23–26]. Also the analysis of a curved beam successive integrations row by row.
could be concreted using a system of twelve linear ordinary differ- As a novelty, the present paper focuses on solving the problem
ential equations [27–29]. This joint statement has permitted ﬁnd- of the arch, because it is a less common type of piece, but of high
ing new resolution procedures and thereby broadening the types of interest in the design and in analysis of structures. From the differ-
pieces to study. But still analytical procedures are limited depend- ential model in global coordinates, we obtain the exact analytical
ing on the complexity of the shape of the axis-line and section of solution. To do this, ﬁrst a particular formulation is provided from
the piece, the characteristics of the material, the action system the general system, representing the arch of symmetrical cross-
and the type of support. Both in the case of approaching the study section loaded into its plane taken into account the axial and
of the pieces by a single differential system or through separate shearing deformation and with whatever independent variable.
equations (equilibrium and compatibility), functions are expressed Sometimes, the length of the arch that deﬁnes the axis-line is not
in natural coordinates using the Frenet reference system and the the fundamental variable of the design. Is remarkable in the formu-
variable independent used is the arc length. lation and in the examples of this article, the employment of a gen-
Authors that subscribe this investigation, published a general eric parameter that can be particularized in function of the
formulation of curved beam elements expressed in the Frenet geometry of the piece element.
The formulation is particularized and solved analytically in two
cases of arches which its axis-line is parabolic without taken into
⇑ Corresponding author. Tel.: +34 948 169225; fax: +34 948 169644. account the shearing deformation. In the ﬁrst example the section
E-mail addresses: faustino@unavarra.es (F.N. Gimena), lazaro.gimena@unavarra.
is of hyperbolic variation. Data have been chosen to be compared
es, lazarogimena@hotmail.com (L. Gimena).

http://dx.doi.org/10.1016/j.engstruct.2013.12.004
190 F.N. Gimena et al. / Engineering Structures 60 (2014) 189–198

with those obtained in the cited literature. In the second example with unit vectors i, j and k instead of unit vectors tangent t, normal
the section is constant and shows how the procedure permits to n and binormal b, through the direction cosines:
parameterize and tabulate the obtaining of results. In both cases 2 3 2 32 3
t ttx tty ttz i
graphs and tables are offered, useful to specify dimensions, testing 6 7 6 76 7
and comparison of results. 4 n 5 ¼ 4 tnx tny tnz 54 j 5 ð3Þ
b tbx tby tbz k
2. Curved beam formula deﬁned in Global Coordinates Fig. 1 represents the axis-line of a curved beam and its systems of
reference, Frenet–Serret and global, associated.
A curved beam is generated by a plane cross section whose cen- The differential system that governs the structural behavior of a
troid sweeps through all the points of an axis curve. The vector ra- curved beam in global coordinates is expressed as [32]:

DV x þqx ¼ 0
DV y þqy ¼ 0
DV z þqz ¼ 0
ttz V y þ tty V z þ DM x þmx ¼ 0
þttz V x ttx V z þDM y þmy ¼ 0
tty V x þ ttx V y þDMz þmz ¼ 0
ð4Þ
cxx Mx  cyx My  czx M z þ Dhx Hx ¼ 0
cxy Mx  cyy My  czy M z þDhy Hy ¼ 0
cxz Mx  cyz M y  czz M z þDhz Hz ¼ 0
exx V x  eyx V y  ezx V z ttz hy þ tty hz þ Ddx  Dx ¼ 0
exy V x  eyy V y  ezy V z þttz hx ttx hz þDdy Dy ¼ 0
exz V x  eyz V y  ezz V z tty hx þ ttx hy þDdz Dz ¼ 0

dius r(s) expresses this curved line, where s length of the arch, is The ﬁrst six rows of the system (4) represent the equilibrium equa-
the independent variable. The reference coordinate system used tions. The functions involved in the equilibrium equations are rep-
here to represent the intervening known and unknown functions resented in Fig. 2 and its expressions are:
of the problem is the Frenet–Serret frame Ptnb. Its unit vectors tan- Internal forces
gent t, normal n and binormal b are: Z Z Z
V ¼ V xi þ V yj þ V zk ¼ r dAt þ sn dAn þ sb dAb ð5Þ
ðt; n; bÞ ¼ ðDr; D2 r=jD2 rj; t ^ nÞ ð1Þ A A A

Internal moments
where D is the derivative with respect to the parameter s.
M ¼ Mx i þ My j þ Mz k
The Frenet–Serret equations [33] describe the movement of the Z Z Z
frame system along the axis line. They are obtained with the vec- ¼ ðsb n  sn bÞdAt þ rb dAn  rn dAb ð6Þ
tors tangent, normal and binormal derivates with respect to the A A A
arch length. Its matricial expression is:
2 3 2 32 3
t 0 vðsÞ 0 t
6 7 6 76 7
D4 n 5 ¼ 4 vðsÞ 0 sðsÞ 54 n 5 ð2Þ
b 0 sðsÞ 0 b

where v(s) and s(s) are the ﬂexure and torsion curvatures
respectively.
Assuming the habitual principles and hypotheses of the
strength of materials [3] and considering the stresses associated
with the normal cross-section (r, sn, sb), the geometric characteris-
tics of the section are: area A(s), shearing coefﬁcients an(s), anb(s),
abn(s), ab(s) and moments of inertia It(s), In(s), Ib(s), Inb(s).
E(s) and G(s) are the longitudinal and transversal elastic moduli
which give the elastic properties of the material.
Equilibrium and kinematics equations compose a system of
twelve linear ordinary differential equations of a curved beam ele-
ment [30].
It is possible to apply a change of basis in the referenced equa-
tions and express the functions in a global coordinate system Pxyz Fig. 1. Curved beam and its reference systems.
F.N. Gimena et al. / Engineering Structures 60 (2014) 189–198 191

Fig. 2. Functions of forces and moment. Fig. 3. Functions of rotations and displacements.

## Load force 3. Formula of arches deﬁned in global coordinates

q ¼ qx i þ qy j þ qz k ð7Þ
In this section it is exposed the procedure to follow and the for-
Load moment mulation to analyze arches. Particularizing the differential system
m ¼ mx i þ my j þ mz k ð8Þ (4) for plane curves of symmetrical section loaded in its plane
(arches):
The last six rows of the system (4) represent the kinematics
DV x þqx ¼ 0
equations. The functions involved in the kinematics equations are
DV y þqy ¼ 0
represented in Fig. 3 and its expressions are:
tty V x þ ttx V y þ DM z þmz ¼ 0
Rotations
M z
EIz
þ Dhz  Hz ¼ 0
h ¼ hx i þ hy j þ hz k ð9Þ exx V x  eyx V y þtty hz þ Ddx Dx ¼ 0
Displacements exy V x  eyy V y ttx hz þDdy Dy ¼ 0
ð15Þ
d ¼ dx i þ dy j þ dz k ð10Þ
In the case of arches, the relation between the reference sys-
Load rotation tems of Frenet–Serret and global could be annotated as:
2 3 2 32 3
H ¼ Hx i þ Hy j þ Hz k ð11Þ t ttx tty 0 i
6 7 6
Load displacement 4 n 5 ¼ 4 tnx tny 0 76 7
54 j 5 ð16Þ
b 0 0 1 k
D ¼ Dx i þ Dy j þ Dz k ð12Þ
It is observed that binormal and z axis coincide. The Fig. 4 rep-
Besides, in formulation (4), the components that multiplied by
resents a generic arch loaded onto his plane.
the internal moments generate rotations could be expressed by:
2 3 2 3 In matricial notation, the system (15) is obtained:
cxx cxy cxz ttx tnx tbx
6c c c 7 6 DeðsÞ ¼ ½TD ðsÞeðsÞ þ qD ðsÞ ð17Þ
4 yx yy yz 5 ¼ 4 tty tny tby 7
5
czx czy czz Vectors involved in the above expression of the arch are:
ttz tnz tbz
2 3 effect vector
1=GIt 0 0 2 3 ð13Þ
6 
2
 
2
 7 ttx tty ttz
eðsÞ ¼ fV x ; V y ; Mz ; hz ; dx ; dy gT ð18Þ
6 0 Ib =E In Ib  Inb Inb =E In Ib  Inb 7 6
6 74 tnx tny tnz 7
5
4    5
0 Inb =E In Ib  I2nb In =E In Ib  I2nb tbx tby tbz

## Also, the components that multiplied by internal forces produce

displacements are:
2 3 2 32 3
exx exy exz ttx tnx tbx 1=EA 0 0
6
4 eyx eyy eyz 7 6 76
5 ¼ 4 tty tny tby 54 0
7
an =GA anb =GA 5
ezx ezy ezz ttz tnz tbz 0 anb =GA ab =GA
2 3
ttx tty ttz
6
4 tnx tny tnz 75
tbx tby tbz
ð14Þ
This new general expression (4), which simulates the structural
behavior of the linear element, has a lower-triangular form. This
important property permits to solve analytically the differential
equation system using successive integrations row by row. Fig. 4. Generic arch with punctual and distributed load.
192 F.N. Gimena et al. / Engineering Structures 60 (2014) 189–198

load vector parabolic shaped arches, with variable cross-section and different
qD ðsÞ ¼ fqx ; qy ; mz ; Hz ; Dx ; Dy g T
ð19Þ heights and spans.

## differential transfer matrix

2 3 4.1. Parabolic arch with variable cross-section under concentrated
0 0 0 0 0 0 force
6 0 0 0 0 0 07
6 7
6 7 The equation that represents the geometry of a parabolic arch in
6 tty ttx 0 0 0 07
½TD ðsÞ ¼ 6
6 0
7 ð20Þ terms of the height f and span l is given by:
6 0 1
EIz
0 0 07
7
6 7
4 exx eyx 0 tty 0 05 2 2
y ¼ 4fx =l ð32Þ
exy eyy 0 ttx 0 0
Integrating the above system directly, row by row, the general In parametric equations:
solution can be written:
xðkÞ ¼ pk; yðkÞ ¼ pk2 =2; zðkÞ ¼ 0 ð33Þ
eðsÞ ¼ ½TT ðsÞC þ qT ðsÞ ð21Þ
being p = l2/8f and k = 8f x/l2, in this example for 1 = kI P k P
where the vector of arbitrary coefﬁcients is:
kII = 1.
C ¼ fC 1 ; C 2 ; C 3 ; C 4 ; C 5 ; C 6 gT ð22Þ A punctual load Q0 = {0,Qy0,0,0,0,0}T is applied at the point
kA = k0 = 0. pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
With the proper change of variable of the arch length s by the
Direction cosines of the curve are ttx ¼ 1= k2 þ 1 and
parameter k, Eq. (17), yields: pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
tty ¼ k= k2 þ 1.
Dk eðkÞ ¼ ½TD ðkÞDk seðkÞ þ qD ðkÞDk s ð23Þ The derivative of the arch length s with respect to the parame-
pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
In the same manner, integrating the above system (7), general ter k is Dk s ¼ p k2 þ 1.
solution is as follows: Properties of the variable cross-section (with height h(k) = h and
pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
eðkÞ ¼ ½TT ðkÞC þ qT ðkÞ ð24Þ width bðkÞ ¼ b0 k2 þ 1) are: A0 = b0h, AðkÞ ¼ A0 k2 þ 1, an = 0,
pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
2
Iz0 = b0h3/12, Iz ðkÞ ¼ Iz0 k2 þ 1 and i0 ¼ Iz0 =A0 .
Let suppose a punctual load applied at a generic point A (see
Fig. 5 show the parabolic arch which is studied in this example.
Fig. 4); the equilibrium and kinematics relate the effects (forces
The differential system (15) with this geometry will be:
and displacements) at this point:
eðkA;I Þ þ eðkA;II Þ þ Q A ¼ 0 ð25Þ Dk V x ¼0
In this equation the vector of punctual action is annotated as: Dk V y ¼0
Q A ¼ fQ xA ; Q yA ; M zA ; 0; 0; 0gT ð26Þ pkV x þpV y þDk M z ¼0
Solution given in Eq. (24) is particularized in the extremes of  EIpz0 M z þDk hz ¼0 ð34Þ
both parts:
p pk
for the ﬁrst part kI P k P kA  ðk2 þ1ÞEA V x  ðk2 þ1ÞEA Vy þpkhz þDk dx ¼0
0 0
1
eðkA;I Þ ¼ ½TT ðkA Þ½TT ðkI Þ ðeðkI Þ  qT ðkI ÞÞ þ qT ðkA Þ ¼ pk pk 2
ð27Þ  ðk2 þ1ÞEA V x  ðk2 þ1ÞEA Vy phz þDk dy ¼ 0
¼ ½TT ðkA ; kI ÞðeðkI Þ  qT ðkI ÞÞ þ qT ðkA Þ 0 0

for the second part kA P k P kII Its direct integration, row by row, gives:

## eðkA;II Þ ¼ ½TT ðkA Þ½TT ðkII Þ1 ðeðkII Þ  qT ðkII ÞÞ þ qT ðkA Þ ¼

ð28Þ
¼ ½TT ðkA ; kII ÞðeðkII Þ  qT ðkII ÞÞ þ qT ðkA Þ
Substituting former values (27) and (28) in Eq. (25), it is
obtained:
 ½TT ðkA ; kI ÞðeðkI Þ  qT ðkI ÞÞ þ ½TT ðkA ; kII ÞðeðkII Þ  qT ðkII ÞÞ
þ QA ¼ 0 ð29Þ
The former expression represents an algebraic system of six
equations with twelve unknowns, six from both extremes of the
arch. Six support conditions are introduced in Eq. (29) to obtain
values at both extremes initial e(kI) and ﬁnal e(kII).
Once knowing these values, the exact solution in both parts is
written:
eðkÞ ¼ ½TT ðk; kI ÞðeðkI Þ  qT ðkI ÞÞ þ qT ðkÞ for kI P k P kA ð30Þ

eðkÞ ¼ ½TT ðk; kII ÞðeðkII Þ  qT ðkII ÞÞ þ qT ðkÞ for kA P k P kII ð31Þ

4. Examples

## In the previous section, a general procedure has been presented

for analyzing arches with distributed and a punctual load. In the
following examples, the geometry of these arches is restrained to Fig. 5. Pinned parabolic arch with variable cross-section.
F.N. Gimena et al. / Engineering Structures 60 (2014) 189–198 193

V x ðkÞ ¼ C1
V y ðkÞ ¼ C2
pk2
M z ðkÞ ¼ 2
C1 pkC 2 þC 3
p2 k3 p2 k2 ð35Þ
hz ðkÞ ¼ 6EIz0
C1  2EI z0
C2 þ EIpkz0 C 3 þ C4
p½30i20 arctan kp2 k5  p½4i20 lnðk2 þ1Þþp2 k4  p2 k3 2
dx ðkÞ ¼ 30EIz0
C1 þ 8EIz0
C2  3EIz0 C 3  pk2 C 4 þ C 5
p½ 12i20 lnðk2 þ1Þþp2 k4  p½6i20 ðkarctan kÞp2 k3  p k2 2
dy ðkÞ ¼ 24EIz0
C1 þ 6EIz0
C2 þ 2EI z0
C 3 þ pkC 4 þ C6

Following the procedure explained in Section 3, considering the In the same way, values at both ends of the ﬁrst part of the
parabolic arch pinned in both extremes, values at both ends of the curve are obtained:
ﬁrst part of the curve are given:  T
 T 25 1 p2
eðkI Þ ¼  Q y0 ; Q y0 ; 0; Q y0 ; 0; 0
5½12i20 ln 25p2  p2 ½15i20 ð4 ln 2þ3pÞp2  32 2 96EIz0
eðkI Þ ¼ Q y0 ; 12 Q y0 ; 0; Q y0 ; 0; 0 ð39Þ
4½15i20 pþ8p2  12EIz0 ½15i20 pþ8p2   T
 25 1 7p p3
5½12i20 ln 25p2  p½60i20 ðln 2þpÞþ7p2  eðk0;I Þ ¼  Q y0 ; Q y0 ;  Q y0 ; 0; 0; Q y0
eðk0;I Þ ¼ Q y0 ; 12 Q y0 ;  Q y0 ; 32 2 64 256EIz0
4½15i20 pþ8p2  8½15i20 pþ8p2 
T In Fig. 6, results of forces, moment, rotation and displacements,
p½p4 þ8p2 i20 ð6pþ25 ln 2þ16Þ60i40 ðp2 4pþ4ðln 2Þ2 Þ
0; 0; 32EIz0 15i20
Q y0 particularized for data given in Benedetti and Tralli [20], l = 42 m;
½ pþ8p2 
f = 10.5 m; b0 = 4 m; h = 1 m; E = 10,000 MPa; Qy0 = 200 kN, are
ð36Þ plotted with and without axial deformation.
Neglecting the axial deformation, the differential system Eq. This graph shows how the hypothesis of neglecting axial defor-
(34), yields: mation is acceptable, and results can be compared with those ob-
tained in the literature.
Dk V x ¼0 In the Table 1, are represented for different values of the param-
Dk V y ¼0 eter k, the functions graphed in the above Fig. 6.
pkV x þpV y þDk M z ¼0 Taking as initial values the data of the Fig. 6 and maintaining
ð37Þ the length l = 42 m, it is presented in Table 2, results of the effects
 EIpz0 Mz þDk hz ¼0
Vx, Mz(0) and dy(0) for different values of slenderness and shallow-
þpkhz þ Dk dx ¼0 ness, with and without axial deformation. Knowing that the slen-
pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
ﬃ pﬃﬃﬃﬃﬃ
phz þ Dk dy ¼0 derness is K ¼ l= Iz0 =A0 ¼ 2 3l=h it has been developed the
Table 2 for ﬁve values of height: h = 0.5, h = 0.707, h = 1, h = 1.414
Its direct integration, row by row, gives:
and h = 2, and ﬁve values of shallowness: f/l = 0.125, f/l = 0.177,
V x ðkÞ ¼ C1 f/l = 0.25, f/l = 0.354 and f/l = 0.5.
V y ðkÞ ¼ C2 This Table 2 also shows how the hypothesis of neglecting axial
pk2 deformation is acceptable.
M z ðkÞ ¼ 2
C1 pkC 2 þC 3
hz ðkÞ ¼ p2 k3
C1 p2 k2
 2EI C2 þ EIpkz0 C 3 þ C4 ð38Þ
6EIz0 z0
4.2. Parabolic arch with constant cross-section under distributed force
p3 k5 p3 k4 p2 k3 pk2
dx ðkÞ ¼  30EIz0 C 1 þ 8EIz0 C 2  3EIz0 C 3  C4 þ C5
2 The equation that represents the geometry of a parabolic arch in
p3 k4 p3 k3 p2 k2 terms of the height f and span l is given by:
dy ðkÞ ¼ 24EIz0
C1  6EI z0
C2 þ 2EI z0
C 3 þ pkC 4 þ C6
2 2
y ¼ f =l ½l  4x2  ð40Þ

Table 1
Forces, moment, rotation and displacements of the parabolic arch with variable cross-section.

k Mz (kN m) hz (103 rad) dx (103 m) dy (103 m) Mz (kN m) hz (103 rad) dx (103 m) dy (103 m)
0 461.71 0.00 0.00 2.34 459.37 0.00 0.00 2.17
0.1 268.10 0.23 0.01 2.08 265.78 0.23 0.02 1.91
0.2 107.25 0.34 0.09 1.46 105.00 0.34 0.11 1.30
0.3 20.84 0.37 0.28 0.69 22.97 0.37 0.30 0.54
0.4 116.16 0.33 0.53 0.05 118.13 0.32 0.56 0.19
0.5 178.71 0.23 0.78 0.65 180.47 0.22 0.81 0.77
0.6 208.50 0.11 0.97 1.01 210.00 0.10 1.00 1.11
0.7 205.53 0.03 1.01 1.10 206.72 0.03 1.04 1.18
0.8 169.78 0.15 0.87 0.92 170.63 0.15 0.89 0.98
0.9 101.27 0.23 0.51 0.53 101.72 0.24 0.53 0.55
1 0.00 0.27 0.00 0.00 0.00 0.28 0.00 0.00
With axial deformation Without axial deformation
Vx(kN) 156.03 156.25
Vy(kN) 100.00 100.00
194 F.N. Gimena et al. / Engineering Structures 60 (2014) 189–198

Table 2
Values of the effect in a parabolic arch with variable cross-section for different data of slenderness and shallowness (f/l).

## Vx(kN) With axial deformation Without axial deformation

Mz(0) (kN m) h = 0.50 h = 0.707 h=1 h=1 h = 1.414 h=2
dy(0) (103 m) K = 290.98 K = 205.76 K = 145.49 K = 145.49 K = 102.88 K = 72.75
f/l = 0.125 312.06 312.50 311.62 312.50 310.74 312.50 309.00 312.50 305.57 312.50
461.69 459.37 464.01 459.37 468.61 459.37 477.76 459.37 495.76 459.37
18.47 17.36 6.92 6.14 2.72 2.17 1.15 0.77 0.54 0.27
f/l = 0.177 220.81 220.97 220.66 220.97 220.35 220.97 219.72 220.97 218.49 220.97
460.54 459.37 461.70 459.37 464.01 459.37 468.63 459.37 477.80 459.37
17.96 17.36 6.56 6.14 2.47 2.17 0.98 0.77 0.42 0.27
f/l = 0.25 156.19 156.25 156.14 156.25 156.03 156.25 155.80 156.25 155.36 156.25
459.96 459.37 460.55 459.37 461.71 459.37 464.05 459.37 468.70 459.37
17.70 17.36 6.38 6.14 2.34 2.17 0.89 0.77 0.36 0.27
f/l = 0.354 110.47 110.49 110.45 110.49 110.41 110.49 110.32 110.49 110.16 110.49
459.67 459.37 459.97 459.37 460.57 459.37 461.76 459.37 464.14 459.37
17.58 17.36 6.29 6.14 2.28 2.17 0.84 0.77 0.32 0.27
f/l = 0.5 78.12 78.13 78.11 78.13 78.10 78.13 78.07 78.13 78.01 78.13
459.53 459.37 459.69 459.37 460.00 459.37 460.62 459.37 461.86 459.37
17.51 17.36 6.24 6.14 2.24 2.17 0.82 0.77 0.31 0.27

Fig. 6. Forces, moment, rotation and displacements of the parabolic arch with variable cross-section.

2
In parametric equations: being p = l2/8f and k ¼ 8fx=l . pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
2 Direction cosines of the curve are ttx ¼ 1= k2 þ 1 and
xðkÞ ¼ pk; yðkÞ ¼ f  pk =2; zðkÞ ¼ 0 ð41Þ pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
tty ¼ k= k2 þ 1.
F.N. Gimena et al. / Engineering Structures 60 (2014) 189–198 195

where
  qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ  qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
a1 ðkÞ ¼ 2k2 þ 1 k k2 þ 1  ln k þ k2 þ 1
rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
 3
a2 ðkÞ ¼ k2 þ 1
qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ  qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
a3 ðkÞ ¼ k k2 þ 1 þ ln k þ k2 þ 1

  qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ    qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
a4 ðkÞ ¼  4k4 þ 4k2 þ 3 k k2 þ 1 þ 3 2k2 þ 64i2z =p2 þ 1 ln k þ k2 þ 1
 qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
a5 ðkÞ ¼ k4 þ 2k2 þ 15i2z =p2 þ 1 k2 þ 1
  qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ    qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
a6 ðkÞ ¼ 2k2  1 k k2 þ 1 þ 4k2 þ 1 ln k þ k2 þ 1
  qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ  qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
a7 ðkÞ ¼ 6k4 þ 7k2 þ 16 15i2z =p2 þ 1 k2 þ 1  15kln k þ k2 þ 1
  qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ    qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
a8 ðkÞ ¼ 2k2 þ 12i2z =p2  5 k k2 þ 1  12i2z =p2 þ 3 ln k þ k2 þ 1
 qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ  qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
Fig. 7. Parabolic arch with constant cross-section vertically loaded. a9 ðkÞ ¼ k2  2 k2 þ 1 þ 3kln k þ k2 þ 1
rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
   3
a10 ðkÞ ¼ 3k2  2 k2 þ 1
 qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ  qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
a11 ðkÞ ¼ 6k4  2k2  315i2z =p2 k  8 k2 þ 1 þ 315i2z =p2 ln k þ k2 þ 1
 qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ  qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
a12 ðkÞ ¼ 8k5 þ 6k3  240i2z =p2 k2  17k þ 480i2z =p2 k2 þ 1  15ln k þ k2 þ 1

2
with iz ¼ Iz =A.
The general solution in function of the arbitrary coefﬁcients is
obtained integrating row by row the differential system (42). Then,
support conditions could be applied to determine the particular
solution.
Neglecting the axial deformation, the differential system Eq.
(42), yields:

Dk V x ¼0
Dk V y pqy ¼ 0
pkV x pV y þDk M z ¼0
pﬃﬃﬃﬃﬃﬃﬃﬃ ð44Þ
p k2 þ1
 EIz M z þDk hz ¼0
pkhz þ Dk dx ¼0
Fig. 8. Bi-ﬁxed parabolic arch with constant cross-section.
phz þDk dy ¼0

## Its direct integration, row by row, gives:

The derivative of the ﬃ arch length s with respect to the parame-
pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
ter k is Dk s ¼ p k2 þ 1. V x ðkÞ ¼ C 1
A distributed load is applied qy in the y direction and the section V y ðkÞ ¼ C2 þpkqy
remains constant. 2 2 3
M z ðkÞ ¼  pk2 C 1 pkC 2 þC 3 þ p 3k qy
Fig. 7 show the parabolic arch which is studied in this section. 2 a ðkÞ 2 a ðkÞ ð45Þ
a3 ðkÞ p3 a10 ðkÞ
The differential system (15) in this example will be: hz ðkÞ ¼  p16EI
1
z
C1  p 3EI
2
z
C2 þ p2EI z
C3 þC 4 45EIz
qy
p3 b1 ðkÞ p3 b2 ðkÞ p2 a6 ðkÞ 2 p4 b5 ðkÞ
Dk V x ¼0 dx ðkÞ ¼ 192EIz
C1  15EIz
C2 þ 16EIz
C 3 þ pk2 C 4 þ C 5 þ 630EIz qy
3 b ðkÞ 3 b ðkÞ 2 a ðkÞ 4
Dk V y pqy ¼0 dy ðkÞ ¼  p240EI
3
z
C1 þ p24EI
4
z
C2 þ p 6EI
9
z
C3 þpkC 4 þ C 6  p720EI
b6 ðkÞ
z
qy
pkV x pV y þDk M z ¼0
pﬃﬃﬃﬃﬃﬃﬃﬃ where
2
 p EIkz þ1 M z þ Dk hz ¼0
qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ  qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
p
 pﬃﬃﬃﬃﬃﬃﬃﬃ Vx pk
þ pﬃﬃﬃﬃﬃﬃﬃﬃ Vy pkhz þ Dk dx ¼0

2
k þ1EA 2
k þ1EA
b1 ðkÞ ¼  4k4 þ 4k2 þ 3 k k2 þ 1 þ 3 2k2 þ 1 ln k þ k2 þ 1
pk pk 2 qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
þ pﬃﬃﬃﬃﬃﬃﬃﬃ Vx  pﬃﬃﬃﬃﬃﬃﬃﬃ Vy phz þDk dy ¼0

2
k þ1EA 2
k þ1EA b2 ðkÞ ¼ k4 þ 2k2 þ 1 k2 þ 1
ð42Þ qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ  qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

## b3 ðkÞ ¼ 6k4 þ 7k2 þ 16 k2 þ 1  15kln k þ k2 þ 1

Its direct integration, row by row, gives: qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ  qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

## V x ðkÞ ¼ C1 b4 ðkÞ ¼  2k2 þ 5 k k2 þ 1  3ln k þ k2 þ 1

qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
V y ðkÞ ¼ C2 þpkqy

## b5 ðkÞ ¼ 6k4  2k2  8 k2 þ 1

2 2 3 qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ  qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
M z ðkÞ ¼  pk2 C1 pkC 2 þC 3 þ p 3k qy

## b6 ðkÞ ¼ 8k5 þ 6k3  17k k2 þ 1  15ln k þ k2 þ 1

2
a1 ðkÞ 2
a2 ðkÞ a3 ðkÞ p3 a10 ðkÞ
hz ðkÞ ¼  p16EI z
C1  p 3EI z
C2 þ p2EI z
C3 þ C4 45EIz
qy
p3 a4 ðkÞ 3
a5 ðkÞ 2a6 ðkÞ 2 4
a11 ðkÞ In Fig. 8 a particular case of a bi-ﬁxed parabolic arch is repre-
dx ðkÞ ¼ C1  p15EI C2 þ p16EI C 3 þ pk2 C 4 þ C 5 þ p630EI qy
192EIz z z z sented with next data, l = 40 m; f = 10 m; b = 2 m; h = 1 m;
3 3 2 4
dy ðkÞ ¼ a7 ðkÞ
 p240EI C1 a8 ðkÞ
þ p24EI C2 a9 ðkÞ
þ p 6EI C 3 þ pkC 4 þ C6 a12 ðkÞ
 p720EI qy E = 20,000 MPa; qy = 5 kN/m.
z z z z

## ð43Þ Results of forces, moment, rotation and displacements are plot-

ted in Fig. 9 with and without axial deformation.
196 F.N. Gimena et al. / Engineering Structures 60 (2014) 189–198

Fig. 9. Forces, moment, rotation and displacements of the parabolic arch with constant cross-section.

Table 3
Forces, moment, rotation and displacements of the parabolic arch with constant cross-section.

k Vy (kN) Mz (kN m) hz (103 rad) dx (103 m) dy (103 m) Vy (kN) Mz (kN m) hz (103 rad) dx (103 m) dy (103 m)
0 0 3.23 0.000 0.000 0.135 0 0.00 0.000 0.000 0.000
0.1 10 3.15 0.002 0.005 0.131 10 0.00 0.000 0.000 0.000
0.2 20 2.88 0.004 0.009 0.120 20 0.00 0.000 0.000 0.000
0.3 30 2.44 0.005 0.012 0.103 30 0.00 0.000 0.000 0.000
0.4 40 1.82 0.007 0.013 0.082 40 0.00 0.000 0.000 0.000
0.5 50 1.03 0.008 0.012 0.060 50 0.00 0.000 0.000 0.000
0.6 60 0.05 0.008 0.008 0.038 60 0.00 0.000 0.000 0.000
0.7 70 1.10 0.008 0.004 0.019 70 0.00 0.000 0.000 0.000
0.8 80 2.42 0.006 0.001 0.005 80 0.00 0.000 0.000 0.000
0.9 90 3.92 0.004 0.003 0.002 90 0.00 0.000 0.000 0.000
1 100 5.60 0.000 0.000 0.000 100 0.00 0.000 0.000 0.000
With axial deformation Without axial deformation
Vx(kN) 99.12 100.00

pﬃﬃﬃﬃﬃﬃﬃﬃﬃ pﬃﬃﬃﬃﬃ
This graph shows how the hypothesis of neglecting axial defor- the slenderness is K ¼ l= Iz =A ¼ 2 3l=h it has been developed
mation is acceptable to obtain the values of the internal forces Vx the Table 4 for ﬁve values of height: h = 0.5, h = 0.707, h = 1,
and Vy. h = 1.414 and h = 2, and ﬁve values of shallowness: f/l = 0.125,
In the Table 3, are represented for different values of the param- f/l = 0.177, f/l = 0.25, f/l = 0.354 and f/l = 0.5.
eter k, the functions graphed in the above Fig. 9.
Taking as initial values the data of the Fig. 9 and maintaining 5. Conclusions
the length l ¼ 40 m, it is presented in Table 4, results of the effects
Vx, Mz(0) and dy(0) for different values of slenderness and Normally, authors use the mobile Frenet system of reference
shallowness, with and without axial deformation. Knowing that with ﬂexure and torsion curvatures to approach the structural
F.N. Gimena et al. / Engineering Structures 60 (2014) 189–198 197

Table 4
Values of the effect in a parabolic arch with constant cross-section for different data of slenderness and shallowness (f/l).

## Vx(kN) With axial deformation Without axial deformation

Mz(0) (kN m) h=0.5 h = 0.707 h=1 h = 1.414 h=2
dy(0) (103 m) K = 277.13 K = 195.96 K = 138.56 K = 97.98 K = 69.28
f/l = 0.125 198.18 200.00 196.39 200.00 192.89 200.00 186.23 200.00 174.12 200.00
3.13 0.00 6.21 0.00 12.21 0.00 23.64 0.00 44.46 0.00
0.82 0.00 0.58 0.00 0.40 0.00 0.27 0.00 0.18 0.00
f/l = 0.177 140.78 141.42 140.15 141.42 138.90 141.42 136.45 141.42 131.78 141.42
1.59 0.00 3.17 0.00 6.28 0.00 12.37 0.00 24.02 0.00
0.45 0.00 0.32 0.00 0.22 0.00 0.16 0.00 0.11 0.00
f/l = 0.25 99.78 100.00 99.56 100.00 99.12 100.00 98.24 100.00 96.54 100.00
0.81 0.00 1.62 0.00 3.23 0.00 6.42 0.00 12.68 0.00
0.27 0.00 0.19 0.00 0.14 0.00 0.09 0.00 0.07 0.00
f/l = 0.354 70.63 70.71 70.56 70.71 70.40 70.71 70.10 70.71 69.49 70.71
0.42 0.00 0.84 0.00 1.68 0.00 3.36 0.00 6.68 0.00
0.19 0.00 0.13 0.00 0.09 0.00 0.07 0.00 0.05 0.00
f/l = 0.5 49.97 50.00 49.95 50.00 49.89 50.00 49.78 50.00 49.57 50.00
0.22 0.00 0.45 0.00 0.89 0.00 1.78 0.00 3.56 0.00
0.15 0.00 0.11 0.00 0.08 0.00 0.05 0.00 0.04 0.00

problem of curved beams elements, when trying to reach exact – In the ﬁrst example (parabolic arch of hyperbolic variable sec-
analytical results. Otherwise, they can use different numerical tion with punctual load) results of internal forces and displace-
approximations or simpliﬁcations in geometry (considering the ments showed are coincident with those cited in the literature.
curved composed of straight beams) to obtain acceptable results. – In both examples, been offered the exact analytical solution, is
In this article, this problem is approached analytically with dif- possible to represent graphically the results as continuous func-
ferential equations and solved using the global reference system. tions in all the points of the axis-line of the arch without
The system of equations to determine the internal forces and dis- discretizations.
placements of curved elements is presented. The method considers – Tables of results, permit the designer to consider or not the
in general, a twisted element with varying cross-sectional area deformations produced by the efforts.
with generalized loads and different boundary conditions.
This article applies the general differential formulation to the Presented formulation, and the solution method developed and
case of the arch, both for its high interest in structural analysis applied in the examples, can be a useful tool in the design and
as for to be covered its calculus from an analytical point of view. structural testing of arches, since is possible to parameterize and
Because of the employment of a system in global coordinates, it tabulate its calculus in function of the conditions of shape, mate-
is obtained a differential system lower triangular of order six. This rial, type of load and support conditions.
system can be solved through the simple successive integration of
its equations.
Once integrated, the notation of the general solution is impor- References
tant, since if you eliminate the integration constants and apply
[1] Love AEH. A treatise on the mathematical theory of elasticity. New
the support conditions for each case, you get a six-order algebraic York: Dover; 1944.
system, which provides the solution to the problem. [2] Parcel JI, Moorman RB. Analysis of statically indeterminate structures. New
On one hand, the application over the annotated system of the York: John Wiley; 1955.
[3] Timoshenko S. Strength of materials. New York: D. Van Nostrand Company;
developed procedure of calculus permits to approach the analysis
1957.
of any type of arches of symmetrical cross-section expressed in [4] Tabarrok B, Farshad M, Yi H. Finite element formulation of spatially curved and
its parametric equations. twisted rods. Comp Meth Appl Mech Eng 1988;70:275–99.
[5] Pestel EC, Leckie EA. Matrix methods in elastomechanics. New York: McGraw-
Hill; 1963.
– It has not been necessary neither for the approach nor for the [6] Washizu K. Some considerations on a naturally curved and twisted slender
calculus, the use of energetic theorems or procedures deriva- beam. J Appl Math Phys 1964;43(2):111–6.
tives thereof. [7] Lee HP. Generalized stiffness matrix of a curved-beam element. AIAA J
1969;7:2043–5.
– Normally, methods of calculus distinguish between arches stat- [8] Norris CH, Wilbur JB, Utku S. Elementary structural analysis. New
ically determinate or not. The solution procedure developed is York: McGraw-Hill; 1976.
independent of the support conditions of the element to be [9] Yang YB, Kuo SR. Effect of curvature on stability of curved beams. J Struct Eng
1987;113:1185–202.
calculated. [10] Rajasekaran S, Padmanabhan S. Equations of curved beams. J Eng Mech
– In the same manner, is possible to take into account or not, 1989;115(5):1094–111.
through the same analytical procedure, the deformation pro- [11] Krishnan A, Suresh YJ. A simple cubic linear element for static and free
vibration analyses of curved beams. Comput Struct 1998;68:473–89.
duced in the arch by the axial and, or shearing effort. [12] Murin J, Kutis V. 3D-beam element with continuous variation of the cross-
– In global coordinates, the results of displacement obtained are sectional area. Comput Struct 2002;80:329–38.
referred to the horizontal and vertical axes of the design, been [13] Papangelis JP, Trahair NS. Flexural torsional buckling of arches. J Struct Eng
1987;113(4):889–906.
directly suitable to the structural testing.
[14] Molari L, Ubertini F. A ﬂexibility-based model for linear analysis of arbitrarily
curved arches. Int J Numer Meth Eng 2006;65:1333–53.
On the other hand, from the two examples developed in this [15] Pi YL, Bradford MA, Uy B. In-plane stability of arches. Int J Solids Struct
article (arch of axis-line parabolic with cross-section variable or 2002;39:105–25.
[16] Tüfekçi E, Arpaci A. Analytical solutions of in-plane static problems for non-
constant, and under a punctual or uniform force load) next conclu- uniform curved beams including axial and shear deformations. Struct Eng
sions are deduced: Mech 2006;22:131–50.
198 F.N. Gimena et al. / Engineering Structures 60 (2014) 189–198

[17] Just DJ. Circularly curved beams under plane loads. J Struct Div [26] Haktanir V, Kiral E. Statistical analysis of elastically and continuously
1982;108:1858–73. supported helicoidal structures by the transfer and stiffness matrix methods.
[18] Litewka P, Rakowski J. The exact thick arch ﬁnite element. Comput Struct Comput Struct 1993;49:663–77.
1996;31:997–1002. [27] Tuma JJ, Munshi RK. Theory and problems of advanced structural
[19] Marquis JP, Wang TM. Stiffness matrix of parabolic beam element. Comput analysis. New York: McGraw-Hill; 1971.
Struct 1989;31:863–70. [28] Kardestuncer H. Elementary matrix analysis of structures. New York: McGraw-
[20] Benedetti A, Tralli A. A new hybrid F.E. model for arbitrarily curved beam-I. Hill; 1974.
Linear analysis. Comput Struct 1989;33:1437–49. [29] Yu A, Fang M, Ma X. Theoretical research on naturally curved and twisted
[21] Lee BK, Oh SJ, Li G, Choi KM. Free vibration analysis of parabolic arches in beams under complicated loads. Comput Struct 2002;80:2529–36.
Cartesian coordinates. Int J Struct Stab Dynam 2003;3(3):377–90. [30] Gimena FN, Gonzaga P, Gimena L. 3D-curved beam element with varying
[22] Wang TM, Merrill TF. Stiffness coefﬁcients of noncircular curved beams. J cross-sectional area under generalized loads. Eng Struct 2008;30:404–11.
Struct Eng 1988;114:1689–99. [31] Gimena FN, Gonzaga P, Gimena L. Stiffness and transfer matrices of a non-
[23] Bergman V. Helicoidal staircases of reinforced concrete. J ACI 1956;28:403–12. naturally curved beam element. Eng Struct 2008;30:1770–81.
[24] Holmes AMC. Analysis of helical beams under symmetrical loading. J Struct [32] Gimena L, Gimena FN, Gonzaga P. Structural analysis of a curved beam
Div 1957;83:1437-1–1437-37. element deﬁned in a global coordinate system. Eng Struct 2008;30:3355–64.
[25] Scordelis AC. Internal forces in uniformly loaded helicoidal girders. J Am Conc [33] Sokolnikoff IS, Redeffer RM. Mathematics of physics and modern
Inst 1960;31:1013–26. engineering. Tokyo: McGraw-Hill; 1958.