0% found this document useful (0 votes)
29 views59 pagesStructural Appendix
Njo
Uploaded by
sufanfan6Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
29 views59 pagesStructural Appendix
Njo
Uploaded by
sufanfan6Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
/ 59
Appendix A
Matrix algebra
A.| Determinants
The determinant of a square matrix is denoted by |S]. This indicates specified arithmetical oper-
ations with the elements of the square matrix [5], which result in a single number, A determinant
1S/ is said to be of order m if (S] is a square matrix of order n x 1.
A second-order determinant is defined as
Siu Sia
= Iso Sua
= Sit So2 — Sy2 Sar (Al)
The operations for a determinant of higher order will be defined later in this section.
The minor My of clement Sy in determinant [S| is defined as the determinant obtained by
omission of the ith row and jth column of |S]. For example, in the third-order determinant
Sur Siz Sis]
1Sl= [Sr Sox Sos
S31 S32 S33)
the minor M32 is given by
Sir Sts
Sar S23
Icis apparent that if |S) is of order n, any minor is a determinant of order n— 1
The cofactor Co; of element S, is obtained from the minor My as follows:
Coy = (—1)IMy (A.2)
Thus, in the above example,
Co32 = (-1)** Myx = —M32
A determinant of order n can be evaktated from the cofactors of any row i, thus:
IS|= > Sy Cox {A3} Appendix A 741
Alternatively, the determinant can be evaluated from the cofactors of any column j, thus:
IS|= > Sy Coy (A)
iat
Equations A.3 and A.4 are called Laplace expansion equations.
Properties of determinants. Some useful properties of determinants are listed below.
1. [fall elements in one row or one column are zero, the determinant is zero.
2. When any two rows or two columns are interchanged, the sign of the determinant is
changed.
3. The determinant of a matrix is the same as the determinant of its transpose.
4. If the elements in a row (or column) are multiplied by a constant and the result added to
the corresponding elements in another row (or column), the determinant is not changed.
5. Ifone row (or column) can be generated by linear combination of other row(s) (or columns),
the determinant is zero. From this it follows that if two rows or columns are identical, the
determinant is zero.
6. The sum of the product of the elements in one row i by the corresponding cofactors of
another row m is zero. Thus, for a matrix [SJnxns
DY $j Cop; =0 (when i xm) (A.5)
iat
This is the general form of Laplace expansion; it can be compared with Eq. A.3, which gives
the value of the determinant |S| when m =i. Similarly, when the product is for the elements in
a column j with the cofactors of another column m, we can write
Y $j Coim =0 (when j #m) (A.6)
iat
The summation is equal to |S| when m=; (see Eq. A.4).
This last property will be used to derive the inverse of a matrix.
Matrix inversion using cofactors: Consider a matrix [S]nxn for which the inverse [S]~! is to
be derived. We form a new matrix [Co] consisting of cofactors Co, of the elements of [S], and
then form the product
[Shue [COljicn = [Bln (A.7)
From the definition of matrix multiplication, any element of [B] is
By => Sip Cop
‘1
From Eqs. A.3 and A.5 we can see that the summation in the above equation is zero when i #),
and equals |S| when i=. Therefore, [B] is a diagonal matrix with all the diagonal elements
equal to |S]. Thus, we can write [B] = |S] [1]. Hence, by substituting in Eq. A.35 and dividing by
(S|, we obtain
1
5 (Co]™ = (5) (S1-* =U 742 Appendix A
whence
(Col?
I
(s]
{A.8)
From Eq, A.36 it is apparent that the division by |S| is possible only when [S| £0. It follows
that only matrices which have a nonzero determinant have an inverse. When its determinant is
zero, a matrix is said to be singular (and its inverse does not exist).
A.2 Solution of simultaneous linear equations
Consider a system of m linear equations with unknowns
Lala (bet = (Chit (A.9)
If all the elements in {c} are zero, the system is called homogeneous. This system occurs in
eigenvalue or characteristic-value problems, such as stability and vibration problems in structural
analysis. In this section, we shall deal only with nonhomogeneous equations for which not all
the clements of {c} are zero.
‘A unigue solution of a system of nonhomogeneous equations exists when the determinant |a|
is nonzero. If |a| =0, that is, if matrix [a] is singular, Eg. A.9 may have no solution or may have
an infinite number of solutions. If the determinant |a| is small compared with the cofactors of [a],
the system of equations is said to be ill-conditioned, in contrast to well-conditioned equations.
‘The determinant |a| of ill-conditioned equaticns is said to be almost singular.
For the purposes of this book, it is useful to mention two methods of solving a system of
simultaneous equations. According to one oF these, known as Cramer's rule, the solution of
Eq. A.9 is:
ee ain
Vez a2... aan
mya ble a 2
{al
fn And +++ Ann
a1 1. ain
lai « a2)
nye ble & 2 (A.10)
alo ee
ant on In|
ai a2 .
Liar az... 6
aya bla an 2
lal
Ant Ann on
Each of the unknowns x; is found by multiplying the reciprocal of the determinant |a| by
the determinant of a matrix which has the same elements as |al except that the ith column is
replaced by the column vector {c}. The evaluation of the determinants involves a large number
of operations, and Cramer's rule is perhaps the most suitable method when the number of
equations is 2 or 3. When the number of equations is greater, other methods require fewer
operations. However, Cramer's rule serves to explain some of the preceding statements related
to the case when the determinant |a| is small or zero. Appendix A 743
As an example of the use of Cramer’s rule, let us consider the equations
Su Sa] [Di|_ fF
[s _ _
Applying Eq. A.38,
\Fi S|
_ |F2 Sxl _ FiS2a— FaS12
and (A.12)
The inverse of [S] may be found, as explained in Section A.7, by solving Eq. A.12 twice: once
with (F), F))=(1, 0} and the second time with (Fi, F2}=(0, 1). Thus,
Sn si (A.13)
Su Say! 1
Su S22) (SisS22 = Si28r1) [Sar Su
‘We can see then that the inverse of a matrix of order 2 x 2 is the product of the reciprocal of
the determinant and of a matrix in which the two elements on the main diagonal of the original
matrix are interchanged, and the other two elements change sign.
‘The method which is most commonly used to solve linear equations (and to invert matrices)
is the method of elimination, introduced in elementary algebra. Several procedures of the elim-
ination process have been developed,' and standard programs are now available for computers.
A discussion of these procedures and of their advantages is beyond the scope of this book, and
we shall limit ourselves here to one procedure suitable for the use of a small calculator.
In the procedure suggested by P.D. Crout,? the calculation is arranged in two tables. Let us
consider it with reference to a system of four equations [a]ax<4 (x}4x1 = {C}4x1- The tables are:
Given equations
ani] a2} ai3| ag |]
ay1| a22| 423) a2a || co
a31| 432 | 433) a34 || cs
qi] 42 | 443) aaa || ca
1 See references on numerical procedures, for example, Crandall, S.H., Engineering Analysis, McGraw-Hill,
New York, 1956.
2.Crout, P. D., “A Short Method for Evaluating Determinants and Solving Systems of Linear Equations with
Real or Complex Coefficients,” Trans, AIEE, 60 (1941), pp. 1235-1240. 744 Appendix A
Auxiliary quantities
bur| 12] O13] bis || at
ba1| b22| b23| b24 || do
b31| b32| b33| bs4 || ds
bar | baz | bas | baa || da
Solution
x1 x2 x3 | x4
‘The first table requires no computation, and is simply formed by the elements of [a] and {c}.
The second table contains auxiliary quantities which are determined following a set pattern
of computation. The elements in the first column of this table {b11, b21, b31, bai) are the same
as the column {ay;, 421, a31, a1}, and are simply copied from the first table. The quantities in
the first row are determined as follows:
1
[bia bis big di = By 413 ays ci]
in
Then the second column (b22 to b42) is completed from the following:
bo (az. — boy b12)
b32 | = } (@32 — bai B12)
ba (as — bay bi)
Next, the second row (b23 to dz) is completed from
1
Uys bos dal= 5 —UUaas ~ bas br3)s(aa4 —bat bia) (ea— bar A]
Proceeding diagonally downward, a column is completed starting from the diagonal element
bj, followed by a row to the right of the diagonal. The elements required to complete the third
column and the third row in the above example are calculated as follows:
bl feta bbe bop
bss)" [lass —bar bis baz b23)
1
[b34 d3]= Ba 4 — b31 b14 — b32 b24), (3 — ba dy — B32 d2)]
Finally, to complete the last row, the elements b44 and da are determined:
bag = a44 ~ bay big — baz bra — bas b34
i
dy = 5 (ca — bas di — ban do — bas ds)
44
From the above we can see that the diagonal elements b11, b22, b33, and b44 and the elements
below the diagonal in Crout’s second table are determined by one pattern of operations. The Appendix A 745
same pattern is followed for the remainder of the table, except that division by the diagonal
element by has also to be carried out.
All these operations can be summarized as follows. Any element bj, such that j < i, is
calculated by
jt
by =a — Yo birdy (A.14)
=
An element bj, such that i birx, (A.18)
rat
where 1 is the number of the unknowns (or of equations).
If we want to invert [a], the columns {c} are replaced by # columns forming a unit matrix.
These columns are treated in the same way as {c} in the preceding discussion, and the column
{d} is replaced by a matrix [d],,y. Back substitution in Eq. A.18 with the elements of d in each
column of this matrix gives a column of the inverse matrix.
If matrix [a] is symmettical, which is usually the case in equations needed in structural analysis,
the arithmetical operations are reduced considerably as each of the auxiliary quantities by to
the right of the diagonal is related to an element by below the diagonal by
bi
bi i ay Coy
-”
and choosing i= 1, we obtain
la|= 3 -A[(11 - (9-4) —7 x 71-7179 -) -7 « 9]
+917 x7- (11-4) x 9]=0
from which
23 — 2327 204 +300=0
Hence, the three values of 4 are 23.31, 3.44, and —3.75. As expected, the number of eigen-
values is equal to the order of the matrix. In many practical problems, it is only the largest
eigenvalue that is of interest.
In general, when [6] is of order, expansion of the determinant in Eq. A.22 gives a polynomial
in 2 of the form
Rea bea? bee toy = (A.23)
This equation is called the characteristic equation of the matrix (b], and its solution gives the
eigenvalues 41, Aas... An which satisfy the conditions
DV xeDd bi (A.24)
= at
and
Ar x AQ x An = 1b] ({A.25)
The eigenvalues obtained in the preceding examples can be checked by these two equations.
The eigenvector corresponding to the 4 values can be found by substituting for 4 in Eq. A.20
and solving, Since Eq. A.20 is homogeneous, only relative values of x can be determined for
each eigenvalue. We can assume x1 = I, solve for the other values using (7t — 1) equations and
the last equation can be used to check the calculations.
Instead of evaluating the determinant, we may use an iteration method which leads to the
cigenvalue with the largest absolute value. In this we assume the value of the eigenvector (x) and 748 Appendix A
substitute it in the left-hand side of Eq. A.21. Hence, we find a value of 4 and a new value of
{x). If this new value of (x) is equal to the assumed value, then we have solved the problem: A is
the required eigenvalue, and (x) the eigenvector. If, however, there is a discrepancy between the
two values of (x}, we use the new value as an assumed value in the second cycle of evaluating
[b]{x}. We repeat this procedure until Eq. A.21 is satisfied.
In many practical problems, a good estimate of (x) corresponding to the largest A can be made
from the physical data available, and rapid convergence is obtained.
‘As an example, let us consider the matrix (b] given earlier, and assume the vector {x} as
below. Then
&} bx)
+ +
3.7 971 (1.0 22.2 1.00
7 UL 7) 41.2} =} 28.6) =22.2 41.29
9 7 9} 44 27.3 1.23
Since {x} # {x}, we repeat the multiplication with {x}:
{x} (x)
+ +
3.7 97 [1.00 23.10 1.00
7 AL 7] 41.29} = 4 29.80 p= 23.10 1.29
9 7 94 [1.23 29.10 1.26
Since {x2} = {x}, we have found the eigenvector, and 4 = 23.1 is the largest eigenvalue. A
third repetition of the multiplication using {x} gives 4 =23, which is very close to the exact
solution.
Problems
A.1 Given the matrices
102 ced
wi=[ 9 i} [3 3]
execute the following matrix operations, if possible:
(a) [A] [B] (b) (AJ? [(B]
(c) (AJTEB)[A]_ (@)_ [A] [BI" [ATT
(e) Prove that, if [B] is symmetrical, the product [A]"[B] [A] must also be symmetrical.
‘A.2 Given [A] and [B] as in Prob. A.1 and
3.20
of} 0 3
execute the following operations:
(a) [BI[A+C] (b) [B] [A] +[B][C)
Note that the distributive law [B] [A + C]=[B] [A] +B] [C] applies to matrices.
A.3 Prove that the transpose of the product of two or more matrices is the product of the
transposes of the matrices in a reversed order. Appendix A 749
A4 Find the inverse of the following matrices:
2 -1 0
[5 Al (b){-1 2-1
0 -1
4
(c)|-4 6-4
1-4 7
A\S The coordinates 1 and 2 in part (a) of the figure represent the positive directions of forces
F, and F; and displacements D; and D2 at the end A of cantilever AB. The displacements
(deflection or rotation) due to unit values of the forces (vertical load or couple) are given
in parts (b) and (c); these displacements are arranged in the flexibility matrix
ifl= fa fa) _[ IED —P/QED)
“Lit fal” L-P/QED B/GED
Assuming that the displacements {D1, Dz} due to forces {F, F2} can be obtained by the
superposition equation
[f | {Ee} {p|
fr fro} {FJ ~ (D2
(a) Find the displacements due to F) = $ and F) =P acting simultaneousy.
(b) Find the forces $j, and S21 corresponding to the displacements Di = 1 and D2 =0.
Sketch the deflected shape of the beam.
(c) Find the forces $12 and S22 corresponding to the displacement D, =0 and D = 1.
Sketch the deflected shape of the beam.
(d) Show that the stiffness matrix formed by the forces obtained in (b) and (c)
—[Su Se
isi=[3u Pal
is the inverse of the flexibility matrix [/1.
pint
ti Moment= nt; +
}+—_——_ 1 —__+|
()
Vertical load =
a
foe
Prob. AS
A.6 The stiffness matrix for the beam in the figure is
,_EI[ 96 ~8.4
=z Ee oe
The forces {F] = {F1, F2) and the corresponding displacements (D] = (D1, D2) along the
coordinates 1 and 2 are related as follows: [S] {D) = (F}. 750 Appendix A
(a) Sketch the deflected shape of the beam when the forces applied are Fy = S11 and Fr =S21.
(b) Find the inverse of the stiffness matrix [S]. What does the resulting matrix signify?
Prob. A.6
A.7 The stiffness matrix of the beam shown in part (a) of the figure is
12
_
2 616 72
Tit
zp
‘The forces P1, P2, P3, and Ps and the corresponding displacements Dy, D2, D3, and Ds
along the coordinates are related as follows:
fies (Sil {ial (Fi) (b)
{Sai} S221} [tDa} > |tFa}
where [S11], [S12] etc., refer to the partitioned matrices in [S] and
Di= aE w= {Bis va={pifs ra={pif
By putting (F:} = {0} in Eq. (b), find a matrix [S*] relating the forces to the displacements
along the coordinates indicated in Prob. A.7(b), such that
[S°} (Di) ={Fi} (c)
fa)
1 or 2
\a 2a ts Io a
-—— n + 2 4
(b)
}—— 2
Prob. A.7 Appendix A 751
A.8_ Solve the following simultaneous equations:
8-2] fm] _f 18 )
-3 1} |xf>]-17 .
4-2 17 [x 2
3.8 -2| {x2}=413 (b)
2 -s 10} xs} [27
108 6 249 (Di _
a ae
EI| 6 1
Tl-7 8 2 Di}=P (d)
24
-—t 2 2 1
t D3 3
1
1 07) [Fi 7
! 1 7
cal N}-|8 °
3 0
A.9_ Find the eigenvalues for the matrices in Prob. A.4 (a) and (c), Calculate the eigenvector
corresponding to the highest eigenvalue.
A.10 Using iteration, evaluate the smallest eigenvalue of the matrix
ee
faj=|-1 2 0
pees eee oe
‘The eigenvalues of [a]~! are the reciprocals of those for [a]. To find by iteration the largest
eigenvalue of (a]~1, assume that the eigenvector represents the buckling mode of a strut
hinged at one end and encastré at the other. The elements of the eigenvector are
proportional to the deflections at equally-spaced points on the axis of the strut. (See
Figure 14.8b.)
* Appendix B
Displacements of prismatic members
The following table gives the displacements in beams of constant flexural rigidity EI and constant
torsional rigidity GJ, subjected to the loading shown on each beam. The positive directions of
the displacements are downward for translation, clockwise for rotation. The deformations due
to shearing forces are neglected.
q per unit length _ 5 alt
f= 394 BT (B.)
M
te fr fA fs h=h =a = (B.2)
———e|
7 2B
fam a fea a re
fe wae —2b2+33) — (B.4)
fia OF am — Bx) when x b (B6)
Pb(l—b) Pb on
A= ep AI h=~ ge —b*) (B.7)
When b=1/2, fy =—fs = PP/(16ED), and fy = PP/48El atx =1/2 (B8) Appendix B 753
*) M
fh fe ih
fia 14 Lt 14 Lt 4 —
f Ah wi
Lemp 4 —-— 14 fra —]
4 per unit length
Vv
h
+ 22 —+.——_ 112 — +
ML
h=3Ry (B.9)
MI
h=-= (B.10)
(B.11)
(B.12)
(B.13)
MI
A= aR (B.14)
9MP
h=- 3365 (B.15)
MP
B= - SE (B.16)
3MP-
=~ s56Ri (B17)
Tl
Arg (B.18)
(Effect of warping ignored)
PB
h=aR (B.19)
f= PP/2EI (B.20)
fr=ftdh (B21)
PP Bb aoe
he SET (1 -at a)
for0
4.39175, 2.78350 0.74227
~S.56701 5.13402 Se aa Ed
1.48454 —2.96907 4.39175
0.37113 0.74227 ~2,59794
0.06186 0.12371 0.43299
Reactions in terms of E/P
oe 10.70103 —4.45361
10.70103 —18.80411 11.81443
4.45361 11.81443 SoG AIL EE
1.11340 —4. 9.58763
0.18557 ee
Number of spans=5
Support moments in terms of EI/P?
4.39227 =2.78453 0.74586
5.56906 5.13812 — se.
—— =2.98342 4.44199
0.39779 0.79558 —2.78453
0.09945, —0.19889 0.69613
—0.01657 0.03315 11602
Reactions in terms of EI/P?
—7.17680 10.70718 —4.47513
10.70718 ~18,82872 11.90055
~4.47513 11.90055 —14.65193
pL She 4.77348 10.70718
0.29834 1.19337 4.17679
0.04972, —0.19889 0.69613
0.69231
-2.76923
3.69231
-1.61538
0.18557
0.74227
-2,59794
3.64948,
1.60825
1.11340
~4.45361
9.58763
—9,89690
3.64948
0.19889
0.79558
~2.78453
4.34254
=2.58563
0.43094
1.19337
~4.77348
10.70718
=14.05524
9.51381
2.58563
0.18557
0.74227
-2.59794
3.64948
1.60825
0.04972
—0.19889
0.69613
~2.58563
3.64641
~1.60773
=0.29834
1.19337
~4,17679
9.51381
=9.87845
3.64641
0.04972
—0.19889
0.69613
=2.58563
3.64641
=1.60773 Appendix E 765
Bending moments
I
Reactions 3.6 ~9.6 8.4
Figure E.1 Illustration of the use of tables of Appendix E.
The values given in each row are the bending moments or the reactions at consecutive supports
starting from the left-hand end. The first row after the heading gives the effect of the settlement
of the first support from the left, the second row gives the effect of the settlement of the second
support from the left, and so on
Figure E.1 shows an example of the use of the tables: the number of spans is 3, the second
support from left settles a unit distance and the values are taken from the second row of the
appropriate table.
In these tables, the reaction is considered positive when acting upward, and the bending
moment is positive if it causes tension in the bottom fiber of the beam. When the reactions are
used to generate a stiffness matrix, appropriate signs should be given according to the chosen
coordinate system,
The effect of deformation caused by the shearing forces is neglected. Appendix F
Properties of geometrical figures
yh
Area x and ¥ coordinates
of centroid
—_!
zal
axl .
roe
a)
~_ 1lai+2a)
Rou
a 3 Gta)
Gt yy :
2 ~_ (a taya2+a})
Y= Xai Faz)
x=teal+)
An ¥=z(@l+)
2
3
a
3 Appendix F 767
Second-degree parabola
y
2
jax!
0 ——+x
[+
Area
Second-degree parabola
y
1
| ba
Tangent| 7 |
Second-degree parabola
y4 Tangent
: oa
0 -
}___+____|
‘Third-degree parabola
| be
Tagen 4
Construction of tangents of a second-degree parabola:
q per unit length
distances
Equal distances “'**
< e £
ee
b
NNN
Centroid
qT
xa
3
q*
3
10
5
8
2
5
xi
i
x
I
i
x
a
Bending moment diagram for a part of a
member subjected to a uniform load Appendix G
Torsional constant J
Ifa circular bar of constant cross section and of length / is subjected to a constant torque T, the
angle of twist between the two bar ends is
ont!
a
where G is the shear modulus and J the polar moment of inertia.
When the cross section of the bar is noncircular, plane cross sections do not remain plane after
deformation, and warping will occur caused by longitudinal displacements of points in the cross
section. Nevertheless, the above equation caa be used with good accuracy for noncircular cross
sections, but J should be taken as the appropriate torsion constant. The torsional constants for
several shapes of cross sections are listed below.
Section | Torsional Constant J
= i
J = 0.14066"
| Closed section
4a
ds f a
a ‘
f where a= area enclosed by a line through the center
of the thickness and the integral is carried out over
the circumference
th
7
:
hl n—| by
2b: — bb. — 4
bit tbh te ty biti tbh + bt}
aa
byt} + bat}
=
t Open section composed of rectangles
ie feb Appendix H
Values of the integral {;M,Mdl
The following table gives the values of the integral J; M,Mdl, needed in the calculation of
displacement of framed structures by virtual work (Eq. 8.2). The same table can be used for the
evaluation of the integrals fj NwN dl, f; VwV dl, fTuT dl, or for the integral over a length / of
the product of any two functions which vary in the manner indicated in the diagrams at the top
and at the left-hand edge of the table. -pjoqesed 22:8p-pu0rs .
ey ok a be T seater,
(d-d-9) Cast qo oe qo et £
z b—I—4
T
(oret nee cages et et ioe Voy
suasuea
£ —
ot ve 4+ 'g ee sak sae wet Go
b—1—4
€ (q+) 5 5 : Sy
pe +e 2 ve
Le, ++ Ie + Digs et Di a
epg rbe pool
Emory cree a | ee cops? od pot
+g oie + 4gine gen? 4 9 4 ef 2
z z z z k—1—
we tq iq ot z
iPr Cat 5 iPr lal 9° 7
br a
bai a 7
LW |e ny
fee-shres4
1
% 4
a Appendix |
Deflection of a simple beam of
constant El subjected to unit
end-moments
The values of the deflection given below are helpful in the calculation of ordinates of influence
lines (see Section 12.5). The deflection y due to end-moments Mag =1 and Mpa =0 (Figure I.1a)
is given from Eq. 12.5 by
Fae -3e +8) (1)
ez ~ .
where £ =x/l, x is the distance from the left-hand end, and / is the member length. The values
of the deflections at different values of x/! are given in Table 1.1.
(b) ly
Figure I.1 Deflection of a simple beam due to unit end-moments,
Table 1.1 Deflections Due to Unit Clockwise Moment at Left-Hand End
523 of or] 02] 03] 04] os] o6| 07 os | 0.9 | 1.0 | Multiplier
y 0} 285} 480] s9s| 640) 625] 560) ss a] 165] 0 we 774 Appendix |
Table 1.2. Deflections Due to Unit Clockwise Moment at Right-Hand End
of o1 | 02 | 03 | o4 Jos | 06 | 07 } 08 | 0.9 | 1.0] Muttiptier
2
Y oe 455 560 625] —640} —S9S| 480] —285| 0 10 eo
‘The deflection y due to Maa =1 and Maz =0 (Figure 1.1) is given by
= gs) (12)
>= CEI .
The values of the deflection at different values of & = x/I are given in Table 1.2. Appendix J
Geometrical properties of some plane
areas commonly used in the method
of column analogy
Property
Figure Representing Analogous Column
WEL
1
a (2xaya + 2xny0
pq tava + 2x
+xaya+xaya)
Las gOk ty ty)
(3 +35 txax8)
6 is measured in radians
2or
aren oe
rea=a= FF
—_ sing
@
P@+sin8 cos)
4 EL
_ P(@=sin8 086)
EI
+P
ep
y=-b+r-yvr—x
Circular arch of constant EI Property
l
Area =a=
4b
= aSEI
Pp
b= DER
Figure Representing Analogous Column
al
El, 3
Parabolic arch with EI assumed to vary
as the secant of the inclination of the
arch axis, Elis the flexural rigidity
at the crown Appendix K
Forces due to prestressing of
concrete members
The figures below show sets of forces in equilibrium representing the forces that prestressed ten-
dons exert on the concrete. The tendon profiles shown are composed of one or more straight lines
or of second-degree parabolas. The symbol P represents the absolute value of the prestressing
force which is assumed constant, and y represents the vertical distance between the tendon and
of the member. @ represents the angle between a tangent to the tendon and
the centroidal a:
the horizontal; @ is assumed small so that # = tan6~sin@ and cos@~ 1.
Pya Pyp
a [Pe | ‘\
Pa fPiéx+4) BP
Pa 12 Peglad 2 Plegmep Mell?
aaa * Na
P CEETETr ore TJ P
2 Peghal?
Py, Po Pe, Py
a) c | pb ly
PA ty f ed td
Po PO
PY4 Py,
(Pq Pe Pea >
aa i
errr)
re VP
O4=(4c+y4-ypll
Op=(4e-yatyp il
‘Two parabolas with a common tangent at
D and horizontal tangents at B and C.
‘The values ca, cx and y may be chosen
arbitrarily, but and cp must satisfy
B= yep/(es— en) and cp = caB? far, Appendix L
Structural analysis computer programs
L.I Introduction
A series of microcomputer programs is available for use as a companion to this book. The
programs can be employed in practical applications, involving the solution of large problems.
They are also written as a learning tool. FORTRAN-language source codes are provided with
extensive comments, such that the steps of the analysis can be followed. Reference is made to
the various sections and equations on which the computer programs are based. Modifications of
the programs to perform additional tasks can be done without difficulty, and can prove to be an
efficient means of fully understanding the analysis techniques. The comments at the beginning
of the source code include definitions of the symbols employed and instructions for input data
preparation.
L.2 Computer programs provided at
bttp:/www.routledge.com/books/Structural-Analysis-isbn978041. 5774338
‘The following password will be required tc access the site:
STRUCTURES
(NB: The password needs to be in capital letters.)
In Section L.4 the computer programs are divided into three groups: (A) Linear analysis
programs. (B) Nonlinear analysis programs. (C) Matrix algebra. Three files described in the
following section can be obtained for each program. For the computer programs in groups A
and C, the files can be unloaded from the Internet at the web site address given above. The files,
for the computer programs of group B can be ordered using the form at end of book.
L3 Files for each program
Three files are provided for each program:
1, The source code in FORTRAN
2. Executable file
3. Input file
Each file has a name composed of up to eight letters descriptive of the program, followed by a
dot and two to three letters. For example, the three files for the program PLANEF which analyzes
plane frames have, respectively, the names: PLANEF.FOR, PLANEEEXE, and PLANEEIN. Appendix L_ 779
‘Typing the letters PLANEF and pressing “Enter” will execute the program, generate a file
named PLANEROUT and - if running the program is with DOS command - cause the words
“Program terminated” to appear on the screen. The file PLANEF. OUT echoes the input data
and gives the results of the analysis of the problem whose data are given in PLANEEIN.
‘The input files provided on the disks are for simple example problems selected from the book.
Typically, an input data line contains numbers on the left-hand side and symbols or words on
the right-hand side, indicating what are the parameters given on the line. Figure L.1 shows, as
an example, the contents of the file PLANEEIN, giving the input data for the plane frame of
Example 21.2 (Figure 21,5 and Table 21.2). To analyze a new structure, it is proposed to revise
the example input file by changing the values of the parameters, rather than creating a new file.
Thus, any input file example provided on the disks should be copied in a back-up file before it
is revised. Preparing an input file for a new problem in this way is similar to completing a form
giving the data required on the right-hand side of the form.
As mentioned above, all symbols used are defined at the beginning of the source code of
each program. We give below definition of the symbols in Figure L.1: NJ, NM, NSJ, NLC are
numbers of joints, members, supported joints, and load cases respectively; JS and JE are start
node and end node of an individual member. The remaining symbols and the restraint indicators
are explained in Section 21.4. Program PLANEF accepts two types of loading, entered as nodal-
force components (Fy, Fy, Mz) on individual nodes, or as six fixed-end forces {A,) for individual
members. Only the nodes subjected to nodal forces and only the members having fixed-end
forces need to be listed. A dummy data line indicates the end of data for each of the two
types of loadings. The dummy lines start with any integer greater than NLC (number of load
cases).
Figure L.2 is part of the file PLANEEOUT, showing the input and the results of case
loading 1.
Plane frame Example 21.2, Figure 21.5.
a oo ‘NJ, NM, NS], NLC
30000. 0. Flasticity modulus, Poisson's ratio
1 0.0 300.0 ‘Node number, x and y coordinates
2 00 0.0
3 300.0 150.0
4 300.0 300.0
11 2 30.0 3000.0 1.6 Member number, JS, JE, cross-sec. a, I, at
2 2 3 40.0 5000.0 1.06
3 3 4 30.0 3000.0 1e6
1 340 340.0 Node, restraint indicators, prse. displs.
4 0 0 1 3x00
1 2 40 0.000 Load case, node, Fx, Fy, Mz
1 3 20 00 0.0
10 0 3x0.0 Dummy, end of forces applied at nodes
12-1342 -26.83 -1500. -13.42 26.83 1500. Ld. Case, member, {Ar}
2 1 -1500. 8. 1200. 1500. -8 1200.
10 0 640. Dummy, end of data of member loads
Figure L.1 Image of file PLANEEIN containing input data for the plane frame of Example 21.2. Plane frame Example 21.2, Figure 21.5.
Number of joints = 4
‘Number of members = 3.
Number of joints with prescribed displacement(s) = 2
Number of Load cases = 2
Elasticity modulus = .30000F+05; Poisson’s ratio = .0000
Nodal coordinates
Node x y
1 .000 300.000
2 000.000
3 300.000 150.000
4 300.000 300.000
Element information
Element 1stnode 2nd node a I ar
1 1 2 -30000E+02 .30000E+04 .10000E+07
2 2 3 -40000E+02 .S0000E+04 .10000E+07
3 3 4 .30000E+02 .30000E+04 .10000E+07
Support conditions
Node Restraint indicators Prescribed displacements
uv theta u v theta
1 o 0 0 00000E+00 .00000E+00 .00000E +00
4 0 0 1 00000E+00 .00000E+00 .00000E+00
Forces applied at the nodes
Load case Node Px. Fy Mz
1 2 .40000E+01 .00000E+00 .00000E+00
1 3 .20000E+01 .00000E+00 .00000E+00
10 0 .00000E+00 .00000E400 .00000E+00
Member end forces with nodal displacement restrained
Ld.case Member Ari Ard, AS Ard Ars Ar6
1 2 -.1342E+02 -.2683E+02 -.1500E+04 -.1342E+02 —2683E+02 .1500E+04
a 1 =1500E+04 8000401 .1200E+04 .1500E+04 -.8000E+01 .1200E+04
10 0 .0000E+00 .0000E+00 .0000E+00 .0000E+00 .0000E+00 .0000E+00
Analysis results
Load case no.1
Nodal displacements
Node u v theta
1 ~.6805SE-07 —.81171E-08 -.16029E-09
2 139935E-01 .S1I71E-U2—.71994E-03
3 .38354E-01 .59412E-02 ~47191E-03
4 27285E-07 .S9412E-08 —.61949E-03
Figure L.2 Image of the file PLANEROUT (see Figure L.1 for corresponding input data file). Results
of case of loading 2 are omitted. This figure is continued on the next page. Appendix L781
Forces at the supported nodes
Node Fx Fy Mz
1 .27222E+01 - 24351402 .19235E+03
4 ~.87311E+01 -.35647E+02 .00000E+00
Member end forces
Member FL F2 F3 F4 FS F6
1 24351E+02 .27222E+01 .19235E+03 ~.24351E+02 -27222E+01 .62431E+03
2 -48777E+01 -.24787E+02 -.62431E+03 -.21962E+02 -.28873E+02 .13097E+04
3 35647E+02 -.87311E+01 -.13097E+04 -.35647E+02 .87311E+01 -.670S7E-13
Figure L.2 (Continued)
L.4 Computer program descriptions
In the following, the available computer programs are mentioned by name and by a short
description. The programs are divided into three groups:
A. Linear analysis programs (basis: Chapter 21)
PLANEE (plane frame)
SPACEF (space frame)
PLANET (plane truss)
SPACET (space truss)
PLANEG (plane grid)
These five programs are, respectively, for the linear analysis of plane frames, space frames,
plane trusses, space trusses, and grids.
B. Nonlinear analysis programs
PDELTA (basis: Chapter 13)
PLASTICE (basis: Section 22.7)
NLST (basis: Chapter 23)
NLPF (basis: Chapter 23)
The program PDELTA performs quasi-linear analysis of plane frames, considering the effect
of axial forces on the stiffness matrix of individual members (Eqs. 13.19 and 13.23).
Analysis of collapse loads of plane frames due to the development of plastic hinges
(Section 22.7) can be performed with the program PLASTICE. The program increases the values
of the loads gradually and indicates the load levels at which the plastic hinges are developed,
the corresponding nodal displacements and member end-forces.
The program NLST performs nonlinear analysis of space trusses, employing Newton-
Raphson’s technique. Cable nets can be analyzed by the same program. Cable nets with
membrane triangular elements can also be analyzed.
Nonlinear analysis of plane frames can be performed by the program NLPF. Similar to the
program PDELTA, the program NLPF calculates the stiffness of the members accounting for the
magnitude of the axial forces. But, in addition, NLPF considers the equilibrium of the nodes in
their displaced positions. The change in length of members is not ignored in determining the 782 Appendix L
geometry of the deformed structure. NLPF can be employed to indicate the nonlinear buckling
loads.
C. Matrix algebra (basis: Appendix A)
‘To enhance the study of this book, the web site provides the following simple programs which
can perform frequently needed matrix operations:
ADD
MULTIPLY
INVERT
SOLVE
DETERM.
EIGEN
The program ADD calculates the sum: [C] = @[A] + y{B], where [A] and [B] are the given
matrices; a and y are the given multipliers. The program MULTIPLY determines the product
of two given matrices. The program INVERT calculates the inverse of nonsingular matrices.
The program SOLVE gives the matrix [X] in the simultaneous linear equations [A] [X]= B],
when [A] and [B) are given. The program DETERM calculates the determinant of a given square
matrix. EIGEN solves the equation: [S]{D} =4[]{D}; where [S] and [71] are square matrices;
the solution gives eigenvalues, & and the corresponding eigenvectors (see Section 20.6). Appendix M
Basic probability theory!
For more comprehensive presentation of probability theory, refer to specialized texts.*
M.I Basic definitions
Consider the outcome of an experiment; the experiment can be testing the strength of a material,
measuring a beam depth, or the occurrence of a truck on a bridge in a specified period of time.
Any outcome of the experiment is called an event. Alll possible events of an experiment comprise
a sample space.
Example M.1: Results of compressive strength tests of concrete
Let x1,%2,...524 be the measured compressive strength (outcome) of n concrete cylin-
ders. The actual compressive strength, ’, varies randomly, and 1 test results provide only
limited information about its variation. For this experiment, the result /’ can take any
positive value. Theoretically, even f’ =0 is possible, but it is unlikely (when the mix is
made without any cement). Events Ey, £2,...,E, can be defined as intervals; for example,
Ey occurs when the outcome of cylinder test falls within the interval (0, 10 MPa), E>
‘occurs when the outcome of cylinder test falls within the interval (10, 20 MPa), and so on.
The defined sample space for concrete cylinder tests is called a continuous sample space.
A certain event is equal to the sample space. An example of this, using the compressive
strength data, is: certain event=all positive real numbers, 0 PE) =
(M.1)
where (JE; =union of all events F1,E2,...sEn-
=
‘Mutually exclusive events exist when the occurrence of any one member in a set excludes
the occurrence of the others. For concrete compressive strength, f!, examples are: Ey =
(O 30 MPa),
(M.2)
M.3 Random variable
In the example of concrete strength, f! is a random variable. A random variable X can be
continuous or discrete. Examples of continuous random variables are: if f/ = 31.5 MPa, then
X(f)=31.5 (no units, a real number), or if X(f!) = {(value in MPa)/10)-2.0 and f= 31.5 MPa,
then X(f)) = 1.15.
‘An example of a discrete random variable is:
lif O 0, even though the mean may be negative.
Vy =k (17)
ix
Example M.3: Calculation of x, E(X"), 0%, and Vx
For the discrete random variable X, whose probability mass function is given by Eq. M.7,
calculate: the mean value %, the second moment E(X*), the variance o3, the standard
deviation a and the coefficient of variation Vx.
Equations M.10, M.12, M.14, M.16 and M.17 give, respectively:
= 10.05) + 2(0.2) + 300.65) +40.1) =2.8
E(X?) = 17(0.05) + 27(0.2) + 37(0.65) +47(0.1) =8.3,
of = 8.32.8)? =0.46
ox =V0.46 = 0.6782
Vx = 0.6782/2.8 = 0.2422
M.6 Sample parameters
If X is analyzed on the basis of test data: x1,2°2,....2y, then the mean can be approximated
by the sample mean, and the standard deviation can be approximated by the sample standard
deviation.
The sample mean (approximating mean) is:
12
(M.18)
(M.19)
M.7 Standardized form of a random variable
Let X be a random variable. Z is called a standardized form of X, and is defined as,
at (M.20)
ox
‘The mean of Z is zero: 2=0. The variance of Z=1 is of = 1. Appendix M 787
M.8 Types of random variables used in reliability of structures
The most important variables used in structural reliability analysis are:
Uniform
Normal
Log-normal
Extreme type I and extreme type II
Pepe
M.9 Uniform random variable
Uniform random variable means that all numbers are equally likely to appear.
1
forasx 0.5, calculate @!(1-p), and use Eq. M.32, which is equivalent to:
o-\(py=—b"(1—p) (M.35)
M.I1 Log-normal random variable
A random variable X has a log-normal distribution when the random variable Y has a normal
distribution; where Y=InX (i. Xe"). With a mean value y and a standard deviation oy, the
density function fx(x) is given by Eq. M.24 substitutinge 2% bye 2°Y ;
ae
Kk@)= (M.36)
1
oyvla 792 Appendix M
‘The cumulative distribution function, CDF, can be calculated using the standard normal values
(Table M.1).
Fy(a) = POX F262
2+ 0.9804
0.950
0,900 +
0.800
Minimum data
point
0.700
0.600
oF 0.500
0.400
0.300
0.050
eile 00020)
0.010
0.005 >
3+ 135x104
Figure M.3. CDF of the test results in Example M.6.
M.14 Covariance of two random variables
Consider a vector (X}n1 of m random variables. A pair of dependent (correlated) variables X;
and X; has a covariance, COV(X;, Xj), a positive or negative value, defined as:
COV(X;, X)) = E[(Xi —¥(Xj — ¥) ] = EX Xj) — Hp (M.45)
The covariance of the two variables is equal to the mean of their product minus the product
of their mean values. Equation M.14 is a special case of Eq. M.45 when X; and X; are the same.
The correlation coefficient between X; and X; is defined as:
COV(X:,X))
(M.46)
Ox;0%;
Px,x,
Tt can be shown that: ~1
