This action might not be possible to undo. Are you sure you want to continue?
Principle of virtual work and the Finite Element Method
On this subject, there exist a large number of textbooks, many of which are on the shelves of the library. A classic is The Finite Element Method by O.C. Zienkiewicz and R.L. Taylor. A text used in ES 128 is Introduction to Finite Elements in Engineering by T.R. Candrupatla and A.D. Belegundu. In this course, we will give only a very brief introduction to FEM, so that the student gains an impression of the basics. Homework, some of which will be carried out using ABAQUS in the computer lab, will help the student gain some practical knowledge. FEM is perhaps the most significant accomplishment in mechanics in the last century. The method has become such a robust tool that today far more people are users than developers. You do not need sophisticated knowledge of FEM to use a commercial package. However, the method can be extended to solve new problems that are not within the scope of commercial packages. To make such extensions, you may benefit from taking a full course on FEM. Field equations and boundary conditions Consider a body of an elastic material. Inside the body, the following equations hold Force balance:
!" ij !x j + bi = 0
Strain-displacement relations: Stress-strain relations:
1 & 'u 'u # ( ij = $ i + j ! ! 2$ % 'x j 'xi "
" ij = Cijpq! pq .
The surface of the body is divided into two parts, Su and St . On Su , the displacement vector u is prescribed. On St , the traction vector t is prescribed, namely,
! ij n j = ti ,
where ni is the unit vector normal to the surface.
x2 . x3 ) be a function defined in a volume. and the second integral extends the surface of the body. and !u = 0 on Su .ES240 Solid Mechanics Fall 2007 Principle of virtual work Let !u be a virtual displacement field in the body: the field is unrelated to the real displacement field in the body. The first integral extends over the volume of the body. The field is arbitrary. We now apply the divergence theorem to the IVW: 10/29/07 Linear Elasticity-2 . The integral extends over the volume of the body. i i "f The integral on the left hand side extends over the volume. defined by 1 & '(ui '(u j # !. except that it vanishes on the part of the boundary where the actual displacement field is prescribed. Associated with virtual displacement field is a virtual strain field !" ij . Note that the traction is known on St . () ij = $ + ! 2$ ' x ' x i " % j Define the internal virtual work by IVW = ! $ ij"# ij dV . Define the external virtual work by EVW = ! bi"ui dV + ! ti"ui dA . !u = 0 on Su . body force and traction are in equilibrium if and only if the IVW equals the EVW for every virtual displacement field. and the integral on the right hand side extends over the surface enclosing the volume. namely. The divergence theorem states that ! "x dV = ! fn dA . Proof Let f (x1 . The principle of virtual work (PVW) states that the stress.
Please do not look for any physical or philosophical meaning of EVW. except that the displacement field is a fake. the PVW recovers the equilibrium equation in the volume of the body. each factor in front of the corresponding virtual displacement must equal. Disclaimers The EVW looks like actual work done by the body force and the prescribed traction. Note that PVW mentions the actual displacement field in only one place: the virtual displacement field must vanish on the part of the surface where the actual displacement is prescribed. There is none. and the stress-traction relation on St . The virtual displacement field is sometimes considered an arbitrary small variation in the actual displacement field. ! ij n j = ti .ES240 Solid Mechanics Fall 2007 IVW = $ " ij!# ij dV = $ " ij =$ %( " ij!ui ) %" dV & $ ij !ui dV %x j %x j %" ij %x j %!ui dV %x j = $ " ij n j!ui dA & $ !ui dV When IVW = EVW for every virtual displacement field. Nor does the PVW invoke actual strain field. so that EVW really looks like work. The body may as well be a liquid or a plastic solid or whatever. 10/29/07 Linear Elasticity-3 . in the above proof. !" ij !x j + bi = 0 . Nor does the PVW invoke any stress-strain relation. the virtual displacement need not be small at all. The same comments apply to IVW. the PVW does not invoke the actual displacement field. Consequently. However. Otherwise.
This will lead to an approximate solution of the actual displacement field. the PVW is a weak statement of the equilibrium conditions. Use the same procedure to express the virtual strain field in terms of the virtual nodal displacements Use the stress-strain relations to express the stress field in terms of the nodal displacements. 1. the element sizes are finite. The element is the segment between the two nodes. 2. as outlined below. hence the name finite element method. 4. The finite element method is a particular way to construct this subset of virtual displacement fields. the nodal displacements approach the actual displacement field. Require that IVW = EVW for every set of virtual nodal displacements. Interpolate a function of one variable Let us first look at interpolation of a one-variable function f (! ). We can approximate the function at an arbitrary point in the element by: 10/29/07 Linear Elasticity-4 . The two nodes are !1 and ! 2 . When the elements become very small. called the nodal displacements.ES240 Solid Mechanics Fall 2007 From every virtual displacement field to some virtual displacement fields To maintain IVW = EVW for every virtual displacement field is asking for a lot. 3. such as quadrilaterals. This leads to a set of algebraic equations for the nodal displacements. Instead. In this sense. Interpolate the displacement field by the nodal displacements. Use the strain-displacement relations to express the strain field in terms of the nodal displacements. In practice. 5. we will only maintain IVW = EVW for a subset of virtual displacement fields. not infinitesimal. Divide the body into elements. 1. The next few paragraphs walk you through the steps to implement this idea. The values of the function at the two nodes are f1 and f 2 . The basic variables are the displacement vectors at all the nodes. using plane-strain problems and four-node elements. Also interpolate the virtual displacement field by the virtual nodal displacements.
N3. 4. where c is a constant. we must consider quadrilaterals of arbitrary shapes. y3 ). in addition. y2 )." plane. N2. we stipulate that the shape functions are linear in ! . N 2 (!2 ) = 1. Map a square to a quadrilateral To solve a plane strain problem. " ) = c (1 # ! ) (1 # " ) ." ) . To represent a body of a general shape. 2. !2 " !1 !2 " !1 The interpolation is linear in ! . y ) plane. and vanish at the other three nodes. We can write the interpolation in the form f (! ) = N1 (! ) f1 + N 2 (! ) f2 . For example. Thus. N 2 (!1 ) = 0." ) plane map to the four nodes of the quadrilateral on the (x. into quadrilateral elements. If. The simplest function that fulfills the second requirement is N1 (!. (x3 . we set the constant to be c = 1 / 4 . so constructed that the four corners of the square on the (! . y ) plane. y1 ). On the (x. The following functions map a point in the (! . and are known as the shape functions. to the quadrilateral in the (x.ES240 Solid Mechanics Fall 2007 f (! ) = !2 " ! ! " !1 f1 + f2 . y4 ) . y ) plane. !1) = 1 . the four nodes have coordinates (x1 . ( ) 10/29/07 Linear Elasticity-5 . are functions of ! . as 1. we map a unit square on a different plane. (x4 . N1 should be 1 at node 1. Let us determine the shape functions. as given above. 3. N4 are functions of ! and ! . N2. y ) plane: x = N1 x1 + N 2 x2 + N 3 x3 + N 4 x4 y = N1 y1 + N 2 y2 + N 3 y3 + N 4 y4 The functions are linear combinations of the nodal coordinates. To fulfill the first requirement. To ease integration in evaluating the virtual work. N1 (!2 ) = 0. The coefficients N1. (! . on the ! . counterclockwise." ) plane to a point in the (x. (x2 . on the (x. we can uniquely determine the shape functions. Now consider a quadrilateral element. Label its four nodes. They are the shape functions. we divide the cross section of the body. y ) plane. N1 (!1. The coefficients N1. They must satisfy the following requirements: N1 (!1 ) = 1.
u3 . v1 ) . v2 . v2 .v) of a point inside the quadrilateral as ( ) ( ) ( ) u = N1u1 + N 2u2 + N 3u3 + N 4u4 . v4 . u4 . u2 . v3 . This is known as isometric interpolation. List the displacements of the four nodes of the quadrilateral in a column matrix: q=! "u1 . T 10/29/07 Linear Elasticity-6 .ES240 Solid Mechanics Fall 2007 N1 = 1 (1 # " )(1 # ! ) 4 We can similarly write out the other three shape functions: N2 = N3 = N4 = 1 (1 + " )(1 # ! ) 4 1 (1 + " )(1 + ! ) 4 1 (1 # " )(1 + ! ) 4 Interpolate displacement field Let the displacements at the four nodes of the quadrilateral be ( u1 . u4 . v3 . v = N1v1 + N 2v2 + N 3v3 + N 4v4 . v1 . v4 # $ . u2 . Interpolate the displacement (u. u3 . We have used the same shape functions to interpolate the coordinates and the displacements.
we obtain that $ #u % $ #x & #! ' & #! & '=& & #u ' & #x & #" ' & #" ( ) ( #y % $ #u % #! ' & #x ' '& '." ).ES240 Solid Mechanics Fall 2007 Write &N N=$ 1 %0 0 N1 N2 0 0 N2 N3 0 0 N3 N4 0 0# . To calculate the strains. we need to calculate displacement gradients. ε = Bq .! ). in the above interpolations. To gain a general impression of the finite element method. y ) plane. However. the displacement vector at a point in the element is an interpolation of the nodal displacements: u = Nq . & u (1 ( " ) ' * 3' & ' )u4 * We need to convert the gradient on the (! . N4 ! " In the matrix form. such as !u / !x . #y ' & #u ' & ' #" ' ) ( #y ) 10/29/07 Linear Elasticity-7 . Express strains in terms of nodal displacements This step contains tedious algebra. Using the chain rule of differentiation. the displacement is given as a function u (! . where B is a matrix depending on (" . but the final result is easy to understand: We can represent the strain field in terms of the nodal displacements. Let the virtual nodal displacements be !q . a beginning student may as well skip the derivation. the virtual displacement field in the elements is !u = N!q ." ) plane is readily calculated: # %u $ & %" ' 1 # ( (1 ( ! ) (1 ( ! ) & '= & & %u ' 4 ) ( (1 ( " ) ( (1 + " ) & %! ' ) * (1 + ! ) (1 + " ) # u1 $ ( (1 + ! )$ & u2 ' & '. The displacement gradient on the (! ." ) to that on the (x.
10/29/07 Linear Elasticity-8 . and obtain that 'u 'u 1 '* = 'x det J 'u ') J11 J12 J 22 1 & 'u 'u # J 22 ( J12 $ !. #" 4 4 #y 1 1 = (1 $ ! )($ y1 + y2 ) + (1 + ! )( y3 $ y4 ). #" 4 4 The determinant of the Jacobian matrix is det J = J11 J 22 ! J12 J 21 We invert the matrix equation by using Cramer’s rule. y ) ." ) # (x. #" 4 4 #x 1 1 = (1 $ ! )($ x1 + x4 ) + (1 + ! )($ x2 + x3 ) . $ 'u det J % '* ') ! " '* These express the displacement gradients as linear combinations of the nodal displacements: & (u # & ' J 22 ( 1 ' ) )+ J 12 ( 1 ' * ) J 22 ( 1 ' ) )+ J 12 ( 1 + * ) J 22 ( 1 + ) )' J 12 ( 1 + * ) ' J 22 ( 1 + ) )' J 12 ( 1 ' * )# & u1 # $ ! $ (x ! $ ! $u 2 ! 4 det J 4 det J 4 det J 4 det J $ (u ! = $ ' J ( ! 1 ' ) ) ' J 11 ( 1 + * )' J 21 ( 1 ' ) ) J 11 ( 1 + * )' J 21 ( 1 + ) ) J 11 ( 1 ' * )+ J 21 ( 1 + ) ) $u 3 ! $ ! $ 11 1 ' * )+ J 21 ( !$ ! ( y $ " ! % 4 det J 4 det J 4 det J 4 det J " u4 % % " We can also write similar expressions for !v / !x and !v / !y . #" 4 4 #y 1 1 = (1 $ ! )($ y1 + y4 ) + (1 + ! )($ y2 + y3 ) .ES240 Solid Mechanics Fall 2007 This equation relates the gradients in the two planes. The elements of the Jacobian matrix are readily calculated: J11 = J12 = J 21 = J 22 = #x 1 1 = (1 $ ! )($ x1 + x2 ) + (1 + ! )(x3 $ x4 ) . $ det J % '* ') ! " = 'u 1 = 'y det J J 21 'u 1 & 'u 'u # ') = J11 ( J 21 $ !. The two by two matrix is the Jacobian matrix J of the map (! .
The stress column relates to the strain column as σ = Dε . the virtual strain field is δε = B! q .* %x + * %y + * %y %x + Consequently. ! y . 4 det J T Similarly. ! xy $ % . The stress column is also linear in the displacement column: σ = DBq . Written explicitly. the strain column of a point in the element is linear in the nodal displacement column. ! 0. the stress-strain relation under the plane strain conditions is &+ x # * &1 ' * E $ ! $ * 1 '* $+ y ! = ( 1 + * )( 1 ' 2* )$ $+ xy ! $ 0 % 0 % " 0 #& ) x # $ ! 0 ! !$ ) y ! . Assemble equilibrium equations The internal virtual work is 10/29/07 Linear Elasticity-9 .5 ' * ! "$ %( xy " Recall that ε = Bq . . ( + ) ' . For example. Express the stress field in terms of the nodal displacements T List the in-plane stresses by a column: σ = " #! x .ES240 Solid Mechanics Fall 2007 Recall that the strain column is !# %u $ # %v $ # %u %v $ " ε = &( ) . The entries to the matrix B can be worked out readily by combing the above relations. ( ) . B11 = 1 [# J 22 (1 # " )+ J12 (1 # ! )]. We write ε = Bq .
Inserting the above expressions into the expression for the internal virtual work. 10/29/07 Linear Elasticity-10 . The second integral is over the surface of the body. The external virtual work is EVW = " ! uT bdV + " ! uT tdA . The first integral is over the volume of the body.ES240 Solid Mechanics Fall 2007 IVW = $ "# T !dV . The element stiffness matrix is an integral over the volume of the element: k = ! BT DBdV . The force column f is the sum of two contributions. The sum is taken over all the elements. giving fb = ! NT bdV . where the traction is prescribed. giving ft = ! NT tdA . Replace the displacement variation by !u = N!q . we obtain that IVW = ! "qT kq . The surface traction term integrates over the surface area of an element on which traction is prescribed. The body force term integrates over the volume of an element. and we obtain that EVW = ! "qT f The sum is over all the elements in the body. The global stiffness matrix K sums over the contributions from all the elements. Write the internal virtual work in terms of the column of the global degrees of freedom: IVW = ! QT KQ .
For plane problems. the computer solves the linear algebraic equation. The global load column F sums over the contribution of the body force from all elements.! ) = #x d! . We have expressed various fields as functions of (! . the differential volume dV is the differential area dA on the (x." ) plane. Consider a rectangular infinitesimal element on the (! . Differential volume and area To calculate the virtual work. namely." ) = #x d! . After enforcing the prescribed displacements. y ) plane times the thickness h of the element. one side of the quadrilateral is the vector dx! = x(! + d! . y ) plane." ) maps to a vector x(" . we will need to integrate over the volume of the body. ! Q . The body is now divided into elements." ) . (" .! ) maps to point x(" + " . #! 10/29/07 Linear Elasticity-11 ." ) . the displacement column Q satisfies KQ = F .! + d! ).! ). ! QT KQ = ! QT F for virtual nodal displacement column." )$ x(! . (" + d" .! ) on the (x. y ) plane.! + d! ) and (" + d" . and the traction from all surface elements. Consequently. We need to carry out the integral on the (! .! + d! )$ x(" .ES240 Solid Mechanics Fall 2007 List all the nodal displacements in the body by the column matrix Q and write the EVW in terms of the global columns: EVW = ! QT F . The point (! . and then sum over all elements. This infinitesimal element maps to a quadrilateral infinitesimal element on the (x." ) plane. Consequently. defined by four points (! . So.! ). #! Another side of the quadrilateral is the vector dx! = x(" . The point (" + d" . The PVW then requires that IVW = EVW. we will integrate over each element.
the area of the differential element is dA = det Jd ! d" .ES240 Solid Mechanics Fall 2007 Recall that the cross product of any two vectors is a vector. and we obtain that dx" + dx! = )x d" )" )x d! )! i )y d" )" )y d! )! j k ( )x )y )x )y % 0 = k& & )" )! * )! )" # #d"d! . Consequently. In calculating the EVW due to the traction. ' $ 0 Consequently. we will integrate over the surface area of the body. ! = 1 . namely. the area of the quadrilateral infinitesimal element is dA = dx" # dx! . the differential surface area becomes the differential line length dL times the element thickness h. whose magnitude is the area of the parallelogram defined by the two vectors. For the plane elasticity problems. Calculate the cross product. & # & (! # 2 2 10/29/07 Linear Elasticity-12 . Say the line is the edge of a quadrilateral element between node 2 and 3. The differential line length is ' (x $ ' (y $ dL = dx! = % " % (! " " +% % " d! .
+ w f (! ). Gaussian quadrature Consider the integral I= "1 ! f (# )d# . The integral is a linear combination of the function values: I= #1 " f (! )d! = w f (! )+ w f (! )+ . Select a set of n points !1 . ! n and the weights such that the above formula is exact for polynomials of as large a degree as possible. ! = 1 .. 1 A generic way to construct a numerical integration method is as follows..1).. f (! 2 ).. 2 where l23 is the distance between node 2 and 3. ! n in the interval (! 1..... 10/29/07 Linear Elasticity-13 . the volume integrals become k= "1 "1 1 1 ! ! hB 1 1 T DB det Jd$d# fb = "1 "1 ! ! hN b det Jd$d# T The surface integral becomes l ft = 23 ! hNT td# 2 "1 1 Only nodes on the edge of the element.. Thus.. ! 2 .. The Gaussian quadrature selects the points !1 .. Evaluate the function at these points: f (!1 ). ! 2 . 1 1 2 2 n n 1 The coefficients are known as the weights.ES240 Solid Mechanics Fall 2007 or dL = l23 d! . f (! n ). The weights are selected so that the above formula gives exact results for some known functions.. Evaluating function values is costly.. contribute to the force.
! 2 = +1 / 3 Many mathematics handbooks list integration points and weights for Gaussian quadrature. 1 1 2 2 1 We need to determine four numbers: !1 . A two dimensional integral is evaluated as #1 #1 "" 1 1 f (% . To ensure the integral is exact for every polynomial of degree up to 3. all we need to do is to ensure that the formula is exact for the four monomials. "1 = !1 / 3 w2 = 1. the 2-point quadrature in one dimension becomes 4-point quadrature in two dimensions: 10/29/07 Linear Elasticity-14 .ES240 Solid Mechanics Fall 2007 For n points and n weights. The solution is w1 = 1. A polynomial of degree 3 is a linear combination of monomials 1. w1 . 2 = w1 + w2 0 = w1!1 + w2! 2 2 = w1!12 + w2! 22 3 0 = w1!13 + w2! 23 This is a set of nonlinear algebraic equations.$ j ). the formula is exact for polynomial of degree 2n-1. ! 2 . x. Thus. " ! wi f (%i . w2 .$ )d%d$ = #1 i =1 %i . for it has 2n terms. As an example. x2 and x3.$ )d$ = !! wi w j f ( n n n j =1 i =1 1 For example. The integral I is a number linear in the function f. consider the two-point Gaussian quadrature: I= $1 # f (! )d! " w f (! )+ w f (! ).
inside the elements. . Three-dimensional problems If you have to solve a three-dimensional model. you would like to avoid as much as possible three-dimensional problems. This represents a larger family of displacement fields. Element type You do not have to divide the plane into quadrilaterals. 3 3" 3" 3 3" % 3 % 3 3" % Additional comments We have walked through the steps of implementing the finite element method. Even if the computer is not tired of the job size. both input and output are made in the plane (r . 10/29/07 Linear Elasticity-15 .' . Such problems dramatically increase the size of computation job. z ). To divide a body into elements. and are important in applications. !+ f$ !+ f$ ! + f $' !. For example.ES240 Solid Mechanics Fall 2007 '1 '1 ' ( ( f (* . ABAQUS provides many options. When the deformation field is axisymmetric. Some elements have nodes on the sides of the elements. you can increase the number of nodes by using either a large number of low order elements. or a small number of high order elements. These options are available in ABAQUS. such as temperature and electrostatic fields. We simply list them.) )d*d) = f $ % 1 1 & 1 1 # 1 # & 1 & 1 1 # & 1 1 # . ABAQUS provides such an option. as well as at the corners of the elements. the displacements in the element vary as a higher order polynomial of the coordinates. triangular elements may be more versatile in modeling odd-shaped bodies. you will. You do not have to restrict the nodes to the corners of an element. for it will take a long time for you to go through the output to extract useful information. with brief comments. The following extensions are straightforward and have little educational value. If an element contains more nodes. Consequently.' . The amount of work is comparable to solving the plane elasticity problems. all the fields are function of the axial coordinate z and the radial coordinate r. Other fields The finite element method has been developed for many other fields. Axisymmetric problems In practice.