Deriving Element Matrices for ACM Plate Bending

Element using the Fundamental of Calculus

Sulaiman Abo Diab

To cite this version:

Sulaiman Abo Diab. Deriving Element Matrices for ACM Plate Bending Element using the Funda-
mental of Calculus. 2020. �hal-02950250�

HAL Id: hal-02950250

https://hal.science/hal-02950250
Preprint submitted on 27 Sep 2020

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
 Deriving Element Matrices for ACM Plate Bending Element using the
Fundamental of Calculus

Sulaiman Abo Diab

Faculty of Civil Engineering, Tishreen University, Lattakia, Syria

[email protected]

Abstracts: This work shows the derivation of the element matrices for a rectangular ACM plate-bending element
using the fundamental of calculus. The double integral over a rectangular element is considered as a special integral
over a polygonal element. The integration over the rectangle is performed as an integration over a polygon enclosed
between two edges, a lower straight edge, an upper one and two coordinate lines. The double integral with two
variables over the element is transformed into two single integrals with one variable of the primitive of the integrand
with respect to the other variable. The sequence is arranged counter clockwise starting from the minimum value of
the variable of integration. Finally, the integration over the rectangle is performed using only one variable in one
direction by making use of two simple formulas for integration over polygonal element. The way of integration is
illustrated by detailed calculation of the stiffness matrix, the mass matrix and the loading vector for a rectangular
plate-bending element. These matrices and vector results in from evaluating the strain energy density, the kinetic
energy and the external work of the element loading, respectively. A program code written in MATLAB is also
presented to assess the results.

keywords: Polygonal element, Fundamental of Calculus, ACM plate bending element

1 Introduction
In [1], two simple formulas for integration of double integrals over arbitrary domain enclosed by curved and/or
straight edges are presented. These formulas will be used here as a basis for deriving the element matrices of the
well-known ACM plate-bending element. Since the formulas are developed to deal with integration over a
polygonal domain, the integration over the rectangular element is considered as a special case of the integration
scheme over a polygon restricted to two edges. The derivation of the finite element approximation is based on
Hamilton’s Principle.
This nonconforming rectangular plate-bending element with three degrees of freedoms each node is one of the
earliest elements developed in the framework of the finite element method. It is already called ACM element in [2]
and the derivation of it is related therein to the works [3], [4], [5]. The mentioned element was a subject of intensive
study in many well-known finite element textbooks published in several editions [6, 7, 8, 9, 10, 11, 12, 13, 14].

In this work, the way of integration of double integrals over a rectangular element using the Riemann integral and
the fundamental of calculus presented in [1] will be used to drive the element matrices for the ACM plate bending
element. The double integral with two variables over the element is transformed into sequences of single integrals
with one variable of its primitive. The sequence is arranged counter clockwise starting from the minimum value of
the variable of integration. Finally, the integration over the element is performed using only one variable in one
direction.
2 Integration over polygon in the Cartesian coordinate system
Let  be a polygonal domain related to a Cartesian coordinate system ( x, y) with the origin o, and the
 
unit vectors (ex , ey ) . Let  be enclosed in the rectangle ( x(1)  x  x(2) ; y(a)  y  y(b) ) and be bounded by
n-edges E(1) , E( 2) ,…, E(n) described counterclockwise through the explicit sequence of equations

E(1) : y  g1( x); x [ x(1) , x( 2) ]

(1.a)
E( 2) : y  g2 ( x); x [ x(2) , x(3) ]
(1.b)
…..

E(n ) : y  gn ( x); x [ x(n) , x(1) ]

(1.n)
The edges are connected by a sequence of vertices (1), (2) (…), (n) (nodal points) with the nodal coordinates

x( p ) , see Figure 1.

 x(1) y (1) 
x y( 2) 
 ; ( x(i )  x(i 1) or y(i )  y(i 1) )
( 2)
x( p ) (2)
 . . 
 
 x( n ) y( n ) 

y (i 1)
y(b ) ( 4)
E(i )
(i  2) E(3)
( 3)

E( n 1)
(n )
A(i ) E( 2)
A( 2) A(1)
E(n ) A( n  2)
( 2)
A( n 1)
(1) E(1)
y(a )
x

x(1) x(n ) x(i  2) x(i 1) x( 4) x(3) x( 2)

Figure 1: Domain enclosed by curved and /or straight edges, Cartesian coordinates of vertices

Suppose that we want to perform double-integrals such as  dxdy

A
or  f ( x, y)dxdy
A
directly in the ( x, y) system,

where f ( x, y) is some function, defined on the domain.

Assume that the total differential of y(r , x) and y  exist, then the following relation applies

dy  ydx
(3)

Now, if the definite integral of the form  dxdy or  f ( x, y)dxdy depends only on the initial and end-values of the
A A
variables x and y , then these can be performed using only one variable as stated in every encyclopaedia for
mathematics, see for instance [15, 16]. Using for example x or y as a variable, the following integral over the total
area of Fig. 1 can be calculated corresponding to Fubini's theorem [17] mentioned in [18] as follows:

I  f ( x, y)dxdy   [ ( f ( x, y)dy)]dx   [ ( f ( x, y)dx]dy

A x y y x
(4)

It is shown in [1] that the definite integral over an arbitrary domain bounded by a sequence of curved and / or
straight edges can be performed using only one variable in x or y direction as a series of single integrals over the
subintervals arranged counter clockwise applying one of the following two formulas:

n x( i 1)

 f ( x, y)dxdy    F ( x, y)|
(5)
Ei dx
A i 1 x ( i )

n y( i 1)

  F ( x, y ) |
(6)

A
f ( x, y )dxdy 
i 1 y( i ) i
Ei dy

F ( x, y) denote the y or x -primitive of f ( x, y) . The indefinite integral or anti-derivative of f ( x, y) with respect to

y or x is as follows:

 f ( x, y) dy  F ( x, y)   ( y)
x
(7)

 f ( x, y) dy  F ( x, y)   ( x)
y
(8)

F ( x, y) |Ei denotes the function F ( x, y) , in which y is replaced by the explicit edge equation
Ei : y  gi ( x) corresponding to Eqns. (1). This means, the edge equation is solved with respect to y and the
expression F ( x, y) |Ei involves terms of x variables only.  ( y ) ,  (x) are free selectable function.

It is understood here that the function f ( x, y) is continuous, measurable and integrable over the subintervals
[ x(1) , x( 2) ],…, [ x( n 1) , x(n ) ],[ x(n ) , x(1) ] .

Now, let us explain how to use this formula for integration over a rectangular element. Consider the rectangular
element depicted in Fig. 2. The domain is bounded by two edges E(1) , E(3) or E( 2) , E( 4) . Every edge-equation can be
determined by the two corresponding vertices spanning the edge. The resulting edge-equations are as follows:
 E(1) : y   b
(9.a)

E( 2) : x   a
(9.b)

E(3) : y   b
(9.c)

E( 4) : x   a
(9.d)

In such case, one of the variables can be explicitly expressed in terms of the other one and the integration can be
performed in one direction. In the following, the integration of the stiffness matrix, the mass matrix and the loading
vector will be demonstrated in details.

2 Deriving the Stiffness matrix of ACM-plate bending element

y
(4) (3)
E(3)
b
E( 4) x
E( 2)
b
E(1)
(1) (2)
a a
Figure 3: Rectangular plate bending element, Cartesian coordinates of vertices, and edges

Detailed information about the classical plate theory can be found in [19, 20]. For studying energy principles the
reader is referred to [21, 22] and for studying structural dynamics the reader may consult [23, 24, 25], just to
mention only some books from a wide range of published books in the literature over the last fifty years.
The finite Element approximation is based on Hamilton’s Principle. The 2D expression for the special case of the
thin plate considered can be written in the absence of the prescribed boundary displacements relating to a Cartesian
coordinate system in the following form:

 1 1 n 

t2
    ij E ijkl  kl dA   q u x0 dA 
3
 
u i  ij

u j dA  F (i ) u x03 (i ) dt  0 (10)
A 2 2A 
t
1
A i 1
where
t1 and t2 are two fixed time points of the vibration process,  is the first variation,
A is the area and dA its differential element, u i is the velocity vector in which both displacement and rotation
components are included,  i j is the corresponding mass density matrix, F (i ) is the concentrated load applied at the
point (i).
 ij is the curvature tensor, which reads expressed in terms of the deflection ux03 ( x1, x 2 ) :
  u 03 1 1 
  x1x1   x ,x x 
   0 
 x 2 x1     u x3 , x 2 x1  (11)
 1 2  0
 u x3 , x1x 2 
 xx   0 

 x 2 x 2  u x3 , x 2 x 2 

E i j k l is the matrix of the force-curvature dependency which reads in a matrix form:

1 0 0  
 0 (1   ) / 2 (1   ) / 2 0
3
Eh  
E ijkl  (12)
12(1   2 )  0 (1   ) / 2 (1   ) / 2 0
 
 0 0 0 

 i j is defined by the following matrix:

 0 0 
 
 ij
 0  h / 12
3
0  (13)
 
0 0  h / 12
3

where  is the material density, E is the modulus of elasticity, h is the plate thickness and  is the Poisson’s
ratio. The indicial notation to indicate the Cartesian variables x1, x 2 is used instead of the x, y - frame and indices
between brackets range over the nodal points.
In Eqn. (10), the internal work associated with the bending and twist moments is only considered.

The plate finite element with the nodal points (i), (j), (k), (l) has three degrees of freedom each node. These are the
displacement normal to the plate surface in x 3 -direction and the two rotations about x1 and x 2 -axes. The total
number of degrees of freedom each element is then represented by the element nodal displacement vector with 12
degrees of freedom

u n (n ) = { u x03 (i ) ,  x1 (i ) ,  x 2 (i ) , u x03 ( j ) ,  x1 ( j ) ,  x 2 ( j ) , u x03 ( k ) ,  x1 ( k ) ,  x 2 ( k ) , u x03 (  ) ,  x1 (  ) ,  x 2 (  ) } (14)

The origin of the coordinate system x1 , x 2 is located at the element centre. The approximation basis is constructed
using the defined local Cartesian variables x1, x 2 in the usual parametric form [2]:

u x03 ( x1, x 2 )  M m ( m)
cm (m)
M  1  x 1
x 2
( x1 ) 2 x1 x 2 ( x 2 )2 ( x1 )3 ( x1 ) 2 x 2 x1 ( x 2 ) 2 ( x 2 )3 ( x1 )3 x 2 x1 ( x 2 )3 

cm(m)  c1(1) c2(1) c3(1) ..........c3(4)  (15)
Linking the free parameters cm (m) to the nodal degrees of freedom using the essential boundary conditions at the
finite element level yield:

ur ( r )  Arm( r( m) ) cm ( m) (16)
 1  a  b a 2 ab b2  a3  a 2b  ab 2  b3 a 3b ab 3 
 
0 0 1 0  a  2b 0 a2 2ab 3b 2  a 3  3ab 2 
0  1 0 2a b 0  3a 2  2ab  b2 0 3a 2b b3 
 
1 a  b a  ab b 2 a b  b3  a 3b  ab 3 
2
a3 2
ab 2

0 0 1 0 a  2b 0 a2  2ab 3b 2 a3 3ab 2 
 
0  1 0  2 a b 0  3a 2 2ab  b2 0 3a 2b b 3  (17)
A m (m)

ab 3 
r (r )
1 a b a2 ab b2 a3 ab 2
ab 2
b3 a 3b
 
0 0 1 0 a 2b 0 a2 2ab 3b 2 a3 3ab 2 
 
0  1 0  2 a  b 0  3a 2  2ab  b2 0  3a 2b  b 3 
1  a b a 2  ab b 2  a3 ab 2
 ab 2
b3  a 3b  ab 3 
 
0 0 1 0 a 2b 0 a2  2ab 3b 2  a 3  3ab 2 
0  1 0 b  3a 2  b2  3a 2b  b 3 
 2a 0 2ab 0

Eliminating the free parameters from Eqn. (15) by solving the linear system of equations (16) and substituting the
result into equation (15), the following relationship between the internal displacements u x03 ( x1 , x 2 ) and the nodal
degrees of freedoms ur (r ) is obtained

ux03 ( x1, x2 )  M m (m) ( Arm( r( m) ) )1ur (r ) (18)

In Eqn. (16), Arm( r( m) ) is the 1212 matrix defined by Eqn. (17) which is derived from M m (m)
by substituting the

coordinates of the element nodes into Eqn. (15) and ( Arm( r( m) ) )1 is the inverse matrix of Arm( r( m) ) .

 1 b a 1 b a 1 b a 1 b a 
 4 8 8 4 8 8 4 8 8 4 8 8 
 3 b 1 3 b 1 3 b 1 3 b 1 
 
 8a 8a 8 8a 8a 8 8a 8a 8 8a 8a 8 
 3 1 a 3 1 a 3 1 a 3 1 a 
 8b 8 8b 8b 8 8b 8b 8 8b 8b 8 8b 
 1 1 1 1 
 0 0 0 0 0 0 0 0 
 8a 8a 8a 8a 
 1 1 1 1 1 1 1 1 1 1 1 1 
 2ab 8a 8b 2ab 8a 8b 2ab 8a 8b 2ab 8a 8b 
 1 1 1 1 
 0 0 0 0 0 0 0 0 
( Arm( r( m) ) ) 1   8b 8b 8b 8b 
 1 1 1 1 1 1 1 1 
0 0 0 0
 8a 3 8a 2 8a 3 8a 2 8a 3 8a 2 8a 3 8a 2 
 1 1 1 1 
 0 0 0 0 0 0 0 0 
 8ab 8ab 8ab 8ab 
 0 1 1 1 1
0 0 0 0 0 0 0 
 8ab 8ab 8ab 8ab 
 1 1 1 1 1 1 1 1 
 3 0 0 0 0 
 8b 8b 2 8b 3 8b 2 8b 3 8b 2 8b 3 8b 2 
 1 1 1 1 
1 1 1 1 (19)
0 0 0 0
 8a 3 b 8a 2 b 8a 3 b 8a 2 b 8a 3 b 8a 2 b 8a 3 b 8a 2 b 
 1 1 1 1 1 1 1 1 
 0 0 0 0 
 8ab 3 8ab 2 8ab 3 8ab 2 8ab 3 8ab 2 8ab 3 8ab 2 
Deriving the curvature tensor  ij using Eqn. (18) and (19) yield

) 1
 ij  p m ( m ) ( Arm( r( m
ij ) ) ur ( r )

0 0 0 2 0 0 6 x1 2x2 0 0 6 x1 x 2 0 
 1 2  (20)
3( x1 ) 2 3( x 2 ) 2 
 
0 0 0 0 1 0 0 2x 2x 0
p m( m)
ij 0 0 0 0 1 0 0 2x 1
2x 2
0 3( x )1 2 2 2
3( x )
 
0 0 0 0 0 2 0 0 2 x1 x2 0 6 x1 x 2 

Applying the expressions of  ij in the first term of Eqn. (10) gives

1 1 1
2A  ij E ijkl  kl dA  ur ( r ) ( Arm( r( m) ) ) 1 (  pijm( m) E ijkl p kln ( n ) dA)( Aqn((qn)) ) 1 uq ( q )  ur ( r ) k r ( r ) q ( q )uq ( q )
2 A
2
(21)

where

k r (r )q(q)  ( Arm( r( m) ) ) 1 (  p mij ( m) E ijkl p nkl ( n) dA)( Aqn((qn)) ) 1 (22)

A
is the element stiffness matrix related to the Cartesian coordinate system.

The integration over the area can be performed using the scheme presented above as follow:
x ( i 1)
n
kr m( m) n ( n )
 ( p 
m( m)
ij
E ijkl
p n( n)
kl
dA)  ( 1 2
p m( m)
ij
E ijkl
p n( n)
kl
dx )dx 2 1
   P
i 1 x ( i )
m( m) n ( n )
r
| Ei dx1 (23)
A x x

where P
r
m( m) n( n)
is the primitive matrix of the 1212 matrix p mij (m) E ijkl p nkl(n) with respect to x 2

P m( m) n ( n )  p
m( m)
E ijkl p n( n) dx 2 (24)
r ij kl
2
x

0 0 0 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0 
 
0 0 0 0 0 0 0 0 0 0 0 0 
 8x 2 6 x1 x 2 9( x 2 ) 2 9 x1 ( x 2 ) 2 
0 0 0 4x2 0 12 x1 x 2 2( x 2 ) 2 6 x1 ( x 2 ) 2 
 5 5 5 5 
 6x2 14 x1 x 2 7( x 2 ) 2 21( x1 ) 2 x 2 7( x )2 3

0 0 0 0 0 0 0 
 5 5 5 5 5 
0 6x2 18 x1 x 2 3( x 2 ) 2 9 x1 ( x 2 ) 2 
 0 0 0 4x2 4 x1 x 2 6( x 2 ) 2 6 x1 ( x 2 ) 2 
5 5 5 5
 1 2 
 1 2 18 x x 1 2 2 1 2 2 18( x1 ) 2 x 2 27 x1 ( x 2 ) 2 1 2 2 2
27( x ) ( x ) 
P m( m) n( n)  D 0 0 0 12 x x 0 36( x ) x 6x (x ) 18( x1 ) 2 ( x 2 ) 2
r 5 5 5 5 
 1 2
14 x x 3( x 2 ) 2 1 2 2
28( x ) x 4( x ) 2 3
17 x1 ( x 2 ) 2 6( x 2 )3 42( x1 )3 x 2 
0 0 0 2( x 2 ) 2 6 x1 ( x 2 ) 2   4 x1 ( x 2 )3 4 x1 ( x 2 )3 
 5 5 5 3 5 5 5 
 6x x 1 2
7( x 2 ) 2 1 2 2
18( x ) x 17 x1 ( x 2 ) 2 28( x 2 )3 21( x ) 
1 4
0 0 0 4 x1 x 2 4( x1 ) 2 x 2  6 x1 ( x 2 ) 2 6( x1 ) 2 ( x 2 ) 2 6( x ) ( x ) 
1 2 2 2

 5 5 5 5 15 10 
0 9( x 2 ) 2 27 x1 ( x 2 ) 2 6( x 2 )3 18 x1 ( x 2 )3 
0 0 0 6( x 2 ) 2 6 x1 ( x 2 ) 2 12( x 2 )3 12 x1 ( x 2 )3
 5 5 5 5 
 1 2 2
21( x ) x 1
9x (x ) 2 2 1 3 2
42( x ) x 1
18 x ( x ) 2 3 1 4 2
63( x ) x 39( x1 ) 2 ( x 2 )3 
0 0 0 6 x1 ( x 2 ) 2 18( x1 ) 2 ( x 2 ) 2  4 x1 ( x 2 )3 6( x1 ) 2 ( x 2 ) 2  12( x1 ) 2 ( x 2 )3 
 5 5 5 5 5 5 
 1
9x (x ) 2 2
7( x 2 ) 3 1 2
27( x ) ( x ) 2 2 1 4
21( x ) 39( x1 ) 2 ( x 2 )3 63( x 2 )5 
0 0 0 6 x1 ( x 2 ) 2 4 x1 ( x 2 )3 6( x1 ) 2 ( x 2 ) 2  12 x1 ( x 2 )3  12( x1 ) 2 ( x 2 ) 3 
 5 5 5 10 5 25 
(25)
The area of the rectangular element is enclosed between the coordinate lines x  a, x  a and the edges 1 1

E1 , E3 so that the previous integration can be simply presented as follows:

n
x ( i 1) a a
kr m( m) n( n)
  
i 1 x ( i )
P m( m) n( n) | Ei dx1  
r  P m( m) n( n) | x 2  b dx1 
r P r
m( m) n( n)
| x 2  b dx1 (26)
a a

P m( m) n( n) | x 2  b is the matrix in which each x 2 is replaced by the edge equation x 2  b .

r
The explicit form of the previous integral takes the following form:

n x( i 1)


kr m( m) n( n)  ( p m( m) E ijkl p n( n) dA) 
ij kl 
1
(
2
p m( m) E ijkl p n( n) dx 2 )dx 1  
ij kl  P
i 1 x ( i )
m(m) n( n)
r
| Ei dx 1
A x x

0 0 0 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0 
 
0 0 0 0 0 0 0 0 0 0 0 0 
 
0 0 0 16ab 0 16ab 0 0 0 0 0 0 
0 0 0 0  8ab(  1) 0 0 0 0 0  8a b(  1)
3
 8ab (  1)
3 
 
0 0 0 16ab 0 16ab 0 0 0 0 0 0 
0 0 0 0 0 0 48a 3 b 0 16a 3 b 0 0 0 
 
kr m( m) n( n)  D 16ab 3 32a 3 b(  1) 
0 0 0 0 0 0 0  0 16ab 
3
0 0
 3 3 
 16a 3 b 32ab 3 (  1) 
0 0 0 0 0 0 16a b
3
0  0 0 0 
 3 3 
0 0 0 0 0 0 0 16ab 3 0 48ab 3 0 0 
 
3 3 72a b(  1)
5
0 0 0 0  8a b(  1)
3
0 0 0 0 0 16a b  8a b (  1)
3 3

 5 
 72 ab 5
(  1) 
0 0 0 0  8ab 3 (  1) 0 0 0 0 0 8a 3 b 3 (  1) 16a 3 b 3  
 5  (27)

Finally, the element stiffness matrix of the rectangular plate bending element can be presented by the product of the
following explicit matrices:

k r (r )q(q)  ( Arm( r( m) ) ) 1 krm( m) n( n) ( Aqn((qn)) ) 1 (28)

The relations presented above can be easily verified using the following MATLAB code.

%===============================================================%
%MATLAB code for deriving the stiffness matrix of the rectangular plate:
%bending element known as ACM integrated using the Fundemental of Calculus :
%===============================================================%
syms x y ;
syms a b E t m;%a=2;b=2;
xip=[-a -b; a -b ;a b;-a b]
D=((E*t^3)/(12.*(1-m^2)))
%===============================================================%
%Symbolic calculation of the stifness matrix of the plate element
%E: Young modulus m Poissons ratio t:thickness
%===============================================================%
M=[ 1 x y x^2 x*y y^2 x^3 x^2*y x*y^2 y^3 x^3*y x*y^3];
Mxx=diff(diff(M,x),x);
Mxy=diff(diff(M,x),y);
Myy=diff(diff(M,y),y);
%c: eleasticity tensor in the x y-system
c=D*[1 0 0 m; 0 (1-m)/2. (1-m)/2. 0; 0 (1-m)/2. (1-m)/2. 0;m 0 0 1];
p=[Mxx;Mxy;Mxy;Myy];
Mi=[M;diff(M,y);-diff(M,x);];
A=[subs(subs(Mi,'x',xip(1,1)),'y',xip(1,2));
subs(subs(Mi,'x',xip(2,1)),'y',xip(2,2));subs(subs(Mi,'x',xip(3,1)),'y',xip(3,2));sub
s(subs(Mi,'x',xip(4,1)),'y',xip(4,2))]
Am1=inv(A)
k=transpose(p)*c*p
%Primitive matrix of the matrix pT*c*p
k=int(k,y)
%Equation of edge E(1)y=-b;equation of edge E(3): y=b
%kr=int(subs(k,y,E(3)),-a,a)-int(subs(k,y,E(1)),-a,a)
kr=-int(subs(k,y,-b),-a,a)-int(subs(k,y,b),a,-a)
k=transpose(Am1)*kr*Am1
simplify(k/D)
%stifness matrix of a biquadrat of side length 4x4 and thhe %following and material
properties %E=1365.0;m=0.3;t=0.2;%D=1
k=subs(subs(subs(subs(subs(k,a,2),b,2),m,0.3),E,1365),t,0.2)
k=vpa(k,6)
%===============================================================%
3 Deriving the mass matrix of the rectangular ACM-plate bending element
The element mass matrix results in from evaluating the third term of Eqn. (10)

1
T3 
2A ui  ij u j dA (29)

The displacement is expanded to include the rotational terms in order to consider their contributions

 u x03 ( x1 , x 2 ) 
 
ui   x01 ( x1 , x 2 )   M im ( m ) ( Arm( r( m) ) ) 1 ur ( r ) (30)
 02 ( x1 , x 2 )
 x 

Where

1 x1 x2 ( x1 ) 2 x1 x 2 (x2 )2 ( x1 ) 3 ( x1 ) 2 x 2 x1 ( x 2 ) 2 ( x 2 )3 ( x1 ) 3 x 2 x1 ( x 2 ) 3 
 
M im( m )  0 0 1 0 x 1
2x 2
0 1 2
(x ) 2x x1 2
3( x ) 2 2 1 3
(x ) 3x1 ( x 2 ) 2 
0 1 0  2 x1  x2 0  3( x1 ) 2  2 x1 x 2  (x2 )2 0  3( x1 ) 2 x 2  3( x )2 2
 

(31)

u x03 ( x1 , x 2 )
 (x , x ) 
0 1 2
(32)
x1
x 2

u x03 ( x1 , x 2 )
 x0 ( x1 , x 2 )   (33)
x1
2

Applying the expressions of ui in the first term of Eqn. (10) gives

1 1 1
2A ui  iju j dA  ur ( r ) ( Arm( r( m) ) ) 1 (  M im( m)  ij M nj ( n ) dA)( Aqn((qn)) ) 1 uq ( q )  ur ( r ) mr ( r ) q ( q )uq ( q )
2 A
2
(34)

where

mr ( r ) q ( q )  ( Arm( r( m) ) )1 (  M im( m)  ij M nj ( n ) dA)( Aqn((qn)) )1 (35)

A
is the element mass matrix related to the Cartesian coordinate system.

The integration over the area can be performed using the scheme presented above as follow:
x ( i 1)
n
I  ( M 
m( m)
i
 Mj
ij n( n)
dA)  ( M m( m)
i
 Mj
ij n( n) 2
dx )dx 1
   Mi 1 x ( i )
m( m) n ( n )
r
| Ei dx1 (36)
A x1 x 2

where M mr ( m) n ( n ) is the primitive matrix of the 1212 matrix M im( m)  ij M nj ( n) with respect to x 2
 M mr ( m) n ( n )   M im( m)  ij M nj ( n ) dx 2 (37)
x2

0 0 0 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0 
 
0 0 0 0 0 0 0 0 0 0 0 0 
 8x2 6 x1 x 2 9( x 2 ) 2 1
9x (x ) 2 2 
0 0 0 4x2 0 12 x1 x 2 2( x 2 ) 2 6 x1 ( x 2 ) 2 
 5 5 5 5 
0 6x2 14 x1 x 2 7( x 2 ) 2 21( x1 ) 2 x 2 2 3
7( x ) 
0 0 0 0 0 0
 5 5 5 5 5 
 6x2 18 x1 x 2 3( x 2 ) 2 9 x1 ( x 2 ) 2 
0 0 0 0 4x2 4 x1 x 2 6( x 2 ) 2 6 x1 ( x 2 ) 2 
 5 5 5 5 
 18 x x1 2
18( x ) x1 2 2
27 x1 ( x 2 ) 2 27( x1 ) 2 ( x 2 ) 2 
M rm ( m ) n ( n )  0 0 0 12 x1 x 2 0 36( x1 ) 2 x 2 6 x1 ( x 2 ) 2 18( x1 ) 2 ( x 2 ) 2 
5 5 5 5
 
0 14 x1 x 2 3( x 2 ) 2 28( x1 ) 2 x 2 4( x 2 )3 17 x1 ( x 2 ) 2 6( x 2 )3 42( x1 )3 x 2 
0 0 2( x 2 ) 2 6 x1 ( x 2 ) 2   4 x1 ( x 2 )3 4 x1 ( x 2 )3
 5 5 5 3 5 5 5 
 6 x1 x 2 7( x 2 ) 2 18( x1 ) 2 x 2 17 x1 ( x 2 ) 2 28( x 2 )3 21( x1 ) 4 
0 0 0 4x x1 2
4( x1 ) 2 x 2  6 x1 ( x 2 ) 2 6( x1 ) 2 ( x 2 ) 2 6( x ) ( x ) 
1 2 2 2

 5 5 5 5 15 10 
0 9( x 2 ) 2 27 x1 ( x 2 ) 2 6( x 2 )3 18 x1 ( x 2 )3 
0 0 0 6( x 2 ) 2 6 x1 ( x 2 ) 2 12( x 2 )3 1
12 x ( x ) 2 3
 5 5 5 5 
 21( x1 ) 2 x 2 9 x1 ( x 2 ) 2 42( x1 )3 x 2 18 x1 ( x 2 )3 63( x1 ) 4 x 2 1 2
39( x ) ( x ) 2 3 
0 0 0 6 x1 ( x 2 ) 2 18( x1 ) 2 ( x 2 ) 2  4 x1 ( x 2 )3 6( x1 ) 2 ( x 2 ) 2  12( x1 ) 2 ( x 2 )3 
 5 5 5 5 5 5 
 9 x1 ( x 2 ) 2 7( x 2 )3 27( x1 ) 2 ( x 2 ) 2 21( x1 ) 4 39( x1 ) 2 ( x 2 )3 63( x 2 )5 
0 0 0 6 x1 ( x 2 ) 2 4 x1 ( x 2 )3 6( x1 ) 2 ( x 2 ) 2  12 x1 ( x 2 )3  12( x1 ) 2 ( x 2 )3 
5 5 5 10 5 25 
(38)
The following simple code based on Fundamental of Calculus gives the exact mass matrix of a rectangular element
and the numerical mass matrix for a biquadrate of side length 4 4

%===============================================================%
%MATLAB code for deriving the mass matrix of the rectangular plate:
%bending element known as ACM integrated using the Fundamental of
Calculus :
%===============================================================%
syms x y ;
syms a b r t;%a=2;b=2;
xip=[-a -b; a -b ;a b;-a b]
%t=0.2;r=5;D=1 E=1365.0;m=0.3
%===============================================================%
%Symbolic calculation of the mass matrix of the plate element
%r: mass density t:thickness
%===============================================================%
M=[ 1 x y x^2 x*y y^2 x^3 x^2*y x*y^2 y^3 x^3*y x*y^3];
%rij: mass density matrix in the x y-system
rij=[r*t 0 0 ; 0 r*(t^3/12) 0; 0 0 r*(t^3/12)];
Mi=[M;diff(M,y);-diff(M,x)]
A=[subs(subs(Mi,'x',xip(1,1)),'y',xip(1,2));
subs(subs(Mi,'x',xip(2,1)),'y',xip(2,2));subs(subs(Mi,'x',xip(3,1)),
'y',xip(3,2));subs(subs(Mi,'x',xip(4,1)),'y',xip(4,2))]
Am1=inv(A)
m=transpose(Mi)*rij*Mi
%Primitive matrix of the matrix MiT*rij*Mi
m=int(m,y)
%E(1)=-b;E(3)=b
%kr=int(subs(k,y,E(3)),-a,a)-int(subs(k,y,E(1)),-a,a)
mr=-int(subs(m,y,-b),-a,a)-int(subs(m,y,b),a,-a)
m=transpose(Am1)*mr*Am1
m=simplify(m)
%mass matrix of a biquadrat of side length 4x4
m=subs(subs(subs(subs(m,a,2),b,2),r,5),t,0.2)
m=vpa(m,6)
%===============================================================%
4 Deriving the load vector of the rectangular ACM-plate bending element
The element load vector results in from evaluating the second term of Eqn. (10)

  q ( x , x )u
1 2 0
T2 x3
( x1 , x 2 )dA (39)
A

The load intensity q ( x1 , x 2 ) at an arbitrary point of the rectangular element subjected to an arbitrary loading
3 3 3 3
(1) ( 2 ) ( 3) ( 4 )
function with the intensities qx (1)
qx ( 2)
,qx (3)
,qx ( 4)
, at the four element nodes , , , is approximated
by the following bilinear form:

q ( x1 , x 2 )  N ( q ) q ( q )
 x 3 (1) 
q 
 3 
x 2  q
x ( 2) 
1 x1 x2 x1 x2 x1 x2 x1 (40)
q ( x1 , x 2 )  (1  )(1  ) (1  )(1  ) (1  )(1  ) (1  )(1  )  
4  a b a b a b a b   x 3 (3) 
q 
 3 
q x ( 4) 
 

Substituting the expressions for the deflection from Eqn.(18) and the loading intensity from Eqn.(36) into Eqn (39)
yield

 ur ( r ) ( Arm( r( m) ) ) 1 ( M  N ( q ) dA) q ( q )  ur ( r ) ( Arm( r( m) ) ) 1 C m( m) q ( q )

m ( m)
T2 (41)
(q)
A
The integration over the area can be performed using the scheme presented above as follow:
x ( i 1)
n
C m( m)
(q)
 ( M 
m ( m)
N ( q ) dA)   1
( M
2
m ( m)
N ( q ) dx 2 )dx1     Cr
i 1 x ( i )
m( m)
(q)
|Ei dx1 (42)
A x x

where Cr m( m)
(q)
is the primitive matrix of the 12 4 matrix M m (m) N(q) with respect to x 2

Cr m( m)  M
m ( m)
N ( q ) dx 2 (43)
(q)
2
x

The area of the rectangular element is enclosed between the coordinate lines x1  a, x1  a and the edges
E1 , E3 so that the previous integral can be simply presented as follows:

n
x ( i 1) a a
C m( m)
(q)
    Cr
i 1 x ( i )
(q)
m( m)
| Ei dx 1
  Cr 
m( m)
(q)
1

| x 2  b dx  Cr m( m) | x 2  b dx1
(q)
(44)
a a

Cr m( m) | x 2  b is the matrix in which each x 2 is replaced by the edge equation x 2  b .

(q)
The explicit form of this matrix is given by the following expression:

 11 19 1 19 
 90 360 45 360 
 b b b b 
 
 30 60 90 45 
 a 
a

a
 
a
 30 45 90 60
 19 11 19 1 
 
 360 90 360 45 
 b b b b 
 60 30 45 90 
 a a a a 
1  
C m( m)   45 30 60 90 
(q)
4ab  1 19 11 19 
 45 360 90 360 
 b b b b
    
 90 45 30 60 
 a a a a 
 90 60 30 45 
 19 1 19 11 
 
 360 45 360 90 
 b 
b

b

b
 45 90 60 30 
 a a a a (45)
 60 
90

45
 
30 

%===============================================================%
%MATLAB code for deriving the load vector for the rectangular plate:
%bending element known as ACM integrated using the Fundemental of
Calculus :
%===============================================================%
syms x y ;
syms a b q1 q2 q3 q4;%a=2;b=2;
xip=[-a -b; a -b ;a b;-a b]
%t=0.2;r=5;D=1 E=1365.0;m=0.3
%D=((E*t^3)/(12.*(1-m^2)))
%===============================================================%
%Symbolic calculation of the load vector for the plate element
%q1,q2,q3,q4: load intensity at nodes 1,2,3,4 respectively
%===============================================================%
M=[ 1 x y x^2 x*y y^2 x^3 x^2*y x*y^2 y^3 x^3*y x*y^3];
Mi=[M;diff(M,y);-diff(M,x)]
A=[subs(subs(Mi,'x',xip(1,1)),'y',xip(1,2));
subs(subs(Mi,'x',xip(2,1)),'y',xip(2,2));subs(subs(Mi,'x',xip(3,1)),
'y',xip(3,2));subs(subs(Mi,'x',xip(4,1)),'y',xip(4,2))]
Am1=inv(A)
%shape function of the load intensity at an arbitrary point of the
element
N=[ 1 x y x*y];
A=[subs(subs(N,'x',-a),'y',-b); subs(subs(N,'x',a),'y',-
b);subs(subs(N,'x',a),'y',b); subs(subs(N,'x',-a),'y',b)];
Nb=N*inv(A)
Cr=transpose(M)*Nb
%Primitive matrix of the matrix MiT*Nb*q
Cr=int(Cr,y)
Cr=int(subs(Cr,y,b),-a,a)-int(subs(Cr,y,-b),-a,a)
L=transpose(Am1)*Cr
L/(4*a*b)
L=simplify(L)*[q1;q2;q3;q4]
%mass matrix of a biquadrat of side length 4x4 under unit load
intensity
L=subs(subs(subs(subs(subs(subs(L,a,2),b,2),q1,1),q2,1),q3,1),q4,1)
L=vpa(L,6)
%===============================================================%
The stiffness matrix of a bi-quadrate with the nodal coordinates

 2.  2. 
~
 2.  2..
x(i p )  
 2 2 
 
 2 2 

associated with the first node reads

 0.66 0.61  0.61

x(i p ) 
  0.61 1.52  0.30
 0.61  0.30 1.52 

As may be seen here, the presented procedure for deriving the element matrices of the rectangular ACM plate-
bending element is easier than that of the conventional one. In the latter one, long expressions for the shape
functions are constructed and complex double integrals are evaluated. The advantages of using the Fundamental of
Calculus are clear even for this simple geometry case. This will be more visible the more complex is the geometry.
Furthermore, the Fundamental of Calculus can deal with special geometry case, too. It is worth mentioning that the
integration over a polygonal element is usually performed by subdividing the polygon into standard elementary
shapes like triangles and quadrilateral and after that applying a mapping technique. For the history of integration
over polygonal element the references [26, 27, 28, 29, 30] are selected to be referred to. In [31], the integration is
curried out without subdivision but the integration is performed numerically after transforming it into boundary
integral using the Stoke's theorem and a mapping procedure.

In the presented procedure the task of integration can be carried out exactly with one variable in one direction by
evaluating boundary integrals. There is also no need for any kind of mapping or any use of Stoke's Theorem or
Gauss Theorem. A MATLAB code [32] is presented with which the results can be assessed.

5 Conclusion

This work shows that the Fundamental of Calculus can also deals with special geometry-domains. The integration
over a rectangular element is considered as a special case of integration over a polygonal element. Element matrices
for the well-known ACM plate-bending element are derived and the derivation is enhanced basically comparing
with the conventional derivation. The same results are obtained. The presented way of integration has several
advantages comparing with the conventional one. There is no need to construct long expressions for the shape
functions or to integrate over the limits of two variables. It reduces also the computational costs especially for
arbitrary domain and enables the construction of explicit element matrices. Such advantages will be clear when we
deal with complex domains enclosed by curved edges.

6 References

[1] Abo Diab, S., Worked Examples on Using the Riemann Integral and the Fundamental of Calculus for
Integration over a Polygonal Element, hal-02306578, (2019).
[2] Clough, R. W., Tocher, J. L., Finite element stiffness matrices for analysis of plate bending, in: Proceedings
of the Conference on Matrix Methods in Structural Mechanics, Wright-Patterson Air Force Base, Ohio,
October 1965, pp. 515-545. Session: Matrix methods in structural dynamics, Finite element properties,
Organizer (Gallagher, R. H.), AFFDL-TR-66-80, 1965.

[3] Adini, A., Clough, R. W. Analysis of plate bending by the finite element method, Rpt. Submitted to the Nat.
Sci. Foundation, Grant G733, 1960.

[4] Melosh, R. J. A stiffness matrix for analysis of thin plates in bending, J. Aeronaut Sci, 1961, 28, pp. 34.

[5] Melosh, R. J. Basis for derivation of matrices for the direct stiffness analysis, AIAA J., 1, (1963), pp. 1631-
1637.

[6] Clough, R. W; Penzien, J., Dynamics of Structures, MC Grow-Hill Book Company , New York ,(1975).

[7] Zienkiewicz, O. C., Taylor R. L., The finite element method, fifth edition, volume 2. Solid Mechanics,
Butterworth-Heinemann, Oxford, 2000.

[8] Zienkiewicz, O. C., Taylor R. L., Zhu, J. Z., The finite element method: Its basis and fundamentals, seventh
edition, Butterworth-Heinemann, Oxford, 2013.

[9] Reddy, J. N., Energy and variational methods in applied mechanics with an introduction to the finite
element method, J. Wiley & Sons, New York, Chichester, Brisbane, Toronto, Singapore, (1984).

[10] Cook, R. D., Malkus, D. S., Plesha, M. E., Witt, R. J., Concepts and applications of finite element analysis,
J. Wiley & Sons, Fourth edition, USA, (2001).

[11] Akin, J. E., Finite Element Analysis with Error Estimators An Introduction to the FEM and Adaptive Error
Analysis for Engineering Students, Elsevier Butterworth-Heinemann,
Amsterdam, …, Tokyo, (2005).

[12] Felippa, C. A.: Introduction to finite element methods, (ASEN 5007) Fall 2017, University of Colorado at
Boulder, Department of Aerospace Engineering Sciences, USA, (2017),

[13] Rao, S. S.: The finite element method in engineering, second edition, United States: N. P., (1986).

[14] Bathe, K. J.: Finite element procedures, second edition, United States, Watertown MA, (2016).

[15] Riemannintegral.EncyclopediaofMathematics.URL:
http://www.encyclopediaofmath.org/index.php?title=Riemann_integral&oldid=38637

[16] Wikipedia.URL: http://en.m.wiki.org/ Riemann_integral

[17] Fubini, G. (1907), Sugli integrali multipli, Rom. Acc. L. Rend. (5), 16 (1): 608– 614, JFM 38.0343.02
Reprinted, in Fubini, G. (1958), Opere scelte, 2, Cremonese, pp. 243–249.

[18] Wikipedia.URL: http://en.wikipedia.org/wiki/Fubini's_theorem.

[19] Timoshenko, S.; Woinowsky-Krieger, S.: Theory of plates and shells, McGraw-Hill Book Company, inc.,
second edition, New York, …, Toronto, (1987).

[20] Ventsel, E.; Krauthammer, T.: Thin Plates and Shells, Theory, Analysis, and Applications, Marcel Dekker,
Inc., New York, Basel, (2001),
[21] Washizu, K., Variational methods in elasticity and plasticity, 3rd ed., Pergamon Press, 1982.

[22] Pilkey, D. W., Wunderlich, W. Mechanics of structures variational and computational methods, CRC
Press, 1994,

[23] Weaver, w., Johnston, P. R., Structural dynamics by finite elements, Prentice-Hall, Englewood cliffs, New
Jersey, (1987).

[24] Leissa, A.W., Vibration of Plates, NASA SP-160. Office of Technology Utilization, Washington, DC,
1969.

[25] Chopra, A. K, Dynamics of Structures: theory and applications to earthquake engineering, fifth edition,
Hoboken, NJ: Pearson Education, Inc., (2017).

[26] Gautam Dasgupta, M.ASCE, Integration within Polygonal Finite Elements, Journal of Aerospace
Engineering / January 2003/ 9, DOI: 10.1061/!ASCE"0893-1321!2003"16:1!9".

[27] Chi, H., Beirão da Veiga, L., Paulino, G.H. , Some basic formulations of the virtual element method (VEM)
for finite deformations. Computer Methods in Applied Mechanics and Engineering, Volume 318, 1 May
2017, pp. 148-192.

[28] Beirão da Veiga, L., Russo, A., Vacca, G., The Virtual Element Method with curved edges, (2018),
arXiv:1711.04306v4[math.NA].

[29] Sukumar N, Tabarraei A, Conforming polygonal finite elements. Int J Numer Methods Eng. 2004;61:2045–
2066.

[30] Sommariva, A., Vianello, M., Product Gauss cubature over polygons based on Green's integration formula,
BIT Numerical Mathematics 47 (2007), 441—453.

[31] Antonietti, P. F., Houston, P., Pennesi, G. , Fast Numerical integration on polytopic meshes with
applications to discontinuous Galerkin finite element methods, Journal of Scientific Computing, August
2018, DOI: 10.1007/s10915-018-0802-y. available at URL:
https://www.mate.polimi.it/biblioteca/add/qmox/03-2018.pdf.

[32] URL: https://www.mathworks.com/

_________________________________________________________________________________________
Address: Sulaiman Abo Diab, Syria, Tartous, Hussain Al Baher
email: [email protected] or [email protected]

This work did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit
sectors.