This action might not be possible to undo. Are you sure you want to continue?

FINITE ELEMENT ANALYSIS AS COMPUTATION

What the textbooks don't teach you about finite element analysis Gangan Prathap

Preface

These course notes are specially prepared for instruction into the principles underlying good finite element analysis practice. You might ask: Why another book on FEA (Finite Element Analysis) and the FEM (Finite Element Method) when there are hundreds of good textbooks in the market and perhaps thousands of manuals, handbooks, course notes, documentation manuals and primers available on the subject. Some time earlier, I had written a very advanced monograph on the analytical and philosophical bases for the design and development of the finite elements which go into structural analysis software packages. Perhaps a hundred persons work on element development in the world and out of these, maybe a dozen would probe deep enough to want to explore the foundational bases for the method. This was therefore addressed at a very small audience. In the world at large are several hundred thousand engineers, technicians, teachers and students who routinely use these packages in design, analysis, teaching or study environments. Their needs are well served by the basic textbooks and the documentation manuals that accompany installations of FEM software. However, my experience with students and teachers and technicians and engineers over two decades of interaction is that many if not most are oblivious to the basic principles that drive the method. All of them can understand the superficial basis of the FEM - what goes into the packages; what comes out; how to interpret results and so on. But few could put a finger on why the method does what it does; this deeper examination of the basis of the method is available to a very few. The FEM is now formally over thirty-five years old (the terminology "finite element" having been coined in 1960). It is an approximate method of solving problems that arise in science and engineering. It in fact originated and grew as such, by intuition and inspired guess, by hard work and trial and error. Its origins can be traced to aeronautical and civil engineering practice, mainly from the point of view of structural engineering. Today, it can be used, with clever variations, to solve a wide variety of problems in science and engineering. There are billions of dollars worth of installed software and hardware dedicated to finite element analysis all over the world and perhaps billions of dollars are spent on analysis costs using this software every year. It is therefore a remarkable commercial success of human imagination, skill and craft. The science of structural mechanics is well known. However, FEA, being an approximate method has a uniqueness of its own and these curious features are taken for closer examination in this course. The initial chapters are short ones that examine how FEM is the art of numerical and computational analysis applied to problems arising from structural engineering science. The remaining chapters elucidate the first principles, which govern the discretisation process by which "exact” theory becomes numerical approximations. We shall close these notes with a telescopic look into the present scenario and the shape of things to come in the future.

iii

Acknowledgements: I am extremely grateful to Dr T S Prahlad. for his encouragement and support. Dr S Rajendran and Mr Vinayak Ugru. Dr Sudhakar Marur and Dr Somenath Mukherjee. I owe a great deal to them. Gangan Prathap Bangalore 2001 iv . N.L. Mr Arjun provided secretarial help and also prepared the line drawings and I am thankful to him. Director. They include Dr B P Naganarayana. I am also thankful to Latha and Rahul for their patience and understanding during the preparation of this manuscript. Many colleagues were associated at various stages of writing this book.A.

Paradigms and some approximate solutions 3. Conclusion iii 1 11 28 33 45 57 79 97 111 v . The locking phenomena 6. Introduction: From Science to Computation 2. Completeness and continuity: How to choose shape functions? 4.Contents Preface 1. Shear locking 7. Membrane locking. parasitic shear and incompressible locking 8. Convergence and Errors 5. Stress consistency 9.

by imitating the structural forms already present in Nature. It is to them that this book is addressed.1 From Art through Science to Computation It will do very nicely to start with a description of what an engineer does.). reliability and efficiency. Every plant and animal. safety. These where then copiously used by engineers to fabricate structural forms for civil and military functions. theories of the various structural elements like beams. Rayleigh. tested and verified to emerge as a science. elasticity. To understand the success of structural mechanics. Lagrange. safety. we see the theoretical and mathematical foundations being laid. By far the most popular method in Computational Structural Mechanics is that called the Finite Element Method. Love. 1 . 1. one must understand its place in the larger canvas of classical mechanics and mathematical analysis. plates and shells etc. reliability and efficiency. The examination of how the laws of nature operate in structural form. etc. it is possible to estimate the number of fem computationalists or analysts as lying between a hundred thousand and two hundred thousand. through Galileo and Hooke to Euler. From Archimedes' use of the Principle of Virtual Work to derive the law of the lever. collated. Solid and Structural Mechanics is therefore the scientific basis for the design.000 seats of another industry standard vendor.Chapter 1 Introduction: From Science to Computation 1. etc. Kirchhoff. analytical methods of solution. evaluation and certification of structural forms made from material bodies to ensure proper function. evaluation and certification of structural forms made from material bodies to ensure proper function. refined. have been replaced by computational schemes ideal for implementation on the digital computer. testing. one is led to codification of the accumulated wisdom as science .the laws of mechanics. and sometimes does so giving time as well . He did this first. before we proceed to define briefly how and where structural mechanics plays a crucial role in the life of the engineer. the railway engineer and the aerospace engineer enable travel in enclosed spaces that provide safety with speed in travel. From advertising claims made by major fem and cad software vendors (5000 sites or installations of one package. 32.2 Structural Mechanics A structure is "any assemblage of materials which is intended to sustain loads". We can therefore think of solid and structural mechanics as the scientific basis for the design. Thus. which are too restricted in application. People who use computational devices (hardware and software) with hardly any reference to experiment or theory now perform design and Analysis. the body of knowledge related to construction of structural form is collected. is what we shall call the field of structural or solid mechanics. testing. Today. The structural engineer's appointed office in society is to enclose space for activity and living. From Art imitating Life. the theory of structural form.the ship builder. every artifact made by man or beast has to sustain greater or less forces without breaking.

For nearly a century. buckling (members should not deflect catastrophically under certain loading conditions). this is measured in terms of economy of cost and/or weight. 1. the description often resulted in what are called differential equations. In a few cases. solids and structures came to exist. it was possible to replace the differential equation description with a simple relation between forces and displacements.3. FEM packages are therefore of crucial relevance here.3 From analytical solutions to computational procedures 1. from great mathematicians such as Cauchy. Strength is therefore the primary driver of design. For a more general continuum structure. came the formulation of basic governing differential equations which originated from the application of the infinitesimal calculus to the behavior of structural bodies. As these mathematical models had to take into account the continuum nature of the structural bodies. we shall in fact see that this is a very simple and direct act of the discretization process that is the very essence of the finite element process. there was no recourse 2 . In fact. etc. Lagrange. Navier. as it relates directly to speed of movement and cost of operation. Often. Gifted mathematicians and engineers were able to formulate precise laws that governed the behavior of such systems and could apply these laws through mathematical models that described the behavior accurately. The efficiency of such a structure will depend on how every bit of material that is used in components and joints is put to maximum stress without failing. the engineer could use this limited library of simple solutions to construct meaningful pictures of more complicated situations. Other design drivers are stiffness (parts of the structures must not have excessive deflections). these differential equations could be very complex. Poisson. one could simplify the behavior to gross relationships .the description of deformation and internal forces in an elastic body under the action of external forces using the same mathematical equipment that was been used in other classical branches of physics. However. It is not possible to do this check for structural integrity without having sophisticated tools for analysis. Euler.for a bar. Therefore.especially the triumph of the infinitesimal calculus of continuum behavior in physics. these solutions were made by analytical techniques however. The study is thus complete only if the solutions to such equations could be found. Sophie Germain. It is the role of the structural designer to ensure that the artifacts he designs serve the desired functions with maximum efficiency. such discretizations or simplifications were not obvious at first. the modern trend is to integrate fem analysis tools with solid modeling and CAD/CAM software in a single seamless chain or cycle all the way from concept and design to preparation of tooling instructions for manufacture using numerically controlled machines. but putting only as much flesh as is needed on the bare skeleton. Thus for vehicles. or a spring. these were possible only for a very few situations where by clever conspiracy. Thus. Depending on the sophistication of the description. by ingenuity.1 Introduction We have seen earlier in this chapter how a body of knowledge that governs the behavior of materials. This therefore encouraged the study of the properties of elastic materials and of elasticity . The development and application of complex mathematical tools led to the growth of the branch of mathematical physics.low weight is of the essence. the loads and geometries were so simplified that the problem became tractable.

In this chapter.1. in contexts more general than solid or structural mechanics.x denotes differentiation with respect to coordinate x. Thus a vast body of solutions to the elastic. viable solutions to the governing differential equations. meaning that one can obtain a complete solution. This can become very formidable in cases of complex geometry. and . much before algebra and analysis arrived.1 A simple bar problem under axial load except to find by mathematical and analytical techniques. viscoelastic behavior of bars. however with ingenuity.1 shows its simplest configuration . In fact. beams. solutions could be worked out for a useful variety of structural forms.1 Analytical solution The problem is categorized as a boundary value problem. u(x) is the function describing the variation of displacement of a point.1) where E is the Young’s modulus. This is a statically determinate problem.2. This is the equation of equilibrium describing the rate at which axial force varies along the length of 3 . This will reveal to us how the discretization process is logically set up for such a problem.xx = 0 (1. It was recognized quite early that where analytical techniques fail or were difficult to implement. approximate techniques could be devised to compute the answers. We shall now compute the analytical solution to the problem depicted in Fig. displacements as well as strains and stresses from considerations of equilibrium alone. 1. in tension or in compression. and often were intractable. The bar is a prismatic one-dimensional structure which can resist only axial loads.Fig. we shall review what we mean by an analytical solution for a very simple problem and then proceed to examine some computational solutions. A is the area of cross-section.2 The simple bar problem Perhaps the earliest application of the discretization technique as it appeared in civil engineering practice was to the bar problem. simple approximation and computational schemes were used to solve engineering and scientific problems. Figure 1.3. 1. plastic. 1.1 using very elementary engineering mathematics. plates and shells has been built over the last two centuries.3.a cantilever bar. and cannot take any bending loads. We presume here that for this problem. the reader is able to understand that the governing differential equation describing the situation is EA u. loading and material behavior.

namely Equations (1. e. Two boundary (1.the bar. i.3. To complete the solution. we can show that the following description completely takes stock of the situation: u(x) = Px/EA ε(x) = P/EA σ(x)= Eε(x)= P/A P(x) = Aσ(x) = P (1.2. A solution is obviously u = ax + b where a and b are as yet to determine a and b in conditions shown in Fig.3a) (1. One obvious method at this stage is to start with the equations we have already derived. We shall briefly describe this next. other expressions can be derived which describe further quantities of practical interest.3). strains. 1.g. stress and axial force (stress resultants) along the length of the bar.2 Approximate solutions It is not always possible to determine exact analytical solutions to most engineering problems. etc. In general.1) is further classified as a field equation or condition as one must find a solution to the variable u which must satisfy the equation over the entire field or domain. for x ranging from 0 to L. We saw in the example above that an analytical solution provides a unique mathematical expression describing in complete detail the value of the field variable at every location in the body.4a) (1.2) undetermined constants. Taking these two conditions into account. stress resultants.e. we must now introduce what are called the boundary conditions are needed here and we state them as u = 0 at x = 0 EA u. This can be achieved using a technique known as the finite difference method. conditions.3b) The first relates to the fact that the bar is fixed at the left end and the second denotes that a force P is applied at the free end.4d) where ε(x). 4 . stresses. in this case. From this expression.4c) (1. In most problems where analytical solutions appear infeasible.1. governing differential equations belong to a category called partial differential equations but here we have a simpler form known as an ordinary differential equation. it is necessary to resort to some approximate or numerical methods to obtain these values of practical interest. a unique sense for the boundary (support) and loading 1.x = P at x = L (1.4b) (1. σ(x) and P(x) are the strain. Equation (1.1) and (1. These are the principal quantities that an engineer is interested in while performing stress analysis to check the integrity of any proposed structural design.

unknown..2. the greater will be the accuracy involved. 1.3. details are omitted. This will be the basis for the finite element method as we shall see in subsequent subsections.5b) 5 .3 Finite difference approximations For a problem where the differential equations are already known. the technique known as the finite difference method can be used to obtain an approximate or numerical solution. We see that the finite difference forms for u. u1..x = (ui+1 – ui)/h u.. then it would become easier to understand that an approximation applied directly to the energy or work quantities and a variation carried out subsequently on these terms will preserve both the physics as well as the approximating or numerical solution in one single process.2 Grid for finite difference scheme Other approximate solutions are based on what are called functional descriptions or variational descriptions of the problem.e. u. Figure 1. If we can appreciate that the governing differential equations were derived analytically from such energy or work principles using what is called a variational or virtual work approach. the smaller the grid spacing h). It is easy to show (again. fluid dynamics and heat transfer. un.xx are required.5a) (1. 1. before the finite element method was established.3).1) and (1..x and u. The larger the number of nodal points used (i. at this stage of the computation) at these nodal points. are the unknown variables or degrees of freedom..1) and (1. ui. it was this method which was used to obtain solutions for very complicated problems in structural mechanics. as these can be found in most books on numerical analysis or fem) that at a grid point i.. In fact.Fig. Thus.. The next step is to rewrite the governing differential equations and boundary conditions in finite difference form.3) above. These describe the problem at a level which is more fundamental in the laws of physics than the more specialized descriptions in terms of governing differential equations as seen in Equations (1..xx = (ui+1 – 2ui + ui-1)/h 2 (1. We deal directly with energy or work quantities and discretization or approximation is applied at that level. The field variable is now taken to be known (or very strictly.2 shows a very simple uniformly spaced mesh or grid of nodal points that can be used to discretize the problem described by Equations (1.

It is in such applications that the finite element method proves to be more versatile than the finite difference scheme. The second method is one that originates from a more fundamental statement of the principles involved. One can start with one axiom and derive the other or vice-versa for all of mechanics or most of classical physics. The above illustrates the solution to a very simple one-dimensional problem for which the finite difference procedure yielded the exact answer.6b) (1. make the problem determinate by introducing stress-strain and strain-displacement relations until the final. u3 = 2PL/3EA and u4 = PL/EA.u3 (1. Thus. For non-conservative systems. we start by measuring the total energy stored in a structural system in the form of elastic strain energy and potential due to external loads. In dynamics. etc. and the one that is most often introduced at the earliest stage of the study of structural mechanics. the equivalent principle is known as Hamilton's principle. In structural mechanics. The first. For more complex shapes. Generalizations to two and three dimensions for more complex continuous problems can be made. algebraic equations: u1 = 0 u3 – 2u2 + u1 = 0 u4 .4a) will confirm that we have obtained the exact answers at the grid points. The governing differential equations and boundary conditions are replaced by the following discrete set of linear. especially if the meshes are of regular form and the boundary lines coincide with lines in such regular grids.6a) and (1. or minimum potential energy. a more 6 . the finite difference approach becomes difficult to use.6c) (1. set of governing equations is obtained. Other quantities of interest like strain.2. stress. u2 = PL/3EA. The reader can easily work out that the solution obtained from this set of simultaneous algebraic equations is. solvable. or virtual work. can be computed from the grid point values using the finite difference forms of these quantities. This statement in terms of minimum energy is strictly true only for what are called conservative systems.We now attempt a solution in which four points are used in the finite difference grid. We then derive the position of equilibrium as that state where this energy is an extremum (usually a minimum if this is a stable state of equilibrium). 1. It is a statement that Nature always chooses the path of minimum economy.3.6d) represent the boundary conditions at x = 0 and L respectively. depending on considerations of static determinacy. the law of least action or minimum total potential is as axiomatic as the basic laws of equilibrium are. It recognizes that one of the most basic and elegant principles known to man is the law of least action.6d) where h=L/3 and Equations (1. Comparison with Equation (1. Then.2u3 + u2 = 0 = Ph/EA u4 . There are two main ways in which it can be done. We did not ask how these equations originated.4 Variational formulation We began this chapter with an analytical solution to the differential equations governing the problem. is to formulate the equations of equilibrium in terms of force or stress quantities.6a) (1.

general statement known as the principle of virtual work would apply. From these brief general philosophical reflections we shall now proceed to the special case at hand to see how the energy principle applies. The total energy is stated in the form of a functional. In fact, most of the problems dealt with in structural mechanics can be stated in a variational form as the search for a minimal or stationary point of a functional. The functional, Π, is an integral function of functions which are defined over the domain of the structure. Thus, for our simple bar problem, we must define terms such as the strain energy, U, and the potential due to the load, V, in terms of the field variable u(x). This can be written as

U =

ò

L 1/ 2 EA u 2 ,x dx 0

(1.7a) (1.7b)

V = P ux=L Then,

Π ( υ(x )) =

ò0 1/ 2 EA

L

u 2 ,x dx - P ux = L

(1.8)

Using standard variational procedures from the calculus of variations, we can show that the extremum or stationary value of the functional is obtained as δΠ = 0 (1.9)

If this variation is carried out, after integrating by parts and by regrouping terms, we will obtain the same equation of equilibrium and boundary conditions that appeared in Equations (1.1) and (1.3) above. Our interest is now not to show what can be done with these differential equations but to demonstrate that the approximation or discretization operation can be implemented at the stage where the total potential or functional is defined.

1.3.2.5

Functional approximation

There are several ways in which an approximation can be applied at the level of energy, virtual work or weighted residual statements, e.g. the Rayleigh-Ritz (R-R) procedure, the Galerkin procedure, the least-squares procedure, etc. We first choose a set of independent functions which satisfy the geometric or essential (also called kinematic) boundary conditions of the problem. These functions are called admissible functions. In this case, the condition specified in Equation (1.3a) is the geometric boundary condition. The other condition, Equation (1.3b) is called a nonessential or natural or force boundary condition. The admissible functions need not satisfy the natural boundary conditions (in fact, it can be shown that if they are made to do so, there can even be a deterioration in performance). The approximate admissible configuration for the field variable, say u (x ) , is obtained by a linear combination of these functions using unknown coefficients or parameters, also known as degrees of freedom or generalized coordinates. These unknown constants are then determined to satisfy the extremum or stationary statement. Methods like the Rayleigh-Ritz procedure

7

(the one going to be used now), the Galerkin method, collocation methods, least square methods, etc. all proceed with approximating fields chosen in this way. For our problem, let us choose only one simple linear function, using a polynomial form for the purpose. The advantages of using polynomial functions will become clear later in the book. Thus, only one constant is required, i.e. u (x ) = a1x (1.10)

Note that this configuration is an admissible one satisfying the geometric condition u (0 ) = 0 . By substituting in Equation (1.8) and carrying out the variation prescribed in Equation (1.9), we are trying to determine the value of a1 which provides the best functional approximation to the variational problem. It is left to the reader to show that the approximate solution obtained is u (x ) = Px/EA (1.11a) (1.11b) (1.11c) (1.11d)

ε (x ) = P/EA σ (x ) = Eε(x) = P/A

P (x ) = Aσ(x) = P

Thus, for this problem, the approximate solution coincides with the exact solution. Obviously, we cannot draw any conclusions here as to the errors involved in approximation by the R-R procedure. In more complicated problems, approximate solutions will be built up from several constants and admissible functions and convergence to the exact solution will depend on the number of terms used and the type of functions used. The R-R procedure, when applied in a piecewise manner, over the element domains that constitute the entire structure, becomes the finite element method. Thus, an understanding of how the R-R method works will be crucial to our understanding of the finite element method.

**1.3.2.6 Finite element approximation
**

We shall now describe an approximate solution by the finite element method. We shall use a form known as the displacement-type formulation (also called the stiffness formulation). The simplest element known for our purpose is a two noded line element such as shown in Fig. 1.3. This element is also known as the bar, truss or rod element. It has two nodes at which nodal displacements u1 and u2 are prescribed. These are also called the nodal degrees of freedom. Also indicated in Fig. 1.3 are the forces P1 and P2 acting at these nodes. Thus, any one-dimensional placed from end to end. The element is that stored as nodal now see how this piecewise Ritz structure can be replaced by contiguous elements only information communicated from element to degrees of freedom and the nodal forces. We can approximation is performed.

8

Fig. 1.3 A two-node bar element We first derive an expression for the approximate representation of the field variable, u (x ) , within the region of the element by relating it to the nodal degrees of freedom. It can be shown that,

u (x ) = u1N 1 + u2 N 2

(1.12)

where N1 =(1-ξ)/2 and N2 =(1+ξ)/2,where ξ=x/l. N1 and N2 are called the shape functions. Note that this form is exactly the same as the linear function used in our R-R approximation earlier, Equation (1.10), with generalized coordinates ai being now replaced by nodal quantities ui. It is possible therefore to compute the energy stored in a beam element as

{u1 u2 }

EA 2l

é 1 - 1 ù ìu1 ü ï ï ê ú í ý ê- 1 1ú ïu2 ï ë û î þ

(1.13)

and the potential due to the applied loads as

{u1 u2 } ìP1 ü í ý P

î 2þ

Thus, if a variation is taken over u1 and equilibrium as EA é 1 - 1ù ìu1 ü ê ú í ý = 1û îu2 þ 2l ë- 1

(1.14)

u2, we can write the equation of

ìP1 ü í ý îP2 þ

(1.15)

This matrix representation is typical of the way finite element equations are assembled and manipulated automatically on the digital computer. We shall now attempt to solve our bar problem using a finite element model comprising just one element. Node 1 is therefore placed at the fixed end and node 2 coincides with the free end where the load P2 = P is applied. The fixity condition at node 1 is simply u1 = 0 and P1 indicates the reaction at this end. It is very simple to show from these that u2 = PL/EA. We can see that the exact solution has been obtained. What we have seen above is one of the simplest demonstrations of the finite element method that is possible. Generalizations of this approach to two

9

and three dimensional problems permit complex plate. finite element theorists and analysts have produced a very large body of work that shows how such solutions are managed for a great variety of problems. which are usually very limited in scope. 1. Over the years. we have briefly investigated how classical analytical techniques. shell and three dimensional elasticity problems to be routinely handled by ready made software packages. it is possible to obtain solutions of greater accuracy. By using a very large number of elements. This is in the first-order tradition of the method. can give way to computational methods of obtaining acceptable solutions.4 Concluding remarks In this chapter. We have seen that the finite element method is one such approximate procedure. we must ask. 10 . The remaining part of this book hopes to address questions like: How good is the approximate solution? What kinds of situation affect these approximations in a trivial and in not-so-trivial ways? Since solutions depend very much on the number of elements used and the type of shape function patterns used within each element. what patterns are the best to assume? All these questions are linked to one very fundamental question: In what manner does the finite element method actually compute the solution? Does it make its first approximation of the displacement field directly or on the stress field? The exercises and exposition in this book are aimed at throwing more light on these questions.

the solution is approximated by using a finite number of admissible functions ƒi and a finite number of unknown constants ai so that the approximate displacement field is represented by a linear combination of these functions using the unknown constants.that the RR or fem process actually seeks to find to best advantage. Now. The "aliasing" paradigm. this is done in a piecewise manner . In this 11 . which will be critically discussed later. belongs to this school. The functional now becomes a function of the degrees of freedom (ai or ui as the case may be). In the RR procedure. Thus. In this book however. The equilibrium configuration is obtained by applying the criterion that Π must be stationary with respect to the degrees of freedom.Chapter 2 Paradigms and some approximate solutions 2.over each sub-region (element) of the structure. Which aspect this actually is has been a matter of some intellectual speculation. Three competing paradigms present themselves. the state of stress or strain in the structure. We also saw earlier that the Rayleigh-Ritz (RR) and finite element (fem) approaches offer ways in which approximate solutions can be achieved without the need to solve the differential equations or boundary conditions. we will seek to establish the currency of a third paradigm . It is possible to believe that by "best" we mean that the functions tend to satisfy the differential equations of equilibrium and the stress boundary conditions more and more closely as more terms are added to the RR series or more elements are added to the structural mesh.1 Introduction We saw in the preceding chapter that a continuum problem in structural or solid mechanics can either be described by a set of partial differential equations and boundary conditions or as a functional Π based on the energy principle whose extremum describes the equilibrium state of the problem. will be less accurate. a solution is complete only if analytical functions can be found for the displacement and stress fields. In the fem process. which describe these states. which is approximated to greater accuracy with improved idealization. This is managed by performing the discretization operation directly on the functional. It is assumed that this solution process of seeking the stationary or extremum point of the discretized functional will determine the unknown constants such that these will combine together with the admissible or shape functions to represent some aspect of the problem to some "best" advantage. a continuum has infinitely many material points and therefore has infinitely many degrees of freedom. It follows from this that stresses. exactly everywhere in the domain of the problem. the displacement field is approximated by using shape functions Ni within each subregion and nodal degrees of freedom ui at nodes strategically located so that they connect the elements together without generating gaps or overlaps. The second school of thought believes that it is the displacement field. which are computed as derivatives of the approximate displacement fields. It is not difficult to imagine that such solutions can be found only for a few problems. a real problem with an infinitely large number of degrees of freedom is replaced with a computational model having a finite number of degrees of freedom. Thus.

The aliasing and best-fit paradigms can be thought of as two competing scenarios. commitments. This does not convey much in our context.1]. hunches.1 Bar under uniformly distributed axial load 12 . we shall briefly introduce a short statement on what is meant by the use of the term "paradigm" in the present context. which attempt to explain how the finite element method computes strains and stresses.scenario. Fig. It can be a collection of metaphysical assumptions. we use the word in the greatly enlarged sense in which the philosopher T. S. This is a word that is uncommon to the vocabulary of a trained engineer. 2. Here. In this sense. a paradigm is a "framework of suppositions as to what constitutes problems. Kuhn introduced it in his classic study on scientific progress [2. the displacement fields are computed from these "best-fit" stresses as a consequence. 2. We shall follow this by examining a series of simple approximations to the cantilever bar problem but with more and more complex loading schemes to see how the overall picture emerges.2 What is a "paradigm"? Before we proceed further it may be worthwhile to state what we mean by a paradigm here. heuristic models. Our task will therefore be to establish which paradigm has greater explanatory power and range of application.2]. which are all shared by a scientific community and which provide the conceptual framework within which they can recognize problems and solve them [2. values. The dictionary meaning of paradigm is pattern or model or example. theories and solutions. Before we enter into a detailed examination of the merits or faults of each of these paradigms.

2 as σ.e. we must note that the distributed axial load is consistently lumped at the nodes. Now. 2. Consider next what will happen if two equal length linear (i. We see once again that the nodal predictions are exact. This is only a coincidence for this particular type of problem and nothing more can be read into this fact.2. produced by the consistent load lumping process (see Fig.3a) σ f (x ) = q 0 A ( ) (L 2 ) (2. we must think of a substitute stairstep distribution σf.1b) Consider a one-term RR solution based on ur = a1x . However.2 will be obtained. It is left to the reader to show that the approximate solution obtained is ur (x ) = (q 0 EA ) (Lx 2 ) (2. Instead. 2.3b) We see that the RR and fem solutions are identical. The solution obtained will be uf (x ) = (q 0 EA ) (Lx 2 ) (2.1 or Fig. from Fig. twonoded) bar elements are used to model the bar.3 Bar under uniformly distributed axial load Consider a cantilever bar subjected to a uniformly distributed axial load of intensity q 0 per unit length (Fig. it is easy to show that the analytical solution to the problem is u (x ) = (q 0 EA ) Lx − x 2 2 ( ) (2. This is to be expected because the fem solution is effectively an RR solution. More striking is the observation that the stresses computed by 13 . 2.1a) σ (x ) = (q 0 A ) (L − x ) (2.2b) We now compare this with an fem solution based on a two-noded linear element. Starting with the differential equation of equilibrium.2) which is sensed by the fem stiffness matrix. We can note also the curious coincidence where all three solutions predict the same displacement at the tip. First. where the subscript r denotes the use of the RR approach. a solution of the set of algebraic equations will result in σ f and uf as the fem solution. 2.. It is left to the reader to work out that the solution described in Fig.1). It is possible to speculate that σ r and σ f bear some unique relationship to the true variation σ. 2. It is also clear from Fig. Thus the physical load system that the fem equations are solving is not that described in Fig. 2. 2.1 that σ r = σ f = σ at the mid-point of the beam.2a) σ r (x ) = q 0 A ( ) (L 2 ) (2.1 we can see that the ur and uf are approximate solutions to u.

As we take up further examples later. there is the promise that by carefully understanding what this rule is. ξ = 2x/L-1 has been chosen so that it will also serve as a natural coordinate system taking on values -1 and 1 at the ends of the single three-node bar element shown as modeling the entire bar in Fig. In other words.1 and 2. it will be possible to derive some unequivocal guidelines as to where the stresses are accurate. are more accurate than the stresses.3ξ 2 ö 3 ö ç ÷ ÷ ç ÷ ç è ø ø è ø è (2. A cursory comparison of Figs. Fig.the finite element stairstep fashion.2 Two element model of bar under uniformly distributed axial load. Also. (1-3ξ ). 2. 14 . ξ.4) Some interesting features about this equation can be noted.2 also indicates that in a general sense.1 and 2. where an element capable of yielding constant stresses is used to model a problem where the true stresses vary linearly. It seems compelling now to argue that this is so because the approximate displacements emerge as "discretized" integrals of the stresses or strains and for that reason.3.2. 2. It is easy to establish that the exact stress distribution in this case will be quadratic in nature. In this example. system now approximates the original true stress in a It also seems reasonable to conclude that within each element. the approximate displacements are more accurate than the approximate stresses. These can be identified with the Legendre polynomials and its relevance to the treatment here will become more obvious as we proceed further. 2.2. A dimensionless coordinate. The cantilever bar has a load distributed in a linearly varying fashion (Fig.3). we can generalize from Figs. that the finite element stress magnitudes are being computed according to some precise rule. this will become more firmly established.æ 1 . 2.4 Bar under linearly varying distributed axial load We now take up a slightly more difficult problem. We have also very 2 curiously expanded the quadratic variation using the terms 1. the centroid of the element yields exact predictions. 2. the true state of stress is captured by the finite element stress in a "best-fit" sense. We can write it as σ (x ) = æ q 0 L2 8A ö æ 4 3 − 2ξ .

2. Taking into account the fact that the Legendre polynomials are orthogonal. A comparison of Equations (2. We have not been able to derive it from any general principle that this must be so for stronger reasons than shown here up till now.only the first two Legendre polynomial terms are retained.6) That is.Fig.5) reveals an interesting fact .3 Bar under linearly varying axial load We shall postpone the first obvious approximation. what this means is that in this problem. 1 ò-1 δσ r (σ r − σ ) dξ = 0 15 .6). 2. the RR procedure has determined a σ r that is a "best-fit" of the true state of stress σ in the sense described by the orthogonality condition in Equation (2. This is chosen so that the essential boundary conditions at x=0 is satisfied. we shall consider a two-term series ur = a1x + a2 x 2 . we have obtained σ r in a manner that seems to satisfy the following integral condition: (2.4) and (2.5) This is plotted on Fig. The reader can verify by carrying out the necessary algebra associated with the RR process that the solution obtained will yield an approximate stress pattern given by σ r (x ) = q 0 L2 8A (4 3 − 2ξ ) ( ) (2. For now. This is a result anticipated from our emerging results to the various exercises we have conducted so far.3 as the dashed line. that of using a oneterm series ur = a1x till later. No attempt is made to satisfy the force boundary condition at x=L.

N2 = 1 − ξ ( 2 ) and N 3 = ξ ξ + 1 2. 16 (2. This implies that both have the capability to represent linear stress fields exactly or more complicated stress fields by an approximate linear field that is in some sense a best approximation. 2.5). that the computed fem stress. If the finite element computations are actually performed using the stiffness matrix for the three-node element and the consistent load vector. At first sight. 2. the load system that is being used is the σf system.9) . In the RR process. the computed σ r was responding to a 2 σ f x = æ q 0 L 8A ö ç ÷ ( ) quadratically varying system σ (see Fig. as the reader can assure himself. We could easily establish through Equation (2.2 and 2. any fem discretization automatically replaces a smoothly varying stress system by a step-wise system as shown by σ f in Figs.6) that σ r responded to σ in a "best-fit" manner. It is logical to start here with the three-node element that uses the quadratic shape functions. pattern will be (2.3.3. The question confronting us now is. this does not seem to be entirely unexpected. in what manner did σ f respond to σf . Both the RR process and the fem process here have started out with quadratic admissible functions for the displacement fields. We first compute the consistent loads to be placed ( ) 1 ( ) at the nodes due to the distributed load using Pi = ò N iq dx .is it also consistent with the "best-fit" paradigm? Let us now assume an unknown field σ = c0 + c1 ξ which is a "best-fit" of the stairstep field given by σ f = 2q 0 L2 3 in 0 <x <L/2 and σf = 0 in L/2 <x <L. P2 = q 0 L2 3 and P3 = 0 . Thus. 2. It is this step-wise system that the finite element solution σ f responds to. The resulting load configuration can be represented in the form of a stress system shown as σf.3). it turns out.7) (4 3 − 2ξ ) è ø This turns out to be exactly the same as the σ r computed by the RR process in Equation (2. N = ξ ξ − 1 2 . 2. However. This results in the 2 following scheme of loads at the three nodes identified in Fig. in the fem process. which varies in the stairstep fashion.8) The reader can work out that this leads to σ (x ) = q 0 L2 8A ( ) (4 3 − 2ξ ) . On second thought however. represented by the dashed-dotted lines in Fig.3: P1 = q 0 L 6 .Let us now proceed to an fem solution. We shall determine the constants c0 and c1 so that the "best-fit" variation shown below is satisfied: ò−1 0 δ σ æ σ − 2q 0 L2 3 ö dξ + ç ÷ è ø ò0 δ σ ç σ − 0 ÷ dξ = 0 è ø 1 æ ö (2. there is some more subtlety to be taken care of.

these are also the points used in what is called the two-point rule for GaussLegendre numerical integration. It seems possible now to be optimistic about our ability to ask and answer questions like: What kind of approximating fields are best to start with? and. The link between them is that in both. A comparison of Equation (2. we shall see how the best-fit nature gets disturbed if the loads are not consistently derived.e. σ f is a "best-fit" of both σ and σf! Later. these points are at ξ = ± 1 3 . How good are the computed results? Before we pass on. It again seems reasonable to argue that the nodal displacements computed directly from the stiffness equations from which the stress field σ f has been processed can actually be thought of as being "integrated" from the "best-fit" stress approximation. Our interpretation so far shows that the fact that such points are sampling points of the Gauss-Legendre rule has nothing to do with the fact that similar points arise as points of optimal stress recovery. Such points are well known in the literature of the finite element method as optimal stress points or Barlow points. to see if it operates in the best-fit sense. i.5) and (2. This would be identical to using a single two-node bar element to perform the same function. whether one exists. Again. This must now approximate the quadratically varying stress field σ(x) given by Equation (2. Thus.the zeros of the Legendre polynomials define the optimal points for numerical integration and these points also help determine where the "best-fit" approximation coincides with a polynomial field which is exactly one order higher. ur = a1x . In other words.7) indicates that these are the points where the quadratic Legendre polynomial. This gives us an opportunity to observe what happens to the optimal stress point. and why in this problem. Very often. named after the person who first detected such points. One can show that the one-term approximate solution would lead to the following computed stress: σ r (x ) = q 0 L2 8A ( ) (4 3 ) (2. the Legendre polynomials play a decisive role . (1 − 3ξ ) 2 vanishes.10) 17 . the algebra is very simple and is omitted here. Our presentation shows clearly why such points exist.4). the two issues are confused. where a quadratic stress state is sought to be approximated by a linear stress state. Curiously. the fem process follows exactly the "best-fit" description of computing stress fields. and whether it can be easily identified to coincide with a Gauss integration point.. We shall now return to the linear Ritz admissible function. it is useful for future reference to note that the approximate solutions σ r or σ f intersect the exact solution σ at two points. Such a field is capable of representing only a constant stress. Another important lesson to be learnt from this exercise is that the consistent lumping process preserves the "best-fit" nature of the stress representation and subsequent prediction.which is identical to the result obtained in Equation (2.4) with Equations (2. at ξ = ± 1 3 .7) by carrying out the finite element process.

We can recognize at once that Barlow [2. of the competing paradigms.the sample data points are now the values of the displacements at the nodes and the alias is the function. ± 3 5 0 ± 1/√3 0. We can compare this with the "best-fit" interpretation where the fem is seen to seek discretized strain/stress fields.3] developed his theory of optimal points using an identical idea . as Barlow did earlier [2. ± 1 ξ4 ξ ξ3 ξ2 0. which interpolates the displacements within the element from the nodal displacements.5) is that the one-term solution corresponds to only the constant part of the Legendre polynomial expansion! Thus. An original sine wave is represented in the output of a sampling device by an altered sine wave of lower frequency . the "displacement aliasing" approach leads to subtle differences in interpreting the relationship between the Barlow points and the Gauss points. ξ. The optimal point here is given by ξ = 1 − 4 3 .4) and the computed stress from the two-term solution. ± 3 5 4 1. or as MacNeal did more recently [2. It is instructive now to see how the alternative paradigm. it is clear that the optimal stress point is not easy to identify to coincide with any of the points corresponding to the various Gauss integration rules. which are the substitutes/aliases of the true strain/stress fields in a "best fit" or "best approximation" sense. if any. This concept can be extended to finite element discretization .3].5 The aliasing paradigm We have made out a very strong case for the best-fit paradigm. 2. Let us now use the aliasing concept to derive the location of the optimal points.. However. we can conclude that we have obtained the "best-fit" state of stress even here.4]. the belief that finite elements seek to determine nodal displacements accurately was the basis for Barlow's original derivation of optimal points [2. ± √5/3 ξ 3 ±1/3. The argument that fem procedures look to satisfy the differential equations and boundary conditions does not seem compelling enough to warrant further discussion. σ r (x ) in Equation (2. Let us now examine the merits.ξ indicate polynomial orders 18 . Table 2. Also.4]..3] and is also the basis for what is called the "aliasing" paradigm [2.this is called the alias of the true signal.the term substitute function is used instead of alias.1 Barlow and Gauss points for one-dimensional case p 1 2 3 Nodes at u u ε ξ ξ2 ξ ξ2 ε 1 Gauss points Barlow points ‘best-fit’ aliasing ± 1 0. We assume here that the finite element method seeks discretized displacement fields. ± 1 ξ ξ3 2 0 ± 1/√3 0 ± 1/√3 0. given the orthogonal nature of the Legendre polynomials. The term aliasing is borrowed from sample data theory where it is used to describe the misinterpretation of a time signal by a sampling device.What becomes obvious by comparing this with the true stress σ(x) in Equation (2. which are substitutes or aliases of the true displacement fields by sensing the nodal displacements directly.

it becomes difficult to identify the optimal points by any simple rule. ε and σ .7. the Barlow points (bestfit). Thus. from Table 2.e. implying that the corresponding elements have 2. In this case.e. while ξb will correspond to the points given in References 2. Note that ξa are the points established by Barlow [2. In the simplest. we see that p=1. and the points determined by the best-fit algorithm as ξb. Nodal displacements will be represented by We shall examine three scenarios. strain and stress field will be designated by u . σ = D ε.4 nodes respectively. we consider the case where the true field u is two orders higher than the discretized field u . we take up a case where the rigidity can vary. having varying no. The aliased displacement. This case corresponds to one in which a straightforward application of Legendre polynomials can be made. strain and stress field will be designated by u. ε and σ . one can determine the points where ε = ε as those corresponding to points. σ = D(ξ) ε. the true displacement field u is exactly one polynomial order higher than what the finite element is capable of representing . the Barlow points (aliasing).3 correspond to basis functions of linear..12) with the weight function D(ξ)=1. We shall show below that in Equation 19 . it becomes ò δε T D (ξ ) (ε − ε ) dV = 0 (2. i.1 A one dimensional problem We again take up the simplest problem.5.2. This point was observed very early by Moan [2.6) and that it can be used interchangeably for stresses and strains.6 and 2. We shall see that once again. for Scenarios a and b.2. Note that the natural coordinate system ξ is being used here for convenience. The definition of an identifiable optimal point becomes difficult. We shall assume that the bar is replaced by a single element of varying polynomial order for its basis function (i. linear and quadratic state of strain/stress. we assume that the rigidity of the bar is a constant. strain and stress field will be designated by u ui.12) Note that we can consider Equation (2. See Table 2.3] and MacNeal [2. Scenario a.2 for a list of unnormalised Legendre polynomials. The discretized displacement. i. In Scenario b. We shall adopt the following notation: The true displacement.e. which are the zeros of the Legendre polynomials. a bar under axial loading. In Scenario c. We shall designate the optimal points determined by the aliasing algorithm as ξa. this leads to T ò δε (ε − ε ) dV = 0 (2. These elements are therefore capable of representing a constant. of equally spaced nodes).3. εa and σa. In both Scenarios a and b.11) whereas for Scenario c.5].1.4].11) as a special case of the orthogonality condition in Equation (2. We shall now take for granted that the best-fit rule operates according to the orthogonality condition expressed in Equation (2.. where strain is taken to be the first derivative of the displacement field.we shall see that the Barlow points can be determined exactly in terms of the Gauss points only for this case. quadratic and cubic order. Thus. a .

Scenario a We shall consider fem solutions using a linear (two-node). the optimal point is ξb = 0.13) å ε s Ps (ξ ) (2. then points where ε a = ε are given by ξa=0. Therefore. Let ε = c0 . a quadratic (threenode) and a cubic (four-node) element. as the element is capable of representing only a constant strain. the points of minimum error are the sampling points of the Gauss Legendre integration rule only if ε is exactly one polynomial order lower than ε. Thus. Best-fit: u =linear. making it difficult to identify the optimal points.ξ = b1 + 2b2 ξ = 2 p =1 s=0 (2. Note also that we have simplified the algebra by assuming that strains can be written as derivatives in the natural co-ordinate system. ui = ui . The true displacement field is taken to be one order higher than the discretized field in each case. (The algebra is very elementary and is left to the reader to work out). the point where the Legendre polynomial P1(ξ) = ξ vanishes. Table 2. the Barlow point (aliasing) is ξa = 0. Thus. Equation (2. a Aliasing: At ξi = ± 1 .11). is undetermined at first.14) Note that we have written ε in terms of the Legendre polynomials for future convenience. Linear element (p = 1) u = quadratic = bo + b1 ξ + b2 ξ ε = linear = u. And in Equation (2.2 The Legendre polynomials Pi Polynomial Pi 1 Order of polynomial i 0 1 2 3 4 ξ (1-3ξ ) 2 (3ξ-5ξ ) 3 (3-30ξ +35ξ 2 4 ) 20 .(2. for this case. the optimal points no longer depend on the nature of the Legendre polynomials.11) will now give ε = c0 = b1 . It is now necessary to work out how the algebra differs for the aliasing and best-fit approaches. the Barlow point (best-fit) for this example is ξb = 0.12).

then points where ε a = ε are given by ξ a = ± 1 3 . Aliasing: then points where εa = ε are given by ξ a = 0. Cubic element (p = 3) u = quartic = b0 + b1ξ + b2ξ 2 + b3 ξ 3 + b4ξ 4 (2. Equation (2. Note that the points where the Legendre polynomial P3 (ξ ) = 3ξ − 5ξ 3 ( ) vanishes are ξ0 = 0. for the linear and quadratic elements.17) ε = cubic = u.11) will now give ε = (b 1 + b 3 ) + 2b 2 ξ . i. Equation (2. the Barlow points from both schemes coincide with the Gauss points (the points where the corresponding Legendre polynomials vanish). Best-fit: u = quadratic. Thus. the algebra is left to the reader to work out). . the Barlow points (best-fit) for 3. (Again.ξ (2. Thus.15) = (b1 + b3 ) + 2b2 ξ − b3 æ 1 − 3ξ 2 ö = ç ÷ è ø s=0 å ε sPs (ξ ) (2. ± 5 3 for this case. the Barlow points (aliasing) are ξ a = 0. the Barlow points (best-fit) for ( ) 21 . the optimal polynomial P2 (ξ ) = 1 − 3ξ 2 this example are ξ b = ± 1 ( points ) are ξb = ± 1 3. for this case. Thus. ± 1.18) = s=0 å ε sPs (ξ ) At a ξi = ± 1 3 . Therefore. 3 5 ! Best-fit: u = cubic. Let ε = c0 + c1ξ + c2 1 − 3ξ 2 .Quadratic element (p = 2) u = cubic = b0 + b1 ξ + b2 ξ 2 + b3 ξ 3 ε = quadratic = u. the points where the Legendre vanishes. Thus. ui = ui .ξ = (b1 + b3 ) + (2b2 + 12b4 5 )ξ − b3 1 − 3ξ 2 − 4b4 5 3ξ − 5ξ 3 p =3 ( ) ( ) (2. In our next example we will find that this will not be so.16) a Aliasing: At ξi = 0. Let ε = c0 + c1ξ as this element is capable of representing a linear strain. Note that in these two examples. ± 5 3 . the Barlow points (aliasing) are ξ a = ± 1 3 . ui = ui . as this element is capable of representing a quadratic strain.e. ± 1.11) will now give ( ) ε = (b1 + b3 ) + (2b2 + 12b4 5 ) ξ − b3 1 − 3ξ 2 .

this example P3 (ξ ) = 3ξ − 5ξ ( 3 ) are ξ b = 0. If P p (ξ ) . the best-fit paradigm shows that as long as the discretized strain is one order lower than the true strain. which offers a discretized strain which is of quadratic order. The experience of this writer and many of his colleagues is that the best -fit model.21) Since b3 and b4 are arbitrary. Scenario b: So far. we have to seek the zeros of b3 1 − 3 ξ 2 ( ) + 4b 5 (3ξ − 5ξ ) 3 4 (2.11) allows us to determine the coefficient ci in terms of bi. depending on the problem. If one were to actually solve a problem where the true strain varies cubically using a 4-noded element.19) Equation (2. It is not however a simple matter to determine whether the optimal points coincide with other well-known points like the Gauss points. in such cases. However. unlike in the case of Scenario a earlier. Therefore. we have. Therefore.20) a representation made very easy by the fact that the Legendre polynomials area orthogonal and that therefore ε can be obtained from ε by simple inspection.e. the simple rule is that he optimal points are obtained as P p (ξ b ) = 0 . denoting the Legendre polynomial of order p. These are therefore the set of p Gauss points at which the Legendre polynomial of order p vanishes. 3 5 . is the one that corresponds to reality. we have examined simple scenarios where the true strain is exactly one polynomial order higher than the discretized strain with which it is replaced. it is not possible to seek universally valid points where this would vanish. the points of optimal strain actually coincide with the Gauss points. we have an example where the aliasing paradigm does not give the correct picture about the way the finite element process computes strains. it turns out that ε = (b1 + b3 ) + (2b2 + 12b4 5 ) ξ . a quadratic element (p = 2) modeling a region where the strain and stress field vary cubically. the points where the Legendre polynomial vanishes. Table 2. (2. Thus. It is sufficient to acknowledge that the finite 22 . Consider now a case where the true strain is two orders higher than the discretized strain . the corresponding Gauss points are the optimal points.g. ε = (b1 + b3 ) + (2b2 + 12b4 5 )ξ − b3 1 − 3ξ 2 − 4b4 5 3ξ − 5ξ 3 ε = c0 + c1ξ ( ) ( ) (2. In this example.1 summarizes the results obtained so far. it is not worthwhile to seek points of optimal accuracy. describes the order by which the true strain exceeds the discretized strain.

element procedure yields strains obtain in the circumstances. Hu proposed a generalized theorem. In fact. we take that the strain fields and stress fields are computed from the straindisplacement and constitutive relationships.6. Thus. The `best-fit' manner in which finite elements compute strains can be shown to follow from an interpretation uses the Hu-Washizu theorem.7] shows how by taking an enlarged view of the variational basis for the displacement type fem approach we will be actually led to the conclusion that strains or stresses are always sought in the best-fit manner.e. As we have seen earlier. strain field ε and stress field σ. The minimum potential theorem corresponds to the conventional energy theorem.) Let us now replace the continuum domain by a discretized domain and describe the computed state to be defined by the quantities u . However. i.12). σ = D(ξ)ε.9]. It is clear that ε is an approximation of the true strain field ε. is to introduce a dislocation potential to augment the usual total potential.2. (We project that the strain field ε is derived from the displacement field through the straindisplacement gradient operators of the theory of elasticity and that the stress field is derived from the strain field through the constitutive laws. where again. 2. we have confined attention to problems where σ is related to ε through a simple constant rigidity term. these theorems are the most basic statements that can be made about the laws of nature. 23 .8]. ε and σ . following the interpretation given by Fraejis de Veubeke [2. What the Hu-Washizu theorem does. One such puzzle is the rationale for the "best-fit" paradigm. The orthogonality condition that governs this case is given by Equation (2. it may not be possible to determine universally valid Barlow points a priori if D(ξ) varies considerably. some recent work [2. for applications to problems in structural and solid mechanics. Let the continuum linear elastic problem have an exact solution described by the displacement field u. Its usefulness came to be recognized when one had to grapple with some of the problems raised by finite element modeling. To see how we progress from the continuum domain to the discretized domain. Consider an exercise where (the one-dimensional bar again) the rigidity varies because the cross-sectional area varies or because the modulus of elasticity varies or both. we will find it most convenient to develop the theory from the generalized Hu-Washizu theorem [2.6 The `best-fit' rule from a variational theorem Our investigation here will be complete in all respects if the basis for the best-fit paradigm can be derived logically from a basic principle. ε . This dislocation which can be potential is based on a third independent stress field σ considered to be the Lagrange multiplier removing the lack of compatibility appearing between the true strain field ε and the discretized strain field ε . which had somewhat more flexibility [2. which are the most reasonable one can Scenario c So far.8] rather than the minimum potential theorem.

all the three fields are subject to variation and leads to three sets of equations which can be grouped and classified as follows: Variation on Nature Equilibrium Equation u ∇ σ + terms from P = 0 (2. a least squares correct) `best approximation' of the true state of strain. This satisfies Equation (2. in this more general three field approach.24c) is a variational condition to restore the equilibrium imbalance between σ and σ . formulation. The three-field Hu-Washizu theorem can be stated as δπ = 0 where (2.24b) ε Orthogonality (Equilibrium ) ò δε T æ σ − σ ö dv = 0 ç ÷ è ø (2. Equation (2. 24 . which is the basis for the derivation of the displacement type finite element formulation in most textbooks. only one field (i. we choose σ = σ .24a) shows that the variation on the displacement field u requires that the independent stress field σ must satisfy the equilibrium equations (∇ signifies the operators that describe the equilibrium condition. We can now argue that this imbalance between the exact strain field ε .24a) σ Orthogonality (Compatibility ) ò δσ T (ε − ε ) dv = 0 (2. σ is now an approximation of σ .24c) Equation (2. the displacement field u) is subject to variation. However. that the strains computed by the finite element procedure are a variationally correct (in a sense.23) and P is the potential energy of the prescribed loads. This leaves us Equation (2.22) π = ò ì T íσ ε 2 + σ î T (ε − ε ) ü + P ý dv þ (2. In the simpler minimum total potential principle. In the displacement type the orthogonality condition seen in with the orthogonality condition in tries to restore the compatibility and the discretized strain field ε can be stated as..Note that.25) Thus we see very clearly.24b).e. In the displacement type formulation this ò δσ T (ε − ε ) dV =0 (2.24c) identically.

Having said that the Ritz type procedures determine strains. We then proposed a mechanism to explain how 25 . elements were believed to be "capable of representing the nodal displacements in the field to a good degree of accuracy. 2.1] and Moan [2.it is the strain fields which are computed. is a consequence of the mean value theorem for derivatives. This proof has been taken up in greater detail in Reference 2. Thus terms like σ and σf was actual physical states that were sought to be modeled. displacements are more accurately computed than the strain and stress field. The stiffness equation actually reflecting this integration process and the continuity of fields across element boundaries and suppression of the field values at domain edges being reflected by the imposition of boundary conditions." Each finite element samples the displacements at the nodes. However.8 Conclusions In this chapter. This follows from the generally accepted axiom that derivatives of functions are less accurate than the original functions.7 What does the finite element method do? To understand the motivation for the aliasing paradigm. Barlow [2.g. almost independently as it were within each element.where the RR or fem operations are mechanically carried out using functional approximation and finite element stiffness equations respectively. We worked out a series of simple problems of increasing complexity to establish that our conjecture that strains and stresses appear in a "best-fit" sense could be verified (falsified. we postulated a few models to explain how the displacements type fem works. in the Popperian sense) by carefully designed numerical experiments. Many attempts have been made to give expression to this idea. It must therefore be argued that displacements are more accurate than strains because integrals of smooth functions are generally more accurate than the original data. do exactly the opposite . It is also argued that strains/stresses are usually most inaccurate at the nodes and that they are of greater accuracy near the element centers . It is argued further that as a result. (e. the Ritz approximation process.10] .2.8]. it follows that the displacement fields are then constructed from this in an integral sense. and internally. This can be derived in a formal way. We noticed certain features that seemed to relate these computed stresses to the true system they were modeling in a predictable or repeatable manner. the displacement field is interpolated using the basis functions. but it appears that the most intellectually satisfying proof can be arrived at by starting with a mixed principle known as the Hu-Washizu theorem [2.3 and was reviewed only briefly in the preceding section. We have thus turned the whole argument on its head.7]). it is thought.. An important part of this exercise depended on our careful choice and use of various stress terms. within the element. The stress terms σ r and σ f were the quantities that emerged in what we can call the "first-order tradition" analysis in the language of Sir Karl Popper [2. The strain fields are computed from these using a process that involves differentiation. we have demonstrated convincingly that in fact. and the displacement type fem which can be interpreted as a piecewise Ritz procedure. Thus.this. it will help to remember that it was widely believed that the finite element method sought approximations to the displacement fields and that the strains/stresses were computed by differentiating these fields.

the generalized theorem of Hu was found valuable to determine that the best-fit paradigm had a rational basis. Kluwer Academic Press. One important conclusion we can derive from the best-fit paradigm is that. Int. 1994. Proceedings of the 1973 Tokyo Seminar on Finite Element Analysis. 1993. H.2.9]. On the local distribution of errors by finite element approximations. Our bold conjecture. J. We assumed that this is indeed the mechanism that is operating behind the scenes and derived predicted quantities σ r that will result from the best-fit paradigm when this was applied to the true state of stress. The Finite Element Method in Structural Mechanics. Engng.4. S. we had to enter into a "second-order tradition" exercise. Dordrecht. that ‘it is useless to look for a better solution by injecting additional degrees of freedom in the stresses. University of Chicago Press. MacNeal. Moan.’ We can see that one cannot get stresses. On which it must `do work' in the energy or virtual work principle is actually self-defeating because the higher order terms cannot be `sensed'. 1994. 2. 2. 243-251 (1976). 26 . 2. we shall use this foundation to examine other features of the finite element method. Meth. was to propose that it is effectively seeking a best-fit state. G.5 T.3. R. In this manner. Tokyo. SADHANA.in this regard. intelligence as an 2. The Structure of Scientific Revolution. Elements: Their Design and Performance. S. we could also establish that this could be derived from more basic principles . 1962. Num. Dasgupta. after examining these numerical experiments. Understanding design: Artificial explanatory paradigm. 5-21. In fact. 43-60. Prathap. Theory and Practice in Finite Element Structural Analysis. 2. These predicted quantities turned out to be exactly the same as the quantities computed by the RR and fem procedures. Marcel 2. T. which are of higher order than are reflected in the strain expressions.9 References 2. This is precisely the basis for de Veubeke's famous limitation principle [2. NY. Another important step we took was to prove that the "best-fit" paradigm was neither gratuitous nor fortuitous. In subsequent chapters.1.6. To confirm that this conjecture is scientifically coherent and complete. Finite Dekker. we could satisfy ourselves that the "best-fit" paradigm had successfully survived a falsification test. 19. Barlow.this could have taken place. 1973. J. Kuhn. an interpolation field for the stresses σ (or stress resultants as the case may be) which is of higher order than the strain fields ε. 10. Optimal stress locations in finite element models.

Struct. 1980. Fontana Press. A variational basis for the Barlow points. On some variational principles in the theory of elasticity and the theory of plasticity. H. 2. 27 . G. de Veubeke. 1988. 1993. London. 2. 49.9. Hu. England. B. F. 381-383. Prathap. C. B. Magee. Displacement and equilibrium models in the finite element method. in Stress Analysis. Popper.10.8.7. 4. Ellis Horwood.2. Scientia Sinica. 2. Comput. 33-54 (1955).

quickly. So that φ is continuous within elements and the completeness requirements (see below) are also met. which is the approximate field chosen as trial function. and if possible. interpretations using Legendre polynomials become very convenient. An obvious choice. We want quantities such as displacements. So special care must be taken otherwise. or essences or requirements that the finite element shape functions must show or meet so that convergence is assured. stresses and strain energies to be exactly recovered. this does not ensure that the displacements are compatible between element edges. A structure is sub-divided into sub-regions or elements. We shall see this happy state of affairs appearing frequently in our study here. Two have been accepted for a long time: continuity (or compatibility) and completeness. This leads to the questions: What kind of assumed pattern is best for our trial functions? What are the minimum qualities. However. φ . in the process of representation. When this is done. 3. compatibility of deformation is assured by the fact that the choice of simple polynomial functions for interpolation allows continuous representation of the displacement field. An important question that will immediately come to mind is . Within each element. field or shape functions that we use. gaps or overlaps will develop.How does the accuracy or efficiency of the approximate solution depends on our choice of shape functions. the sequence of trial solutions must approach the exact solution. we mean that as we add more terms to the RR series. then we obviously require that within each element.Chapter 3 Completeness and continuity: How to choose shape functions? 3. and the one most universally preferred is the use of simple polynomials. Accuracy is represented by a quality called convergence. The specification of continuity also depends on how strains are defined in terms of derivatives of the displacement fields. the more important requirement now is that continuity of field variable φ must be maintained across element boundaries - 28 . The overall deformation of the structure is built-up from the values of the displacements at the nodes that form the net or grid and the shape functions within elements. It is possible to use other functions like trigonometric functions but we can see that the least squares or best-fit nature of stress prediction that the finite element process seeks in an instinctive way motivates us to prefer the use of polynomials.1 Introduction We can see from our study so far that the quality of approximation we can achieve by the RR or fem approach depends on the admissible assumed trial. We know that a physical problem can be described by the stationary condition δΠ=0. These functions can be chosen in many ways. surely. where Π=Π(φ) is the functional.2 Continuity This is very easy to visualize and therefore very easy to understand. If Π contains derivatives of the field variable φ to the order m. must contain a complete polynomial of degree m. By convergence. or as we add more nodes and elements into the mesh that replaces the original structure. However.

It has been noticed that formulations that take liberties. Compatibility between adjacent 0 elements for problems which require only C -continuity can be easily assured if the displacements along the side of an element depend only on the displacements specified at all that nodes placed on that edge. we are trying to replace an unknown function φ(x). We desire a trial function set that will ensure that the approximation approaches the exact solution as the number of trial functions is increased. a simple continuity of the displacement fields across element 0 edges suffices . In this case. strains are defined by first derivatives of the displacement fields. In such cases. e. where the total domain is sub-divided into smaller sub-regions. By bringing in the new perspective of a best-fit strain or stress paradigm. as in the classical Kirchhoff-Love theories of plates and shells. We shall however find a large class of problems (Timoshenko beams. Elements. This is called the completeness requirement. We have also seen that what we seek is a best-fit arrangement in some sense between φ (x ) and φ(x). this is known as C -continuity. where strains are based on second derivatives of displacement fields. plane stress/strain flexure. in any approximate method. This phenomenon will be taken up for close examination in a later chapter. completeness must be assured for the shape functions used within each domain. shape or basis functions.this requires that the trial function φ and its derivatives through order m-1 must be continuous across element edges.3 Completeness We have understood so far that in the finite element method. incompressible elasticity) where this simplistic view of continuity does not assure reasonable (i. continuity of first derivatives of displacement fields 1 across element edges is demanded. The continuity requirements then provide the compatibility of the functions across element edges. In the finite element context. It can be argued that the convergence of the trial function set to the exact solution will take place if φ (x ) will be sufficient to represent any well behaved function such as φ(x) as closely as possible as the number of functions used becomes indefinitely large. are called conforming or compatible elements.e. practical) rates of convergence. Mindlin plates and shells. we are able to look at the completeness requirements entirely from 29 . or for that matter. This has important implications in further narrowing the focus of the completeness requirement for finite element applications in particular. we also have the insight that this best-fit can be gainfully interpreted as taking place between strain or stress quantities. significantly improve convergence.this is called C continuity.g. which satisfy the continuity conditions. which is the exact solution to a boundary value problem over a domain enclosed by a boundary by an approximate function φ (x ) which is constituted from a set of trial. the non-conforming or incompatible approaches. From Chapter 2. There are problems. 3. In most natural formulations in solid mechanics.

g. x in a one-dimensional C 0 problem. This becomes very clear from the orthogonality condition derived in Equation 2.e.) 1. if nodal displacements applied to the element are compatible with a constant strain state. i. we are now in a position to make a more meaningful statement about what we mean by completeness.25. the shape functions satisfy the continuity requirement described above. Consider a field φ (usually a displacement) which is interpolated for an n-noded element according to 0 φ= i=1 å N iφi n (3. i.e. y in a two-dimensional C problem so that each element is certain to be able to recover a constant state of strain.1) natural co-ordinates (e. This is called the strain free rigid body motion condition. If polynomial trial functions are used. etc. 30 . It is assumed that these shape functions allow elements to be connected together without generating gaps or overlaps.i. this must be reflected in the strains computed within the element. We would like displacement functions to be so chosen that no straining within an element takes place when nodal displacements equivalent to a rigid body motion of the whole element are applied. In addition. C shape functions have some interesting properties that derive from the completeness requirements imposed by rigid body motion considerations. Most finite element formulations in practical use in general 0 purpose application require only what is called C continuity . Thus i) Ni = 1 when x = xi (or ξ = ξi) and Ni = 0 when x = xj (or ξ = ξj) for i ≠ j. Of paramount importance now is the idea that the approximate strains derived from the set of trial functions chosen must be capable in the limit of approximation (i.what we desire the strains or stresses to be like. argue that one property the the distribution of φ within at node i has unit value and where the Ni are assumed to be interpolated using η for a 2-dimensional problem).1 Some properties of shape functions for C0 elements The displacement field is approximated by using shape functions Ni within each element and these are linked to nodal degrees of freedom ui. From the foregoing discussion. 1. as the number of terms becomes very large) to describe the true strain or stress fields exactly. will meet this requirement. it will be necessary that each element must be able to reproduce a state of constant strain. x. a field quantity φ must have interelement continuity but does not require the continuity of all of its derivatives across element boundaries. (e. 3.e. ξ. There are simple rules that allow these conditions to be met and these are called the completeness requirements. It is simple to shape functions must have is that Ni must define the element domain when the degree of freedom φi all other nodal φ’s are zero. then a simple assurance that the polynomial functions 0 contain the constant and linear terms.g.e.3.

3) will contain enough terms to ensure completeness. the computed strain vanishes at the origin! 31 . The reader is encouraged to show that this entails conditions such as iii) Σ Ni. The omission of this term has led to the loss of completeness. the reader can verify that as long as this term is left out of any trial set containing polynomials of higher order. We could at the outset arrange for a0 to be zero to satisfy the essential boundary condition.1 shows how erroneous the computed stresses σ2. This offers a very simple check for shape functions. where derivatives of φ are also used as nodal degrees of freedom. Thus from Eq 3.ξ = 0. 1 3. a linear variation is seen instead of the actual constant state of stress. because of the absence of the linear term in u (the constant strain term therefore vanishes). With u1 (x ) = a1x we were able to obtain the exact solution to the problem. σ3 and σ2+3 are. The exact solution for this problem is given by a state of constant strain. i. We can now investigate what will happen if we omit the a1x term and start out with the trial function set u2 (x ) = a2 x 2 . 2. Another useful check comes again from considerations of rigid body motion . Note that for C elements.Consider now.1 we have ii) Σ Ni = 1. The term that we have omitted is the one needed to cover the state of constant strain.e. The reason for this is now apparent. this very term also provides the exact answer. The reader is now encouraged to experiment with RR solutions 2 based 3 on other incomplete sets like u2 +3 (x ) = a2 x u3 (x ) = a3 x 3 and + a3 x . Thus. In the present example. It is left to the reader to show that this results in a computed stress variation σ 2 = 3Px 2EAL .that a condition of zero strain must be produced when rigid body displacements are applied to the nodes.1. In this case.η = 0 for a 2-dimensional problem. the case of the element being subjected to a rigid body motion so that at each node a displacement ui=1 is applied. It is obvious from physical considerations that every point in the element must have displacements u=1. Σ Ni. Figure 3.4 A numerical experiment regarding completeness Let us now return to the uniform bar carrying an axial load P as shown in Fig. the polynomial series u (x ) = a0 + a1x (see Section 2. In fact. there will not be any sort of clear convergence to the true state of strain. these rules apply only to those Ni which are associated with the translational degrees of freedom.

Unfortunately. The problems encountered were called locking. These difficulties are most pronounced in the lowest order finite elements. In the next chapter. elements derived rigorously from only these basic paradigms can behave in unreasonably erratic ways in many important situations. This was the received wisdom or reigning paradigm around 1977. parasitic shear. As a very good example. solutions from coarse meshes will be reasonably accurate and that convergence will be acceptably rapid with mesh refinement. more recently. for a very long time. 32 .Fig. 3. It is also hoped that if these requirements are met. Some of the problems may have gone unrecorded with no adequate framework or terminology to classify them. etc. This shall be taken up from Chapter 5. the correct error inducing mechanism has been discovered and these problems have come to be called membrane locking. An enlargement of the paradigm is now inevitable to take these strange features into account. we shall look at the twin issues of errors and convergence from this point of view. However.5 Concluding remarks We have now laid down some specific qualities (continuity and completeness) that trial functions must satisfy to ensure that approximate solutions converge to correct results. it was believed that curved beam and shell elements performed poorly because they could not meet the strain-free rigid body motion condition.1 Stresses from RR solutions based on incomplete polynomials 3.

To understand its proper place. which can be said to belong to the "second-order tradition" of knowledge.1 Introduction Error analysis is that aspect of fem knowledge. We shall assume here that in most problems we deal with. Here. Error analysis therefore belongs to this tradition of epistemological enquiry. let us briefly review what has gone before in this text. The real issue that is left for us to grapple with is that the computational model prepared to simulate the mathematical model may be faulty and can lead to errors. If the mathematical model does not explain the actual physical system as observed through carefully designed empirical investigations. We have seen structural engineering appear first as art and technology. It too grew first as art. errors originate in many ways. The finite element method is one such device. the need for an outer loop or second-order tradition of enquiry becomes obvious. we can argue that this will depend on the type and shape of elements we use. From physical intuition. How do we know why we should do it this way? If the first stage is the stage of action. we can see this as a first-order tradition or level or inner loop of fem art and practice. This is one stage of the learning process and it is now assumed that over the last few centuries we have gone through this stage successfully enough to accept our mathematical models without further doubt or uncertainty. then the models must be refined or revised so that closer agreement with the experiment is obtained. the way distributed loads are assigned to nodes. We must first understand that the fem starts with the basic premises of its paradigms. the number of elements used and the grading or density of the mesh used. The science of solid and structural mechanics codified this knowledge in more precise terms. starting as the imitation of structural form by human fabrication. this second stage is now of reflection. there are errors which occur due to the fact that information in the form of numbers can be stored only to a finite precision (word length as it is called) at every stage of the computation. by systematic practice and refinement. The answer may be wrong because the mathematical model wrongly described the physical model. Let us now translate what we mean by error analysis into simpler terms in the present context. In epistemological terms. These are called round-off errors. the manner in which boundary 33 . With the emergence of cheap computational power. its definitions and its operational procedures to provide numerical results to physical problems which are already described by mathematical models that make analytical quantification possible. Mathematical models were then created from this to describe the structural behavior of simple and complex systems analytically and quantitatively. In the process of replacing the continuum region by finite elements. we know what to do and how to do it. However. Since our computational models are now created and manipulated using digital computers. We must take care to understand that this is not the issue here. word length is sufficient so that round-off error is not a major headache.Chapter 4 Convergence and Errors 4. techniques for computational simulation emerged.

analytical quantification of these errors is possible. error analysis. must ideally be founded on a priori projections computed from intelligent paradigms which can be verified (falsified) by digital computation. has introduced a certain predictable degree of error and that this converges at a predictable rate. the error is removed in a predictable manner as the discretization is improved in terms of mesh refinement. In this chapter. Therefore. due to finite element discretization must diminish 2 rapidly. instead of being only a posteriori or post mortem studies. It seems possible that to some extent. as it is usually practised. i. or engineering methodology. Error analysis is the attempt to make such predictions a priori. after the errors are found. This is a very crucial element of our study. We next take note of the newly established paradigm that the Ritz-type and fem approximations seek strains/stresses in a `best-fit' manner. We can only rely on heuristic judgement to understand how best to minimize errors. or rationalize the errors in a logical way. We can recognize two kinds of discretization error belonging to this category. can be said to have acquired a scientific basis only when it has incorporated within itself. whether in displacements. we examine if it is possible to argue that errors. at least in a (l/L) manner or better. a posteriori. These error estimates or predictions can be seen as consequences computed from our guesses about how the fem works. The first kind is that which arises because a model replaces a problem with an infinitely large number of degrees of freedom with a finite number of degrees of freedom.conditions are modeled by specification of nodal degrees of freedom. or for that matter.e. Therefore. Thus. However. These must be set apart from the first-order tradition procedures that carry out the discretization (creating the computational model from the mathematical model) and solution (computing the approximate results). then we can assure ourselves that we have carved out a scientific basis for error analysis. In both cases. which led to the computational model. we must design auxiliary procedures that can trace errors in an a priori fashion from basic paradigms (conjectures or guesses). Thus. stresses or energies. except in very rare cases. To carry out error analysis. Most of such errors are difficult to quantify analytically or determine in a logically coherent way. 34 . these auxiliary procedures that permit its own self-criticism. the mathematical model can alter the physics of the problem. the governing differential equations and boundary conditions are satisfied only approximately. These errors must now be verified by constructing simple digital computation exercises. new procedures have to be invented. The fem. which are an approximation of the true solution to the mathematical model. If this cycle can be completed. The second kind of error appears due to the fact that by overlooking certain essential requirements beyond that specified by continuity and completeness. where a large structure (domain) of dimension L is sub-divided into elements (sub-domains) of dimension l. we shall first take stock of the conventional wisdom regarding convergence. any body of engineering knowledge. These are the discretization errors that can occur. This is what we seek to do now. etc. From such an interpretation. we shall now look only at that category of discretization error that appears because the computational or discretized model uses trial functions. This is based on the old paradigm that fem seeks to approximate displacements accurately. we must be able to satisfy ourselves that the discretization process.

3). in fact while the true solution required a quadratic variation of the displacement field u(x). The simple example below will highlight the essential features of this approach.1 Error analysis of the axially loaded bar problem Let us now go back to the case of the cantilever bar subjected to a uniformly distributed axial load of intensity q0 (Section 2. if a series of approximate solutions using grids whose mesh sizes are uniformly reduced is available. provided there is some means to establish the rate at which errors are removed with mesh size. 4. In subsequent chapters. there are some inherent limitations in this approach that will become clear later.1) We shall use the two-node linear bar elements to model this problem. 4. It is usual to describe the magnitude of error in terms of the mesh size. It is also possible with suitable interpretations to cast finite difference approximations as particular cases of weighted residual approximations using finite element type trial functions. If for convenience we choose q 0 AE = 1 .new paradigms need to be introduced and falsified. 4.the discretization fail in a dramatic fashion. then we can show that u(x) = Lx − x 2 2 u (x) = Lx −(l 2 (4. which gives exact displacements at the nodes. However. Fig. This is the usual range of problems where the continuity and completeness paradigms explain completely the performance of finite elements.xx + q 0 AE = 0 (4. It was also clear to us that this did not mean that an exact solution had been obtained. Thus. it may be possible to obtain more information about the exact answer by some form of extrapolation. Thus. the finite element solution u (x) was piecewise linear. within each element. The equilibrium equation for this problem is u. we shall discover however that a class of problems exist where errors are much larger .2a) 2 xi ) + 2xxi − 2 (4.2 Traditional order of error analysis The order of error analysis approach that is traditionally used is inherited from finite difference approaches. This seems reasonable because quite often simple finite difference and finite element approximations result in identical equations. errors must not be more than a few percent.2b) 35 .with ten elements in a one-dimensional structure. there is some error at locations between the nodes. Convergence and error analysis must now be founded on a more complex conceptual framework .1 shows the displacement and strain error in the region of an element of length 2l placed with its centroid at xi in a cantilever bar of length L.2. We have seen that this leads to a solution. The order of error analysis proceeds from the understanding that the finite element method seeks approximations for the displacement fields. The errors are therefore interpreted using Taylor Series expansions for the true displacement fields and truncations of these to represent the discretized fields. This will be postponed to subsequent chapters.

x=xi.1. 4.1 Displacement and strain error in a uniform bar under distributed axial load q 0 ε(x)=L–x (4. which is the dimensionless quantity that indicates the mesh division relative to the size of the structure.u ∈ (x ) ∈ ∈ (x ) Fig.3a) (4. in terms of the parameter h = l L . the strain error vanishes at the element centroid. What is more important to us is to derive measures for these errors in terms of the element size l. From Equations (4. the displacement error is a maximum at x=xi. 4. e u max = O h 2 ( ) 36 (4. we can argue that in this case.2c) (4. and that it is a maximum at the element nodes. It is also clear that for this problem.2d) ε(x) = L − xi If we denote the errors in the displacement field and strain field by e(x) and e’(x) respectively. It will also be useful to have these errors normalized with respect to typical displacements and strains occurring in the problem.3b) From this.3).4a) .2) and (4. This is clear from Fig. where “O” stands for "order". or more usefully. we can show that e (x ) = (x − xi + l ) (x − xi − l ) 2 e ′(x ) = (x − xi ) (4. we can estimate the maximum normalized errors to be of the following orders of magnitude.

it is very easy to associate this load with the deflection under the load using Castigliano's theorem. then the computed displacement field will have an error O(h ). It is possible to generalize these findings in an approximate or tentative way for more complex problems discretized using elements of greater precision. 4. If the exact solution requires a polynomial field of degree q at q least. for errors will also depend on how the loads are lumped at the nodes and so on. Our challenge is therefore to see if the best-fit paradigm can be used to predict the convergence rate for this example from first principles.e′ ε max = O (h ) (4. Thus. If this is accepted. which are exact. is not suitable for this purpose. It is not clear to us in this analysis that the discretized strain ε (x ) is a "best-fit" of the true strain ε (x ) . Since the bending moment varies linearly over the beam for this problem. Of course. We should therefore look for an example where nodal displacements are not exact. then it is possible to show that the error in 2 the strain energy stored in such a situation is O(h ) as well. 4. We must now choose a suitable example to demonstrate in a very simple fashion how the convergence of the solution can be related to the best-fit paradigm. It can actually be proved that this arises from the special mathematical nature of this problem. The example that is ideally suited for this purpose is a uniform beam with a tip load as shown in Fig. A case such as the uniform bar with linearly varying axial load modeled with two-node elements gives nodal deflections. the trial functions are of degree q-1.4b) Note that the order of error we have estimated for the displacement field is that for a location inside the element as in this problem. these measures are tentative. if an element with q nodes is used. the finite element will replace this with a stairstep approximation. Thus. only a linear approximate trial function is possible. We shall model it with linear (two-node) Timoshenko beam elements which represent the bending moment within each element by a constant. It will then be possible to discern the convergence for the tip deflection. the nodal deflections are exact. Since the applied load is at the tip. If strains th for the problem are obtained as the r derivative of the displacement field. the error in strain vanishes.2. in each element. the stress pattern will approach the true solution more closely and therefore the computed strain energy due to bending will also converge. The bar under axial load. At element centroids. with increase in number of elements. 37 . We shall now use this argument to show how convergence and errors in a typical finite element model of a simple problem can be interpreted. q-r then the error in the strain or stress is O(h ). We shall see that a much better insight into error and convergence analysis emerges if we base our treatment on the `best-fit' paradigm. even though the true displacement field is cubic but between nodes.3 Errors and convergence from "best-fit" paradigm We have argued earlier that the finite element method actually proceeds to compute strains/stresses in a `best-fit' manner within each element. The strain error is O(h) while the displacement error is 2 O(h ). an example we have used quite frequently so far.

2 and 4 element idealizations of the present problem using the linear element. After correcting for the inconsistent shear strain.the simplest representation possible under the circumstances and therefore an advantage in seeing how it works in this problem.000 38 .000 2 . the elements pick up the bending moment at the centroid correctly . as that of the deflection at the tip under the load. We shall now interpret accuracy in the conventional sense. The example chosen represents a thin beam so that the influence of shear deformation and shear strain energy is negligible. 4. it is doing so in a `best-fit' manner.0. this element permits constant bending and shears strain accuracy within each element .2) are chosen such that the tip deflection under the load will be w=4.Fig. We shall restrict attention to the bending moment variation as we assume that the potential energy stored is mainly due to bending strain and that we can neglect the transverse shear strain energy for the dimensions chosen.1 Cantilever beam idealized with linear Timoshenko beam elements The dimensions of a cantilever under tip load (Fig. This element serves to demonstrate the features of shear locking which arise from an inconsistent definition of the shear strains (It is therefore useful to take up this element for study later in Chapter 6). The computed (i. The true bending moment (shown by the solid line) varies linearly.e.984 . Figure 4.i. Table 4.3.2 Cantilever beam under tip load 4.000 3 .750 .750 1. discretized) bending moments are distributed in a piecewise constant manner as shown by the broken lines. L/t = 100 No.938 . In each case.938 1.1 shows the Table 4. We shall now discretize the beam using equal length linear elements based on Timoshenko theory.3 shows the bending moment diagrams for a 1.1 Normalized tip deflections of a thin cantilever beam. What we shall now attempt to do for this problem is to relate this to the accuracy of results.e. We use this element instead of the classical beam element for several reasons.984 1. 4. of elements Predicted rate Element without RBF Element with RBF 1 .

We can show that Fig. Thus the true bending moment over the element region for our problem can be taken to vary as M+Vx. from a continuum analysis) and that given by the finite element (discretized) model.e. Let the moment and shear force at the centroid be M and V. 4.3 Bending moment diagrams for a one-. 4. 4.4). We shall now compute the actual bending energy in the element region (i. If error is measured by the norm {w − w (fem ) } w . The discretized bending moment sensed by our linear element would therefore be M. two. it turns out that this is given exactly by the formula 1 4N 2 where N is the number of elements. Consider a beam element of length 2l (Fig. An interesting pattern emerges.3. It can now be seen that this relationship can be established by arguing that this feature emerges from the fact that strains are sought in the `best-fit' manner shown in Fig.normalized tip deflection with increasing idealizations (neglecting a very small amount due to shear deformation).and four-element dealizations of a cantilever beam under tip load. 39 .

40 . Fig. Of course. We are interested now to discover how this error in strain energy translates into an error in the deflection under load P. equivalently.5) (4.6) Thus. It is left to the reader to show that the tip deflections of the continuum and discretized model will differ as {w − w (fem ) } w = 1 4N 2 . This can be very easily deduced using Castigliano's second theorem. the difference in work or energy) will converge with idealization. Our foregoing analysis shows that this follows from the fact that if any linear variation is approximated in a piecewise manner by constant values as seen in Fig. in a problem where the bending moment is constant. as a result of the discretization process involved in replacing each continuum segment of length 2l by a linear Timoshenko beam element which can give only a constant value M for the bending moment. the 2 total reduction in strain energy of the discretized model for the beam is U/4N 2 3 where U=P L /6EI is the energy of the beam under tip load. the rate of convergence will be decidedly less. Table 4. 4.4 Bending moment variation in a linear beam element. 4. this is the manner in which the square of the error in the stresses/strains (or.Energy in continuum model Energy in discretized model = = (l EI) (M 2 + V 2l 2 3) (l EI) (M 2 ) (4.3. It is simple now to show from this that for the cantilever beam of length L with a tip load P. exact) and in the case where the bending moment is varying quadratically or at a higher rate.1 shows this predicted rate of convergence. the rate of convergence will be better than this (in fact. there is a reduction 2 2 (error) in energy in each element equal to (l/EI) (V l /3).

We can see briefly below (the reader is in fact encouraged to work out the case in detail) where the cantilever beam has a uniformly distributed load acting on it using the same linear Timoshenko beam element for discretization that this is not the case. With these assumptions. this variation is piecewise linear. two levels of discretization error are introduced. This can be deduced from the fact that the discretized potential energy U is less than the actual potential energy U for this problem. We see from the foregoing analysis that using linear Timoshenko beam elements for the tip loaded cantilever. we now extend the concept of "best-fit" and variationally correct rate of convergence to the case of uniformly distributed load of intensity q on the cantilever with a little more effort. Thus. We also see that this order of convergence carries over to the estimate of the tip deflections for this problem. However. This can deceive an unwary analyst into believing that an exact solution has been reached. O(h ) for this problem. We can see that for this problem (see Fig. It is frequently believed that the finite element displacement approach always underestimates the potential energy and a displacement solution is consequently described as a lower bound solution.3) this estimate is meaningful if we consider the maximum error in strain to occur at element nodes (at centroids the errors are zero as these are optimal strain points). 4. when a finite element model is made. Thus. It turns out that tip rotations converge from above (in a 1 2N 2 rate) while the tip deflections are fortuitously exact. as far as tip deflections are concerned. the energy for this problem converges 2 as O(h ) where h=2l=L/N is the element size. which are concentrated at element nodes. Many text-books are confused over such relationships. Now.e. this is not a universally valid generalization. Inspection of the tip rotation confirms that the solution is approximate and converging. especially those that proceed on the order of error analysis. even with one element!) and a tip rotation that converges at the rate 1 2N 2 from above to the exact value θ = qL3 6EI .e.e. The next level of error is due to the approximation implied in developing the stiffness matrix which we had considered above this effectively senses a "bestapproximated" constant value of the bending moment within each element of the linear bending moment appearing to act after load discretization.We also notice that convergence in this instance is from below. i. As promised above. these errors in strain vanish as ( ) ( ) 41 . Over the entire beam model. it is a simple exercise using Castigliano's theorem and fictitious tip force and moment P and M respectively to demonstrate that the finite element model of such a problem using two-noded beam elements will yield a fortuitously correct tip deflection w = qL4 8EI for all idealizations (i. The lower bound solution nature has been disturbed because of the necessity of altering the load system at the nodes of the finite element mesh under the `consistent load' lumping procedure. the first level of discretization error is due to the replacement of the quadratically varying bending moment in the actual beam with a linear bending moment within each beam element. We also see that with element discretization. i. the uniformly distributed load is replaced by consistent loads. Firstly. These approaches arrive at conclusions such as strain error is proportional to element size. O(h) and 2 displacement error proportional to the square of the element size. the two levels of discretization errors have nicely cancelled each other to give correct answers.

This conclusion is valid only for cases where the discretized strains are `best-fit' approximations of the actual strains. then strain energies are accurate to O(h ) because strain energy expressions contain squares of the strain. the only valid conclusion that could be drawn is that the strains that have O(h) error will produce errors in strain energy that are O(2h). It is possible to improve performance beyond this limit only by resorting to some kind of extra-variational manipulation.O(h).4 The variationally correct rate of convergence It is possible now to postulate that if finite elements are developed in a clean and strictly "legal" manner. In these C elements (Timoshenko theory for beam and Mindlin theory for plate) there is provision for transverse shear strain energy to be computed in addition to the bending energy (which is the 1 only energy present in the C Euler-Bernoulli beam and Kirchhoff plate formulations). 4. without violating the basic energy or variational principles. If the `best-fit' paradigm did not apply. To understand that this is strictly an extra-variational trick (or "crime"). 4.5.5 earlier. We can also see that the strain energies are now converging at the rate 2 of O(h ) and this emerges directly from the consideration that the discretized strains are `best-fits' of the actual strain. where B and D are the strain-displacement and stress-strain matrices respectively. this prescription of strictly legal formulation is not always followed.1 The mechanics of the RBF correction. it is necessary to understand that a variationally correct procedure will yield linear elements that have a predictable and well-defined rate of convergence. Consider a case where only one element of 42 .4. The RBF correction is a deceptively simple device that enhances performance of the linear elements by modifying the shear energy term to compensate for the deficient bending energy term so that superior performance is obtained. as observed in the present example.2] is a device used to enhance the performance of 2-node beam and 4-node rectangular plate elements (these are linear elements) so that they achieve results equivalent to that 0 obtained by elements of quadratic order. This has been carried out earlier in this chapter. which often argue that the 2 strains are accurate to O(h). However.5) and (4. MacNeal [4. there is a certain rate at which solutions will converge reflecting the fidelity that the approximate solution maintains with the exact solution. This conclusion is not so readily arrived at in the order of error analysis methods. We can call this the variationally correct rate of convergence.5 Residual bending flexibility correction The residual bending flexibility (RBF) correction [4.1.6) above. It is not uncommon to encounter extra-variational devices that are brought in to enhance performance. we mean that the stiffness matrix is derived strictly using a BTDB triple product. By a variationally correct procedure.2] describes the mathematics of the RBF correction using an original cubic lateral displacement field (corresponding to a linearly varying bending moment field) and what is called a aliased linearly interpolated field (giving a constant bending moment variation for the discretized field). 4. We might recall that we examined the aliasing paradigm in Section 2. We find that a quicker and more physically insightful picture can be obtained by using Equations (4.

the discretizations fail in a dramatic fashion.g.8). due to finite element discretization must converge rapidly. The RBF is one such extra-variational device. stresses or energies. one expects errors of the order of (l L )2 . we could establish that there is a variationally correct rate of convergence for each element. It is also possible to argue that errors. Using a simple example and this simple concept. at least 2 in a O(h ) manner or better. We shall discover however that a class of problems exist where errors are much larger . w L = P L2 3EI + 1 kAG 2 (4. We have shown that the best-fit paradigm is a useful starting point for deriving estimates about errors and convergence. The RBF correction proposes to manipulate the shear flexibility in the discretized case so that it compensates for the deficiency * in the bending flexibility for this problem. 4. Applying Castigliano's theorem. errors must not be more than a few percent. The missing L2 12EI (or l 2 3EI ) is now brought in as a compensation * through the shear flexibility. Subsequent chapters will undertake a comprehensive review of the various manifestations of such errors. This "fudge factor" therefore enhances the performance of the element by giving it a rate of convergence that is not variationally permissible.9) It is not accurate to say here that the correction term is derived from the bending flexibility.8) Note the difference in bending flexibilities. This describes the inherent approximation involved in the discretization process if all variational norms are strictly followed. from Eq (4. i. Thus. with ten elements in a one-dimensional structure.1 shows how the convergence in the problem shown in Fig. A preliminary study of the issues involved will be taken up in the next chapter. the two-noded beam element used here) is L2 4EI . or do they? Table 4. It is felt that this detailed treatment is justified. If a large structure (domain) of dimension L is sub-divided into elements (sub-domains) of dimension l.2 is changed by this procedure.e. Two wrongs make a right here.6 Concluding remarks. Thus if k is the compensated shear correction factor. 4.7) L2 3EI + 1 kAG = L2 4EI + 1 k* AG 1 k* AG = 1 kAG + l 2 3EI (4. k is changed to k . This can be improved only by taking liberties with the formulation.e. we have ( w L = P (L ) 4EI + 1 kAG ) (4. we can show very easily that the continuum and discretized solutions will differ by. and this cannot be resolved by the classical (pre-1977) understanding of the finite element method. as an understanding of 43 .length 2l is used to model a cantilever beam of length L with a tip load P. i. The bending flexibility of an element that can represent only a constant bending moment (e. by introducing extra-variational steps.7) and (4. whether in displacements. the linear Timoshenko beam element serves to expose the factors clearly.

H. NY. 4. Most of the early techniques to overcome these difficulties were ad hoc. 175-183. 1994. Comput. 8. 1978. 44 .such errors has been one of the most challenging problems that the finite element procedure has faced in its history. rational scientific criteria. In the subsequent chapters. MacNeal. 4. H.7 References 4. more in the form of `art' or `black magic'. R.1. Struct. MacNeal. A simple quadrilateral plate element.. R.2. our task will be to identify the principles that establish the methodology underlying the finite element procedure using critical. Marcel Dekker. Finite Elements: Their Design and Performance.

Failure to do so causes solutions to lock to erroneous answers. due to finite element discretization must manner or better. whether in displacements. Studies in recent years have established that an aspect known as consistency must be taken into account.1 Introduction A pivotal area in finite element methodology is the design of robust and accurate elements for applications in general purpose packages (GPPs) in structural analysis. 45 . we mean that finite element solutions vanish quickly to zero (errors saturating quickly to nearly 100%!) as certain parameters (the penalty multipliers) become very large.as the element libraries of GPPs stabilized these elements came to be favored for reasons we shall discuss shortly. it is possible to argue that errors. with ten elements in a onemore than a few percent. These are what are now called the ‘locking’ 0 problems in C finite elements . The paradigm showed how elements can be designed to be free of these errors. as it was understood around 1977. The consistency paradigm requires that the interpolation functions chosen to initiate the discretization process must also ensure that any special constraints that are anticipated must be allowed for in a consistent way. stresses or energies. These same errors are responsible for the locking seen in the displacement type finite element models of similar problems. at least in a (l L )2 (domain) of dimension L is sub-divided l. In this chapter. the elements work without difficulty. The displacement type approach we have been examining so far is the technique that is overwhelmingly favored in general purpose software. This permits us to discuss the concept of consistency and the nature of errors that appear during a Ritz type approximation. where a large structure into elements (sub-domains) of dimension dimensional structure. We have seen earlier that the finite element method can be interpreted as a piecewise variation of the Rayleigh-Ritz method and therefore that it seeks strains/stresses in a `best-fit' manner. we shall first introduce the basic concepts needed to understand why such discretized descriptions fail while others succeed. should produce for such problems. We 0 compare the equations of the Timoshenko beam theory (a C theory) to the 1 classical beam theory (a C theory) to show how constraints are generated in such a model. From such an interpretation. answers that were only a fraction of a per cent of the correct answer with a practical level of discretization. It was not clear why the displacement type method. By locking. It also enabled error-analysis procedures that allowed errors to be traced to the inconsistencies in the representation to be developed. Thus. errors must not be converge rapidly. We can now develop a family of such error-free robust elements for applications in structural mechanics.Chapter 5 The locking phenomena 5. By and large. However. there were spectacular failures as well.

and this was considered to be an advantage. it turned out that to capture the physics of deformation of thick or moderately thick structures. It may be instructive to note that the general two-dimensional (i. So. as would be found in triangular or quadrilateral planforms of a plate bending element. or the behavior of plates and shells made of newly emerging materials such as high performance laminated composites.namely the EulerBernoulli beam theory. the use of the fiction of the Kirchhoff effective shear reactions.e.g. The orders of the governing equations were correspondingly reduced. reflecting the confidence structural analysts had in such theories for over two centuries . Some of the 1 contradictions that arose as a result of the old C theories . w. In the former. There were some consequences arising from such an assumption both for the mathematical modeling aspect as well as for the finite element (discretization) aspect.the three translations and the three rotations at a point. In the older C formulations. as general purpose packages ideal for production run 0 analyses and design increasingly found favour in industry. could now be removed.5. continuity of slope was also required and to achieve this in arbitrarily oriented edges. it was necessary to retain curvature degrees of freedom (w. plane stress. restoring the more physically meaningful set of three boundary conditions on the edge of a plate or shell (the Poisson boundary conditions as they are called) to be used. plate 1 and shell elements slowly began to replace the older C equivalents. the Kirchhoff-Love plate theory and the equivalent shell theories. w.thus this 0 development was welcome in that universally valid C shape functions and their derivatives could be used for a very wide range of structural applications. plates and shells. This was ideal from the point of view of the organization of a general purpose package. One single governing differential equation resulted.xx. their element libraries replaced what were called the C elements with what 0 were known as the C elements.yy) at the nodes and rely on quintic polynomials for the element shape or basis functions.xy. plane strain and axisymmetric) elements and three-dimensional (solid or brick as they 0 are called) elements were in any case based on C shape functions .2 From C1 to C0 elements As the second generation of GPPs started evolving around the late 70s and early 1 80s.these required the definition of rotations of normals which were different from the slopes of the neutral surface. the transverse deflection of a point on what is called the neutral axis (in a beam) and neutral surface of a plate or shell. w. elements needed only simple basis functions requiring only the continuity of the fields across element boundaries . The former were based on well known classical theories of beams. A salutary effect that carried over to finite element modeling was that the elements could be designed to have nodal degrees of freedom which were the six basic engineering degrees of freedom . it was necessary to turn to more general theories accounting or transverse shear deformation as well . although of a higher order (in comparison to other theories we shall discuss shortly). 46 . the C beam. These theories did not allow for transverse shear strain and permitted the modeling of such structures by defining deformation in terms of a single field.these 0 1 are called the C requirements. Also. The strains could then be computed quite simply from the assumption that normal to the neutral surface remained normal after deformation.e.

produces poorer convergence than would otherwise be expected of the higher order of approximation involved. Π = 2 ò (1 2 EI w.1 The classical beam theory Consider the transverse deflection of a thin cantilever beam of length L under an uniformly distributed transverse load of intensity q per unit length of the beam. The consistency approach argues that it is necessary in such problems to discretize the penalty-linked strain fields in a consistent way so that only physically meaningful constraints appear.1]. show how the Ritz approximate solution is very sensitive to the way in which its terms are chosen. rank and singularity of penalty-linked stiffness matrix.a simple example will illustrate this. and consistency of strain-field When locking was first encountered. illustrate the concepts involved using a simple Ritz-type variational method of approximation of the beam problem via both classical and Timoshenko beam theory [5. we can state this problem in a weak form with a quadratic functional given by.e. on reflection. efforts were made to associate it with the rank or non-singularity of the stiffness matrix linked to the penalty term (e. It is therefore necessary to trace this to the origin. This should produce a linear shear force distribution increasing from 0 at the free end to qL at the fixed end and correspondingly. a bending moment that varies from 0 to qL2 2 . see the discussion below). 5. By making it `consistent'. The high rank and non-singularity is the outcome of certain assumptions made (or not made. It is again demonstrated that a Ritz approximation that ensures ab initio consistent definition will produce the expected rate of convergence .xx − qw ) dx 0 L (5.However. it is obvious that these are symptoms of the problem and not the cause.surprisingly dramatic failures came to be noticed and the greater part of academic activity in the late seventies.1) This theory presupposes a constraint condition. leaving certain unanticipated requirements unsatisfied) during the discretization process. most of the eighties and even in the nineties was spent in understanding and eliminating what were called the locking problems.3. `inconsistency' does not lead to locked solutions but instead. However. An `inconsistent' choice of parameters in a low order approximation leads to a full-rank (non-singular) penalty stiffness matrix that causes the approximate solution to lock. locking can be eliminated. assuming zero transverse shear strain. In higher order approximations.g.3 Locking. we would not enter into a formal finite element discretization (which would be taken up in the next section) but instead. 5. i. It is possible to show how the Timoshenko beam theory can be reduced to the classical thin beam theory by using a penalty function interpretation and in doing so. the shear stiffness matrix in a Timoshenko beam element which becomes very large as the beam becomes very thin. In this section. life was not very simple . and this allows the deformation of a beam to be described entirely 47 . Using what is called the classical or Euler-Bernoulli theory.

1 A two-term Ritz approximation Let us 2 now 3 consider an approximate Ritz solution based on two terms.xxx = − q (L − x ) ( ) (5.3c) If the expressions in Equations (5.xx = 5qL2 12 − (qL 2 ) x EI w .3b) (5.5) ( ) (4 3 − 2ξ ) The approximate solution for the bending moment is seen to be a `best-fit' or `least-squares fit' of the exact solution. Note that the constant and linear terms are dropped.3a) so that the approximate bending moment and shear force determined in this Ritz process are. we see that the shear force predicted by the Ritz approximation is a 'best-fit' of the actual shear force variation. with the points ξ = ± 1 3 . w (x ) = qL2 4EI x 2 − (qL 6EI ) x 3 − (q 24EI ) x 4 EI w.2) and (5. 48 .3c). w = b2 x + b3 x .1.xx = qL2 8 4 3 − 2ξ − 1 3 (1 − 3ξ 2 ) EI w .xxx = − qL 2 ( ) (5. where x = (1 + ξ ) L 2 so that the ends of the beam are represented by ξ = -1 and +1.2b) (5.3) are written in terms of the natural co-ordinate system ξ. EI w. w (x ) = 5qL2 24EI x 2 − (qL 12EI ) x 3 ( ) (5. anticipating the boundary conditions at the fixed end. An application of the principle of minimum total potential allows the governing differential equations and boundary conditions to be recovered. From Equations (5.4) (5.xx = (q 2 ) (L − x )2 EI w.in terms of a single dependent variable.2a) (5.2c) 5. One can easily work out that the approximate solution will emerge as. EI w .3. which are the points where the second order Legendre polynomial vanishes. Once again we confirm that the Ritz method seeks the 'best-approximation' of the actual state of stress in the region being studied.xx = qL2 8 ( )( ) (5. but this will not be entered into here.2c) and (5. the exact and approximate solutions can be expanded as. A simple exercise will establish that the exact solution satisfying the governing differential equations and boundary conditions is. the transverse deflection w of points on the neutral axis of the beam. emerging as points where the approximate solution coincides with the exact solution.

the shear strain γ=θ-w. the difference between the face rotation and the slope of the neutral axis. The shear strain is now given by γ = a1x − b1 (5.3. the transverse deflection w and the face rotation θ so that the shear strain at each point is given by γ = θ − w.x .6) where the curvature κ=θ.x + 1 2 α (θ − w.x→0 must emerge in the limit.x. and it will now account for the bending (flexural) energy and an energy of shear deformation.2.5. For a very large α. The factor k accounts for the averaged correction made for the shear strain distribution through the thickness.1] offers a physically more general formulation of beam flexure by taking into account the transverse shear deformation. we get (5. This requires that the Kirchhoff constraint θ-w. for a beam of rectangular cross-section.8) a1 = − 3qL2 12EI + αL2 .7) and it would seem that this can represent a linearly varying shear force. α→∞. b1 = − qL 1. The description of beam behavior is improved by introducing two quantities at each point on the neutral axis.e. 5. i. The Timoshenko beam theory will asymptotically recover the elementary beam theory as the beam becomes very thin.1 A two-term Ritz approximation Consider now a two term Ritz approximation based on θ = a1x and w = b1x . This permits a constant bending strain (moment) approximation to be made. these equations lead directly to the simple fourth order differential equation for w of elementary beam theory. or as the shear rigidity becomes very large.9) 49 . Π = L 2 2 ò (1 2 EI θ . ì é L3 3 − L2 2 ù ü ìa1 ü ï 0 ï ì ü éL 0 ù ï ï ú ý í ý=í 2 ý + αê íEI ê ú 2 ê− L 2 L ú ï î b1 þ ïqL 2 ï ë0 0 û ï î þ ë û þ î Solving. For example. this is usually taken as 5/6.2 The Timoshenko beam theory and its penalty function form The Timoshenko beam theory [5.3. this is secured very easily in the infinitesimal theory but it is this very same point that poses difficulties when a simple Ritz type approximation is made. The Ritz variational procedure now leads to the following set of equations. The total strain energy functional is now constructed from the two independent functions for w(x) and θ(x).x and α=kGA is the shear rigidity.x ) − q w ) dx 0 (5.5qL3 − 2α 12EI + αL2 (5. Thus.

x→0 as well this spurious constraint produces the locking action on the solution. This could have been anticipated from a careful examination of Equations (5. The Ritz variational procedure leads to the following equation structure: ì é L3 3 − L2 2 éL 0 0 ù ï ê ï ú ê 2 L í EI ê0 0 0 ú + α ê − L 2 ê ï 3 ê0 0 0 ú ê − 2L 3 L2 û ë ï ë î − 2L3 3 ù ú L2 ú ú 4L3 3 ú û ü ï ï ý ï ï þ ì 0 ü ìa1 ü ï ï ï 2 ï ï ï í b1 ý = íqL 2 ý ï 3 ï ïb ï ïqL 3 ï î 2þ î þ (5. We shall see how this is done next.2 A three-term consistent Ritz approximation Consider now a three term approximation chosen to satisfy what we shall call the consistency condition.12b) (5.the consistently represented field in Equation (5.2. a meaningful approximation can be made for a penalty function formulation only if the penalty linked approximation field is consistently modeled. the conditions emerging are a1→0 and b1→0. The shear force is now correctly captured as a linear variation . only a constant bending strain is permitted. a1→0 imposes a zero bending strain θ. θ =− qL2 x 6EI (5. Clearly. γ = − b1 + (a1 − 2b2 )x (5.x = q (L − x ) ( ) There is no locking seen at all . This is a clear case of a solution `locking'.12d) α θ − w .12a) w = qL q 2 qL2 x− x + x2 α 2α 12EI − EI θ .the bending moment is now a least squares correct constant approximation of the exact quadratic variation (this can be seen by comparing Equation (5.6) introduces a penalty condition on the shear strain and this requires that the shear strain must vanish in the Ritz approximation process . The penalty limit α→∞ in Equation (5. the constraint α1-2b2→0 is consistently balanced and will not result in a spurious constraint on the bending field.12c) with Equations (5.As α→∞.7).10) Note that as α→∞.4) and (5.11) It can be easily worked out from this that the approximate solution is given by.6) and (5. Thus.x = qL2 6 (5. Choose θ = a1x and w = b1x + b2 x 2 .5) earlier). But we now have shear strain of linear order as.12c) (5. both a1 and b1→0.12) being able to recover this even as a α→∞! 50 . 5.from Equation (5.7).3. Again.

5.e.11) shows that while in the former.3 A four-term inconsistent Ritz approximation We now take up a four term approximation which provides theoretically for a linear variation in the approximation for bending strain.as a direct consequence.8) and (5. It is clear also that the singularity (or reduced rank) is a result of the consistent condition represented by (α1-2b2) in the linear part of the shear strain definition in Equation (5.A comparison of the penalty linked matrices in Equations (5. The Ritz variational procedure leads to the following set of equations: ì éL ï ê ï ê 2 L ï EI ê í ê ï ê0 ï ê ï ê0 ë î L2 4L3 3 0 0 0 0 0 0 0ù ú ú 0ú + ú 0ú ú 0ú û (5.13). We shall now see what effect this spurious constraint has on the approximation process.3. The rank of the 51 . we have a non-singular matrix of rank 2.10) . in the latter we have a singular matrix of rank 2 for a matrix of order 3.x = a1 + 2a2 x and the shear strain is γ = − b1 + (a1 − 2b2 )x + a2 x 2 (5. We can further show that non-singularity of penalty linked matrix arises in an inconsistent formulation only when the order of approximation is low.this arises from the consistent representation (a1 − 2b2 ) of the linear part of shear strain in Equation (5. as seen for the two-term Ritz approximation. We can go for a quadratic inconsistent approximation (with linear bending strain variation) in the next section to show that there is no locking' of the solution and that the penalty linked matrix is not non-singular .2. the third row of the penalty linked matrix in Equation (5.14) é L3 3 ê ê L4 4 α ê ê − L2 2 ê 3 ê− 2 L 3 ë L4 4 L5 5 − L3 3 − L4 2 − L2 2 − L3 3 L L2 − 2L3 3 ù ú − L4 2 ú ú L2 ú ú 4L3 3 ú û ü ï ï ï ý ï ï ï þ ì 0 ü ìa1 ü ï ï ï ï ïa2 ï ï 0 ï í ý = í 2 ý ï b1 ï ïqL 2 ï ï b2 ï ï 3 ï î þ îqL 3 þ It can be seen that the penalty linked 4×4 matrix is singular as the fourth row is exactly twice the second row . It is this aspect that led to a lot of speculation on the role the singularity or rank of the matrix plays in such problems.however the effect of inconsistency is to reduce the performance of the approximation to a sub-optimal level.11) is exactly twice the first row. i. θ = a1x + a2 x 2 and w = b1x + b2 x 2 so that κ = θ .13) Note now that the condition α→∞ forces a2→0 this becomes a spurious constraint on the bending strain field.

52 . This can be achieved by noting that in Equation (5. Our experience with consistent finite element formulations 2 ) [5. the bending moment and shear force variation are given by. − EI θ .16b) The solution can sense only a constant bending moment in the thin beam limit.x = qL2 6 (5. ( ) can be shown to vanish at the points ( 3 ) and 5. or ξ = ± 1 3 . We for the shear strain defined as also see from Equation (5. one can work out from Equation (5.matrix is therefore 3 and the solution should be free of "locking" .e.5qL (6 (x L )2 − 6 (x L ) + 1 ( ) ) x L = 1 2 1+1 (5. a1 = − qL2 15qL2 − 6EI 60EI + αL2 15qL2 60EI + αL2 (5. There are now violent quadratic oscillations in the shear force and these oscillations 1 2 1 − 1 3 . What it means is that this four term inconsistent approach will produce answers only as efficiently as the three term consistent Ritz formulation.15d) As α→∞.15c) b2 = − qL2 q + 6EI 2α (5.2] allows us to replace the x term in Equation (5. The effect of the inconsistent representation has been to reduce the effectiveness of the approximation. i.13) with Lx − L2 6 so that. We shall next see how the effectiveness can be improved by making the approximation consistent before the variational process is carried out.13).2. the offending term is the quadratic term associated with a2.however the inconsistent constraint forces a2 → 0 and this means that the computed bending strain κ → a1 .16a) α θ − w .x = q (L − x ) + 2.15a) a2 = (5.16b) that this leads to a spuriously excited quadratic form (6x 2 − 6Lx + L2 . Indeed. which are the points corresponding to the 2 point Gauss-Legendre integration rule. it will have only a constant bending moment prediction capability.14) that.15b) b1 = − qL 5qL3 − α 2 60EI + αL2 ( ) (5.3.4 A four-term consistent Ritz approximation Let us now take up a Ritz approximation with a consistently represented function γ .

12b) so that the shear force is now being exactly computed.2b2 )x In fact.17) ( )+ (a1 + a2 L . We also see that Equation (5.b1 + a2 L2 6 and (a1 + a2 L . ì éL ï ê ï ê 2 ï êL í EI ê ï ê0 ï ê ï ê0 ë î L2 4L3 3 0 0 0 0 0 0 0ù ú ú 0ú + ú 0ú ú 0ú û (5. In other words.19a) with Equation (5.18) é L3 3 ê ê L4 4 α ê ê − L2 2 ê 3 ê− 2 L 3 ë L4 4 7L5 36 − L3 3 − L4 2 − L2 2 − L3 3 L L2 − 2L3 3 ù ú − L4 2 ú ú L2 ú ú 4L3 3 ú û ü ï ï ï ý ï ï ï þ ì 0 ü ìa1 ü ï ï ï ï ï 0 ï ïa2 ï í ý = í 2 ý ïqL 2 ï ï b1 ï ï 3 ï ï b2 ï î þ îqL 3 þ It is important to recognize now that since the penalty linked matrix emerges from the terms .3b) we see that the approximate solution to the Timoshenko equation for this problem with a consistent shear strain assumption gives exactly the same bending moment as the Ritz solution to the classical beam equation could provide. b1 and b1 all yield physically meaningful conditions representing the state of equilibrium correctly.2b2 ) there would only be two linearly independent rows and therefore the rank of the matrix is now 2. ò T δ γ æ γ − γ ö dx = 0 ç ÷ è ø and this will yield precisely the form represented in Equation (5.17). ( ) − EI κ = 5 qL qL2 − x 12 2 (5.19b) is identical to Equation (5.γ = − b1 + (a1 − 2b2 )x + a2 Lx − L2 6 = − b1 − a2 L2 6 ( ) (5. as α→∞.19b) α γ = q (L . 53 .x ) Comparing Equation (5.19a) (5. a2. it can be proved using the generalized (mixed) variational theorem such as the Hellinger-Reissner and the Hu-Washizu principles [5. The Ritz variational procedure now leads to. The approximate solution is then given by.2] that the variationally correct way to determine γ from the inconsistent γ is. the terms a1.

of the strain-fields that are constrained in the penalty regime. poor convergence and 0 displacement type finite element violent stress oscillations seen in C formulations are due to a lack of consistent definition of the critical strain fields when the discretization is made . In some limiting cases of physical behavior. It is easy to predict all this by examining the constrained strain-field terms from the consistency point of view rather than performing a post-mortem examination of the penalty-linked stiffness matrix from rank or singularity considerations as is mostly advocated in the literature.i. the shape functions chosen for the displacement fields cannot do this in a meaningful manner . We shall now attempt a very preliminary heuristic definition of these requirements as seen from the point of view of the developer of a finite element for an application in a constrained media problem. the shear strain in a Timoshenko theory will have θ and the first derivative of w). The consistency condition demands that the discretized strain field interpolations must be so constituted that it will enforce only physically true constraints when the discretized functionals for the strain energy of a finite element are constrained.4 Consistency and C0 displacement type finite elements We shall see in the chapters to follow that locking. these strain fields can be constrained to be zero values. From these definitions.) have. the coefficients of the strain field interpolations may have constants from all the contributing field variable interpolations or from only one or some of these. the vanishing shear strain in a thin Timoshenko beam. The number of constants used will depend on the number of nodal variables and any additional nodeless variables (those corresponding to bubble functions). quadratic. the results are in very great error and that these errors grow without limit as the beam becomes very thin. linear.g. one can compute the strain fields using the strain-displacement relations. These are obtained as interpolations associated with the constants that were introduced in the field variable interpolations. We saw in our Ritz approximation of the Timoshenko beam theory in this section that as the beam becomes physically thin. It is also seen that in higher order approximations. We can elaborate on this definition in the following descriptive way. constant. coefficients from all the independent interpolations of the field variables that appear in the definition of that 54 .spurious constraints are generated which cause locking. Where the strain-field is such that all the terms in it (i. etc. This is so even when the shape functions for the w and θ have been chosen to satisfy the completeness and continuity conditions. but is damaging as to produce poor convergence and stress oscillations. The foregoing analysis showed how the lack of consistency translates into a non-singular matrix of full rank that causes locking in low-order Ritz approximations of such problems. e. Depending on the order of the derivatives of each field variable appearing in the definition of that strain field (e. In the development of a finite element. the field variables are interpolated using interpolations of a certain order. In the finite element statement of this problem.e.g. the shear strains must vanish and it must begin to enforce the Kirchhoff constraint and that this is not possible unless the approximate field can correctly anticipate this.e. the situation is not as dramatic as to be described as locking.5. associated with it. We shall see later that if a simple finite element model of the Timoshenko beam is made).

The same conditions are true for the various finite element formulations where locking. vanishing membrane strains in curved beams and shells etc.causing what is often described as locking. In Chapter 8 we proceed to a few problems where consistency between strain and stress functions are needed even where no constraints on the 55 . to penalty multipliers. The constraints thus enforced are true constraints. poor convergence and stress oscillations are known to appear. Again.we shall investigate this closely. For simple low order elements. This example serves as the simplest case that could be employed to demonstrate all the principles involved in the consistency paradigm. vanishing shear strains in thin Timoshenko beams or Mindlin plates. It sets the scene for the need for new paradigms (consistency and correctness) to complement the existing ones (completeness and continuity) so that the displacement model can be scientifically founded. the constraints may incorrectly constrain some of these terms. these constraints are severe enough to produce solutions that rapidly vanish . solutions are obtained which are sub-optimal to solutions that are possible if the formulation is carried out with the consistency condition imposed a priori. This is followed up with studies of parasitic shear and incompressible locking. The correctness concept is then brought in to ensure that the devices used to achieve a consistent strain interpolation are also variationally correct. The membrane locking phenomenon is investigated in Chapter 7. We shall call such a representation field-consistent. which are linked. We shall see later that the use of devices such as reduced integration permits the consistency requirement to be introduced when the penalty linked matrix is computed so that the correct rank which ensures the imposition of the true constraints only is maintained. e. It is also clear why the imposition of the consistency condition into the formulation allows the correct rank or singularity of the penalty linked stiffness matrix to be maintained so that the system is free of locking or suboptimal convergence. This fieldinconsistent formulation is said to enforce additional spurious constraints. The phenomenon of shear locking is the most well known . The application of these concepts to Mindlin plate elements is also reviewed.strain-field. In Chapters 5 to 7 we investigate the effect of constraints on strains due to the physical regimes considered.5 Concluding remarks These exercises show us why it is important to maintain consistency of the basis functions chosen for terms in a functional. The simple curved beams are used to introduce this topic. In the next chapter. In higher order inconsistent formulations. Where the strain-field has coefficients in which the contributions from some of the field variables are absent. 5. we shall go to the locking and other over-stiffening phenomena found commonly in displacement finite elements. The Timoshenko beam element allows the problem to be exposed and permits a mathematically rigorous error model to be devised. as for the four-term inconsistent Ritz approximation. the constraint that appears in the limit can be correctly enforced. The consistency paradigm is introduced to show why it is essential to remove shear locking. it is worthwhile to observe that non-singularity of the penalty linked matrix occurs only when the approximate fields are of very low order as for the two-term Ritz approximation.g.

Method in Structural Mechanics. On the transverse vibrations of bars of uniform crosssection. Prathap. 1993.1 S. 5. correct and consistent variational basis for the displacement type finite element method.2 G. Philosophical Magazine 443. The Finite Element Academic Press.6 References 5. 125-131. 1922. These examples show the universal need of the consistency aspect and also the power of the general Hu-Washizu theorem to establish the complete.strains appear. Timoshenko. 5. Dordrecht. Kluwer 56 . P.

Our conclusions from the preceding chapter would therefore apply to this as well. However.2a) and (6. the element was beset with problems. (a) 1 w w.2b). T T ò 1 2 EI χ χ + 1 2 kGA γ γ dx ( ) (6. It was therefore very attractive for general purpose applications.x 2 w θ Fig.1.1 The conventional formulation of the linear beam element The strain energy of a Timoshenko beam element of length 2l can be written as the sum of its bending and shear components as. I and A are the moment of inertia and the area of cross-section.Chapter 6 Shear locking In the preceding chapter.1) (6.1 The linear Timoshenko beam element An element based on elementary theory needs two nodes with 2 degrees of freedom at each node. as we shall presently see. E and G are the Young's and shear moduli and the shear correction factor used in Timoshenko's theory.x γ = θ − w.x (b) 1 w θ 2 w w. 6.1 (a) Classical thin beam and (b) Timoshenko beam elements. the transverse deflection w and slope dw/dx and uses cubic 1 interpolation functions to meet the C continuity requirements of this theory (Fig.2b) where χ = θ . 6. We shall now take up the finite element representation of this problem.1).2a) (6. w is the transverse displacement and θ the section rotation. which is essentially a piecewise Ritz approximation. 57 . respectively. we saw a Ritz approximation of the Timoshenko beam problem and noted that it was necessary to ensure a certain consistent relationship between the trial functions to obtain accurate results. 6.x In Equations (6. A similar two-noded beam element based on the shear flexible 0 Timoshenko beam theory will need only C continuity and can be based on simple linear interpolations. 6.

1.In the conventional procedure. and the element stiffness matrix is calculated in an analytically or numerically exact (a 2 point Gauss Legendre integration rule) way.1). of elements 1 2 4 8 16 “Thin” beam -5 0.Normalized tip deflections No. t=1. as. exact answers would have been obtained with one element for this case.1 to see the trend as the number of elements is increased.1 shows that the normalized tip displacements are dramatically in error. We can compute the bending and shear strains directly from these interpolations using the strain gradient operators given in Equations (6. The tip deflections obtained. using a fictitiously large value of G to simulate the "thin" beam condition.320 × 10 -3 0. a bending related part and a shear related part.1 . Table 6. These are then introduced into the strain energy computation in Equation (6. are directly related to the square of the number of elements used for the idealization.3a) (6. This is clearly unrelated to the physics of the Timoshenko beam and also not the usual sort of discretization errors encountered in the finite element method.200 × 10 -5 0.h 2 0 0 0ú ê ú ê 2 −1 0 1ú û êh 2 h 6 ë 0 0 h 2ù ú . displacement field variables as. for a length h the stiffness matrix can be split into two parts. which are several orders of magnitude lower than the correct answer. We can carefully examine Table 6. the discretization process has introduced an error so large that the 2 resulting answer has a stiffness related to the inverse of N . It is this very phenomenon that is known as shear locking. In fact with a classical beam element model.128 × 10 -3 0. Table 6.2b).3b) where the dimensionless coordinate ξ=x/l varies from -1 to +1 for an element of length 2l. G=37500000. For the beam element shown in Fig. 6.800 × 10 -4 0.2a) and (6. considering the case of a "thin" beam with E=1000.h 2 h2 6 ú ú 1 -h 2 ú ú ú 2 -h 2 h 3 ú û −1 We shall now model a cantilever beam under a tip load using this element. This ensures that the element is capable of strain free rigid body motion and can recover a constant state of strain (completeness requirement) and that the displacements are continuous within the element and across the element boundaries (continuity requirement).512 × 10 58 . L=4. linear interpolations are chosen for the N 1 = (1 − ξ ) 2 N 2 = (1 + ξ ) 2 (6. kb é0 ê ê0 = EI h ê ê0 ê ê0 ë h2 é 1 0ù ê ú êh 2 h 2 3 1 0 − 1ú ú ks = kGt h ê ê -1 . In other words.

An experiment with a one-point integration of the shear strain energy component causes the shear related stiffness matrix to change as shown below.1.x at x=0.x at x=0. and therefore when a beam of overall length L is divided into N elements of equal length h.1 for a typical convergence trend with this element).1) is integrated with a one-point Gaussian rule as this is the minimum order of integration required for exact evaluation of this strain energy.4b) θ = b0 + b1 We can relate such constants to the field-variables obtaining in this element in a discretized sense. we now have to look for a mechanism that can explain how this 2 spurious stiffness of (h/t) can be accounted for by considering the mathematics of the discretization process. The performance of this element was extremely good. The bending component of the strain energy of a Timoshenko beam element of length 2l shown in Equation (6. thus. We shall propose a requirement for a consistent interpolation of the constrained strain fields as the necessary paradigm to make our understanding of the phenomena complete. b0=θ and b1/l=θ. The magic formula proposed to overcome this locking is the reduced integration method. kb é0 ê ê0 = EI h ê ê0 ê ê0 ë 0 0 0 ù ú 1 0 − 1ú ú 0 0 0 ú ú − 1 0 1ú û h 2 h 2ù −1 é1 ê ú êh 2 h 2 4 − h 2 h 2 4 ú ú ks = kGt h ê ê− 1 − h 2 1 −h 2 ú ê ú ê ú 2 2 êh 2 h 4 − h 2 h 4 ú ë û 6. This denotation would become useful when we try to explain how the discretization process can alter the infinitesimal description of the problem if the strain fields are not consistently defined. numerical experiments showed that the locking stiffness progresses without limit as the element depth t decreases. showing no signs of locking at all (see Table 4. Thus. However. It is this rule that resulted in the shear stiffness matrix of rank two that locked. 59 . the additional stiffening introduced in each element due to shear 2 locking is seen to be proportional to h . If we start with linear trial functions for w and θ.4a) (6. In fact.2 The field-consistency paradigm It is clear from the formulation of the linear Timoshenko beam element using exact integration (we shall call it the field-inconsistent element) that ensuring the completeness and continuity conditions are not enough in some problems. as we had done in Equation 6.The error in each element must be related to the element length.3 above. a mathematically exact evaluation of the shear strain energy will demand a two-point Gaussian integration rule. a1/l=w. we can associate two generalized displacement constants with each of the interpolations in the following manner w = a0 + a1 (x l ) (x l ) (6.

i. this is equal to the condition (θ-w. A constraint imposed on this will lead to an undesired restriction on θ. this is equal to the condition θ. 6. This is the `spurious constraint' that leads to shear locking and violent disturbances in the shear force prediction over the element. An examination of the conditions produced by these requirements shows that the following constraints would emerge in such a limit b0 − a1 l → 0 b1 → 0 (6.7a) is field-consistent as it contains constants from both the contributing displacement interpolations relevant to the description of the shear strain field. These constraints can then accommodate the true Kirchhoff constraints in a physically meaningful way. We must now devise an operational procedure that will trace the errors due to an inconsistent representation of the constrained strain field and obtain precise a priori measures for these.6b) must vanish. We must then show by actual numerical experiments with the original elements that the errors are as projected by these a priori error models.3 An error model for the field-consistency paradigm We must now determine that this field-consistency paradigm leads us to an accurate error prediction. in an infinitesimal sense.4.x)→0 at the element centroid.7b) In the new terminology that we had cursorily introduced in Section 5. a procedure we shall call the functional reconstitution technique makes it possible to do this verification. constraint (6. The consistency requirement has been offered as the missing paradigm for the errorfree formulation of the constrained media problems. 60 . the shear strain energy in Equation (6.6b) It is possible to see from this that in the constraining physical limit of a very thin beam modeled by elements of length 2l and depth t.e. we get χ = (b1 l ) γ = (b0 − a1 l ) + b1 (x l ) (6.If the strain-fields are now derived from the displacement fields given in Equation (6.6a) (6. Only such an exercise will complete the scientific validation of the consistency paradigm. constraint (6.7b) contains only a term from the section rotation θ.5b) An exact evaluation of the strain energies for an element of length h=2l will now yield the bending and shear strain energy as Us {(b1 l )}2 2 = 1 2 (kGA ) (2l ) {(b0 − a1 l )2 + 1 3 b1 } U B = 1 2 (EI ) (2l ) (6. We know that the discretized finite element model will contain an error which can be recognized when digital computations made with these elements are compared with analytical solutions where available. no bending is allowed to develop in the element region). as we shall see presently. In direct contrast.x→0 at the element centroid (i.1.5a) (6. Fortunately.4). In an infinitesimal sense.7a) (6.e.

we can postulate that a beam of length L modeled by equal elements of length 2l will have a re-constituted functional Π = 2 2 2 ò {1 2 (EI + kGAl 3 ) (θ. we note that in each element. its flexural rigidity) by the factor kGAl2 3EI .6b) can be traced to the constants in Equations (6. It indicates that an exactly-integrated or field-inconsistent finite element model tends to behave as a shear flexible beam with a much stiffened flexural rigidity I’.8) and (6. This can be related to the original rigidity I of the system by comparing the expressions in Equations (6. I ′ I = 1 + kGAL2 3EI (6.6).12) 61 .7b).x ) }dx 0 L (6.6a) and (6.e.x ) + 1 2 (kGA ) (θ − w.x and w. which were introduced due to the discretization process.x )2 dx x } (6. θ.6. and produces the additional stiffening effect described as shear locking.4 Functional re-constitution We have postulated that the error of shear constraint in Equation (6.x at the centroid of the element. distinct from the direct finite element procedure that yields the stiffness matrix) allows us to critically examine the consistency paradigm.4a) and (6. Thus. the discretization process produces energy for the element of the form given in Equation (6. the infinitesimal) description. Thus.x )2 (6.1. Thus. From this reconstituted functional. the constants in Equations (6.x )2 (6.11) We now understand that the discretized beam is stiffer in bending (i.8) If an element of length 2l is isolated. the constants. For a thin beam.9) Thus the constants in the discretized strain energy functional have been reconstituted into an equivalent continuum or infinitesimal form. 6. π e = 1 2 (EI ) (2l ) (θ . case where the inconsistent element depth t. π e = 1 2 EI + kGAl2 3 ( ) (θ.x )2 + 1 6 (kGAl )2 (θ .1.4b) and can be replaced by the values of the field variations θ. the strain energy of deformation in an element is. The strain energy for such a shear locking originates from the spurious We must now devise an error model for the is used to model a beam of length L and beam can be set up as. we can argue that an idealization of a beam region of length 2l into a linear displacement type finite element would produce a modified strain energy density within that region of.5 Numerical experiments to verify error prediction Our functional re-constitution procedure (note that this is an auxiliary procedure.10) This strain energy density indicates that the original physical system has been altered due to the presence of the inconsistent term in the shear strain field. In this equation.x )2 + 1 2 (kGA ) (2l ) (θ − w. this can be very large.e.2 + 1 2 kGA (θ − w. Π= L ò {1 2 EI 0 θ . can be replaced by the continuum (i.11) as. x )2 + 1 2 (kGA ) (θ − w.

128 × 10 -3 0.200 × 10 -4 0. It will be instructive to note here that conventional error analysis norms in the finite element method are based on the percentage error or equivalent in some computed value as compared to the theoretically predicted value. This has shown us that the consistency paradigm can be scientifically verified.13) and (6.2 . which has been established entirely a priori.14) We can now re-interpret the results shown in Table 6.800 × 10 -3 0. N 1 2 4 8 16 Computed 0. efem = W Wfem − 1 (6.1 for the thin beam case. Since e>>1. Table 6.512 × 10 62 . We have seen now that the error of shear locking can be exaggerated without limit.We must now show through a numerical experiment that this estimate for the error. we already have an a priori prediction for this factor as. Exact solutions are available for the static deflection W of a Timoshenko cantilever beam of length L and depth t under a vertical tip load. anticipates very accurately.512 × (fem) -4 10 -4 10 -3 10 -3 10 -3 10 Predicted -4 0. Using Equations (6. starting from the consistency paradigm and introducing the functional re-constitution technique.Normalized tip deflections for the thin beam (Case 2) computed from fem model and predicted from error model (Equation (6. If W fem is the result from a numerical experiment involving a finite element digital computation using elements of length 2l.800 × 0.320 × 0. we have W fem W = N 2 5 × 10 −5 ( ) (6.2 shows how the predictions made thus compare with the results obtained from an actual finite element computation using the fieldinconsistent element.12).15) for the thin beam.13) From Equation (6.200 × 0. or computing the rank or condition number of the stiffness matrices could offer only a heuristic picture of how and why locking sets in.15)). e = I ′ I − 1 = kGAl2 3EI (6.14). as the structural parameter that acts as a penalty multiplier becomes indefinitely Table 6. Traditional procedures such as counting constraint indices. the behavior of a field-inconsistent linearly interpolated shear flexible element in an actual digital computation. we can argue a priori that the inconsistent element will produce normalized tip deflections (W fem W ) = 1 (1 + e ) .320 × 10 -3 0. the additional stiffening can be described by a parameter as.128 × 0.

the fact that the spurious constraints give rise to a spurious energy term and consequently alters the rigidity of the system being modeled. A new error norm called the additional stiffening parameter. I’. This was found to be true of the many constrained media problems. Essentially. this takes into account. curved beams etc. we would expect the relationship described by Equation 6. 2 2 for a 63 .large. 6. if w is the deflection of a reference point as predicted by an analytical solution to the theoretical description of the problem and wfem is the fem deflection predicted by a field inconsistent finite element model.2 Variation of error norms e. of the field consistent system and the rigidity. Mindlin plates. e can be introduced to recognize the manner in which the errors of locking kind can be blown out of proportion by a large variation in the structural parameter. were related to the structural parameters in the form. Thus.14. E with structural parameter kGL /Et cantilever beam under tip shear force. By way of illustration of the distinction made Fig. A logarithmic plot of the new error norm against the parameter (l/t) will show a quadratic relationship that will continue indefinitely as (l/t) is increased. I. I’/I = α(l/t) where l is an element dimension and t is the element thickness. The percentage error norms therefore saturate quickly to a value approaching 100% and do not sensibly reflect the relationship between error and the structural parameter even on a logarithmic plot.g. of the inconsistent 2 system.) it was seen that the rigidity. In many other examples (e.

It was not understood for a very long time that in many cases. In each case.by this definition. very accurately.3 shows the shear force oscillations in a typical problem .1. and plot both E and the new error norm e from the results for the same problem using 4 FI 2 elements against the penalty multiplier (l/t) on a logarithmic scale.5b). stress oscillations originated from the inconsistent constraints. Over the length of the element. the results above. we shall anticipate again. If we represent the conventional error norm in the form E = (W − W fem ) W . We can arrive at a prediction for these oscillations by applying the functional re-constitution technique. the dependence is as shown in Fig. On the other hand.16b) We see that V has a linear term that relates directly to the constant that appeared in the spurious constraint. V=0 is reproduced. this spurious energy is so large as to completely dominate the behavior of the beam and cause a locking effect. 64 .6 Shear Force Oscillations A feature of inconsistently modeled constrained media problems is the presence of spurious violently oscillating strains and stresses. in fact it is a measure of the bending moment at the centroid of the element. is the prediction made by the functional re-constitution exercise. it contributes a finite energy in Equation (6. Figure 6. For a cantilever beam under constant bending moment modeled using linear Timoshenko beam elements.2.a straight cantilever beam with a concentrated moment at the tip. the oscillations are seen to be linear functions corresponding to the kGA b1 (x/l) term. 6. We shall explore this now. One to ten equal length field-inconsistent elements were used and shear forces were computed at the nodes of each element.e. only the variation within the element at the fixed end is shown.5b). e captures this relationship. we can write from Equation (6. The oscillation is self-equilibrating and does not contribute to the force equilibrium over the element. as the pattern repeats itself in a saw-tooth manner over all other elements. 6.16a) (6. Thus.17) that b1 will not be zero. At element mid-nodes.9) and in the modeling of very slender beams. the correct shear force i. Also indicated by the solid lines. this constant will activate a violent linear shear force variation when the shear forces are evaluated directly from the shear strain field given in Equation (6. the shear force (stresses) displays a saw-tooth pattern (we shall see later that a plane stress model using 4-node elements will also give an identical pattern on the neutral bending surface). It can be seen that E saturates quickly to a value approaching 100% and cannot show meaningfully how the error propagates as the penalty multiplier increases indefinitely. We shall see below from Equation (6. If V is the shear force predicted by a field-consistent shear strain field (we shall see soon how the field-consistent element can be designed) and V the shear force obtained from the original shear strain field. However. V = kGA (b0 − a1 l ) V = V + kGA b1 (x l ) (6.7b). Equation (6. in a field-inconsistent formulation.

Fig. 6.3 Shear force oscillations in element nearest the root, for N element models of a cantilever of length L = 60.

Consider a straight cantilever beam with a tip shear force Q at the free end. This should produce a linearly varying bending moment M and a constant shear force Q in the beam. An element of length 2l at any station on the beam will now respond in the following manner. Since, a linear element is used, only the average of the linearly varying bending moment is expected in each finite element. If the element is field-consistent, the constant b1 can be associated after accounting for discretization, to relate to the constant bending moment M0 at the element centroid as,

M 0 = EI b1 l b1 = M 0l EI

or (6.17)

In a field-inconsistent problem, due to shear locking, it is necessary to consider the modified flexural rigidity I’ (see Equation 6.17) that modifies b1 ′ to b1 , that is,

′ b1 = M 0l EI ′ = {M 0l EI (1 + e )} = b1 (1 + e )

where e = kGAl2 3EI .

(6.18)

Thus, in a field-inconsistent formulation, the constant b1 gets stiffened by the factor e; the constant bending moment M0 is also underestimated by the

65

same factor. Also, for a very thin beam where e>>1, the centroidal moment M0 2 predicted by a field-consistent element diminishes in a t rate for a beam of rectangular cross-section. These observations have been confirmed through digital computation. The field-consistent element will respond with V = V0 = Q over the entire element length 2l. The field-inconsistent shear force V from Equations (6.16) and (6.18) can be written for a very thin beam (e>>1) as,

V = Q + (3M 0 l ) (x l )

(6.19)

These are the violent shear force linear oscillations within each element, which originate directly from the field-inconsistency in the shear strain definition. These oscillations are also seen if field-consistency had been achieved in the element by using reduced integration for the shear strain energy. Unless the shear force is sampled at the element centroid (i.e. Gaussian point, x/l=0), these disturbances will be much more violent than in the exactly integrated version.

**6.1.7 The consistent formulation of the linear element
**

We can see that reduced integration ensures that the inconsistent constraint does not appear and so is effective in producing a consistent element, at least in this instance. We must now satisfy ourselves that such a modification did not violate any variational theorem. The field-consistent element, as we now shall call an element version free of spurious (i.e. inconsistent) constraints, can and has been formulated in various other ways as well. The `trick' is to evaluate the shear strain energy, in this instance, in such a way that only the consistent term will contribute to the shear strain energy. Techniques like addition of bubble modes, hybrid methods etc. can produce the same results, but in all cases, the need for consistency of the constrained strain field must be absolutely met. We explain now why the use of a trick like the reduced integration technique, or the use of assumed strain methods allows the locking problem to be overcome. It is obvious that it is not possible to reconcile this within the ambit of the minimum total potential principle only, which had been the starting point of the conventional formulation. We saw in Chapter 2, an excellent example of a situation where it was necessary to proceed to a more general theorem (one of the so-called mixed theorems) to explain why the finite element method computed strain and stress fields in a `best-fit' sense. We can now see that in the case of constrained media problems, the mixed theorem such as the Hu-Washizu or Hellinger-Reissner theorem can play a crucial role in proving that by modifying the minimum total potential based finite element formulation by using an assumed strain field to replace the kinematically derived constrained field, no energy, or work principle or variational norms have been violated.

66

To eliminate problems such as locking, we look for a consistent constrained strain field to replace the inconsistent kinematically derived strain field in the minimum total potential principle. By closely examining the strain gradient operators, it is possible to identify the order up to which the consistent strain field must be interpolated. In this case, for the linear displacement interpolations, Equations (6.5b), (6.7a) and (6.7b) tell us that the consistent interpolation should be a constant. At this point we shall still not presume what this constant should be, although past experience suggests it is the same constant term seen in Equation (6.7a). Instead, we bring in the Hellinger-Reissner theorem in the following form to see the identity of the consistent strain field clearly. For now, it is sufficient to note that the Hellinger-Reissner theorem is a restricted case of the Hu-Washizu theorem. In this theorem, the functional is stated in the following form,

T T ò − 1 2 EI χ χ + EI χ χ − 1 2 kGA γ

(

T

γ + kGA γ

T

γ dx

)

(6.20)

where χ

and γ

are the new strain variables introduced into this multi-field

principle. Since we have difficulty only with the kinematically derived γ we can have χ = χ and recommend the use of a γ which is of consistent order to replace

γ. A variation of the functional in Equation (6.20) with respect to the as yet undetermined coefficients in the interpolation for γ yields

ò δγ

T

(γ

− γ ) dx = 0

(6.21)

This orthogonality condition now offers a means to constitute the coefficients of the consistent strain field from the already known coefficients of the kinematically derived strain field. Thus, for γ given by Equation (6.5b), it is possible to show that γ = (b0 − α 1 l ) . In this simple instance, the same result is obtained by sampling the shear strain at the centroid, or by the use of onepoint Gaussian integration. What is important is that, deriving the consistent strain-field using this orthogonality relation and then using this to compute the corresponding strain energy will yield a field-consistent element which does not violate any of the variational norms, i.e. an exact equivalence to the mixed element exists without having to go through the additional operations in a mixed or hybrid finite element formulation, at least in this simple instance. We say that the variational correctness of the procedure is assured. The substitute strain interpolations derived thus can therefore be easily coded in the form of strain function subroutines and used directly in the displacement type element stiffness derivations.

**6.1.8 Some concluding remarks on the linear beam element
**

So far we have seen the linear beam element as an example to demonstrate the principles involved in the finite element modeling of a constrained media problem. We have been able to demonstrate that a conceptual framework that includes a condition that specifies that the strain fields which are to be constrained must satisfy a consistency criterion is able to provide a complete scientific basis for the locking problems encountered in conventional displacement type modeling. We have also shown that a correctness criterion (which links the assumed strain variation of the displacement type formulation

67

the engineering theory of beams predicts a tip deflection of w=4. We shall now proceed to explain these features using the field-consistency paradigm. in fact it produces reasonably meaningful results. the element does not lock. the quadratic beam element shall be used to examine problems such as encountered in curved beam and shell elements and in quadrilateral plate elements due to non-uniform mapping.e.to the mixed variational theorems) allows us to determine the consistent strain field interpolation in a unique and mathematically satisfying manner. This is a valuable exercise as later.4 shows a typical comparison between the linear and quadratic beam elements in its application to a simple problem. We can now use the quadratic beam element to demonstrate that this is true in finite element approximations as well. on the Figure). as in the field-consistent linear element.2. It is seen that for this simple problem. In a higher order formulation. We shall consider three finite element idealizations of this problem with the linear 2-node field-consistent element considered earlier in this section (2C. 6. width 0. on the Figure) and the quadratic 3-node fieldconsistent element which we shall derive later (3C). instead of being able to produce linearaccurate bending moments could now yield only a constant bending moment within each element. i.1 The conventional formulation Consider a quadratic beam element designed according to conventional principles.3.01 m has a vertical tip load of 100 N applied at the tip. the 3C element produces exact results. This resulted in sub-optimal performance of the approximation. the quadratic 3-node field-inconsistent element being discussed now (3I. which were identical. We saw in Chapter 5 that the conventional formulation with lowest order interpolation functions led to spurious constraints and a non-singular assembled stiffness matrix. As the beam becomes very thin. A uniform cantilever beam of length 1. 6. 6.01 m and 10 depth 0. It also produced moment predictions. i. Further. The curious aspects that call for further investigation are: the quadratic element (3I) behaves in exactly the same way as the fieldconsistent linear element (2C). exact integration of all energy terms arising from a minimum total potential principle. The 3I and 2C elements show identical convergence trends and behave as if they are exactly alike. For E=10 2 N/m and µ=0. 68 . the matrix was singular but the spurious constraints resulted in a system that had a higher rank than was felt to be desirable. there were now quadratic oscillations in the shear force predictions for such an element. which result in locking. Note now that these curious features cannot be explained from the old arguments. as it is able to simulate the constant shear and linear bending moment variation along the beam length.. which linked locking to the non-singularity or the large rank or the spectral condition number of the stiffness matrix.e.0 m. giving exactly the same accuracy for the same number of elements although the system assembled from the former had nearly twice as many nodes. It will be useful now to see how these concepts work if a quadratic beam element is to be designed. Fig. the quadratic beam element.0 m.2 The quadratic Timoshenko beam element We shall now very quickly see how the field-consistency rules explain the behavior of a higher order element.

we can argue that the coefficients of the strain field expanded in terms of the Legendre polynomials must vanish separately. when we introduce the penalty limit condition that for a thin beam the shear strain energies must vanish. Indeed. In this case.Fig. 6. U s = 1 2 (kGA ) (2l ) {(b 0 2 + b2 3 − a1 l )2 + 1 3 (b1 − 2a2 l )2 + 4b2 45 } (6. γ = (b0 + b2 3 − a1 l ) + (b1 − 2a2 l )ξ − b2 3 1 − 3ξ 2 ( ) (6. we emphasize the usefulness of expanding the strain field in terms of the Legendre polynomials.24a) 69 . If quadratic isoparametric functions are used for the field-variables w and θ in the following manner w = a0 + a1 (x l ) + a2 (x l )2 θ = b0 + b1 (x l ) + b2 (x l )2 the shear strain interpolation will be. the strain energy due to transverse shear strain is.23) Therefore.4 A uniform cantilever beam with tip shear force convergence trends of linear and quadratic elements.22) Again. three constraints emerge: (b0 + b2 3 − a1 l ) → 0 (6. because of the orthogonal nature of the Legendre polynomials the discretized energy expression becomes the sum of the squares of the coefficients multiplying the Legendre polynomials. When the strain energies are integrated.

4). it will diminish the rate of convergence of the element and would induce disturbances in the form of violent quadratic oscillations in the shear force predictions. is "locked" to a vanishingly small value.1). iii) there are quadratic oscillations in the shear force field within each element. is an unnecessary restriction on the freedom of the interpolation for θ. in a consistent manner.2 Functional reconstitution We can use the functional re-constitution technique to see how the inconsistent terms in the shear strain interpolation alter the description of the physics of the original problem (we shall skip most of the details.2. ii) the 3I element can predict only a constant moment within each element. We can argue that this accounts for the poor behavior of the field-inconsistent quadratic beam element (the 3I of Fig. unlike the linear beam element seen earlier. that the b2 term. This constraint. exactly as the 2C element does. when enforced.24b) (6.1 derives the effect more precisely. This is therefore the fieldconsistent part of the shear strain interpolation.24b) represent constraints having contributions from the field interpolations for both w and θ. 6. This has been explained by showing using the functional re-constitution technique.xx→0 in a discretized sense over each beam element region. which it replaces. as we shall see in the next section.24c) however contains a constant only from the interpolation for θ. demonstrating that the following features can be fully accounted for: i) the displacement predictions of the 3I element are identical to that made by the 2C element on an element by element basis although it has an additional midnode and has been provided with the more accurate quadratic interpolation functions. true constraints that reflect a physically meaningful imposition of the thin beam Kirchhoff constraint. However. 6. which describes the linear variation in the bending strain and bending moment interpolation. The 3I element then effectively behaves as a 2C element in being able to simulate only a constant bending-moment in each region of a beam. field-inconsistency in this element would not cause the element to lock. 6.(b1 − 2a2 l ) → 0 b2 → 0 (6. They can therefore reproduce. reflecting its origin through the spurious constraint. constraining it in fact to behave only as a linear interpolation as the constraint implies that θ. as the material is available in greater detail in Ref. The b2 term that appears in the bending energy also makes an appearance in the shear strain energy.24a) and (6. Therefore. The spurious energy implied by such a constraint does not contribute directly to the discretized bending energy. 70 . 6.24c) Equations (6. We have already discussed earlier that the 3I element (the fieldinconsistent 3-noded quadratic) converges in exactly the same manner as the 2C element (the field-consistent linear). Equation (6. Ref.

3 The consistent formulation of the quadratic element As in the linear element earlier. plate elements are the most commonly used elements in general purpose structural analysis. Such theories had difficulties and limitations and a1so attention 0 turned to what are called the C theories. At first. 71 .e. we are required to replace γ in Equation (6. We use the inverted commas to denote that the strain is not assumed in an arbitrary fashion but is actually uniquely determined by the consistency and the variational correctness requirements. In a hybrid assumed strain approach. 6.22) which had been derived from the kinematically admissible displacement field interpolations using the strain-displacement operators with an `assumed' strain field γ which contains terms only upto and including the linear Legendre polynomial in keeping with the consistency requirement. This determines how c0 and c1 should be constituted from b0.3 The Mindlin plate elements A very large part of structural analysis deals with the estimation of stresses and displacements in thin flexible structures under lateral loads using what is called plate theory.25) should replace γ over the length of the element. The quadratic Legendre polynomial and its coefficient are simply truncated and c0=b0 and c1=b1 represent the variationally correct fieldconsistent `assumed' strain field.2. The re-constitution of the field is to be done in a variationally correct way. Thus. especially as they can cover applications to moderately thick and laminated plate and shell constructions.21) dictates how γ (6. The use of such an interpolation subsequently in the integration of the shear strain energy is identical to the use of reduced integration or the use of a hybrid assumed strain approach. the simple four-node quadrilateral plate element (the QUAD4 element) may account for as much as 80% of all usage. Fortunately. i. b1 and b2. the field-consistent element (3C) can be formulated in various ways.2] is now the most commonly used basis for the development of plate elements. most General Purpose Packages (GPPs) for structural analysis used plate elements based on what are called the 1 C theories. It is therefore important to understand that the evolution of the current generation of QUAD4 elements from those of yester-year. Reduced integration of the shear strain energy using a 2-point Gauss-Legendre formula was the most popular method of deriving the element so far. The Mindlin plate theory [6.6. over a span of nearly three decades was made difficult by the presence of shear locking. Let us now derive this element using the `assumed' strain approach. such a consistent re-constitution is automatically implied in the choice of assumed strain functions and the operations leading to the derivation of the flexibility matrix and its inversion leading to the final stiffness matrix. Let us write this in the form γ = c0 + c1ξ The orthogonality condition in Equation (6. We shall now see how this takes place. It has been estimated that in large scale production runs using finite element packages. the orthogonal nature of the Legendre polynomials allows this to be done for this example in a very trivial fashion.

as the plates became thinner. In Mindlin's theory [6. iii) the selective 2×3 and 3×2 rule for the transverse shear strain energies from γxz and γyz recommended for a 8-node element also failed to remove shear locking. which have permitted such an improved description of plate behavior. rotations of lines normal to the midsurface of the undeformed plate) and the mid-surface deflection w. the section rotations θx and θy (i. This is in fact the two-dimensional manifestation of the same problem that we encountered for the Timoshenko beam element. deformation is described by three quantities. identified as membrane locking [6. The degenerate shell elements that we discussed briefly at the beginning of this section can be considered to correspond to a Mindlin type representation of the transverse shear effects.The history behind the discovery of shear locking in plate elements is quite interesting. the deformation is completely described by the transverse displacement w of the mid-surface. It was first recognized when an attempt was made to represent the behavior of shells using what is called the degenerate shell approach [6. We shall confine our study now to plate elements without going into the complexities of the curved shell elements. These elements behaved very poorly in representing even the trivial example of a plate in bending and the errors progressed without limit. Unlike classical plate or shell theory. The bending strains are now derived from the section rotations and do not cause any 72 .3]. the transverse shear deformation is ignored. In such a description. ii) the 2×2 rule in the 9-node Lagrangian element removed locking but introduced zero energy modes. it is necessary to introduce additional degrees of freedom. The difficulty was attributed to shear locking.4].2].6. In Kirchhoff-Love thin plate theory. due to an additional factor.6]. Such an approach was therefore equivalent to a Mindlin theory formulation. the transverse shear strain and its energy is therefore accounted for in this formulation. Some of these were. the degenerate shell elements performed poorly when trying to represent the bending of curved shells.5. originating now from the need for consistency of the membrane strain interpolations.7]. We shall now consider Mindlin's approximations. This was only partially successful and many issues remained unresolved. The remedy proposed at once was the reduced integration of the shear strain energy [6.e. To account for transverse shear effects. iv) the same selective 2×3 and 3×2 rule when applied to a 9-noded element is optimal for a rectangular form of the element but not when the element was distorted into a general quadrilateral form. ironically it was noticed first in the degenerate shell element and was only later related to the problems in designing Timoshenko beam and Mindlin plate elements [6. In this the shell behavior is modeled directly after a slight modification of the 3D equations and shell geometry and domain are represented by a 3D brick element but its degrees of freedom are condensed to three displacements and two section rotations at each node. v) even after reduced integration of the shear energy terms. i) the 2×2 rule failed to remove shear locking in the 8-node serendipity plate element. We shall consider the membrane-locking phenomenon in another section.

x γ yz = θ y − w.3. It was not clear why the quadrilateral form locked even after reduced integration. we shall postpone the discussion on why such a strategy cannot preserve consistency if the element was distorted to a later section.y ) dx dy [ ] } 73 .difficulty when a finite element model is made. an optimum integration rule can be found which ensures consistency if the element is rectangular. U = UB + US = 24 1 − ν ( Et2 2 ) {ò ò [θ 2 x .x 6k (1 − ν ) t 2 ( )2 ] dx dy (6. linear elastic plate element according to Mindlin theory can be constituted from its bending and shear energies as. Also.4] as it is in this configuration that the consistency requirements can be easily understood and enforced. Locking was seen to vanish for the rectangular element if the bending energy was computed with a 2×2 Gaussian integration rule while a reduced 1-point rule was used for the shear strain energy. γ xz = θ x − w. after distortion to a quadrilateral form. y + (1 − ν ) 2 θ x. The shear strains are now computed as the difference between the section rotations and the slopes of the neutral surfaces. This rectangular element behaved very well if the plate was thin but the results deteriorated as the plate became thicker. 6.y (6. We can now demonstrate from our consistency view-point why the 1-point integration of the shear strain energy is inadequate to retain all the true Kirchhoff constraints in a rectangular thin plate element.1 The 4-node plate element The 4-node bi-linear element is the simplest element based on Mindlin theory that could be devised.y +2ν θ x .27) + 2 2 ò ò (θ x − w. We shall first investigate the rectangular form of the element [6. the strain energy for an isotropic.there were two zero energy mechanisms in addition to the usual three rigid body modes required for such an element. thus.4]. [6. locking re-appeared.x 2 +θ y .x ) + (θ y − w. 6. It is the inconsistent representation of the latter that causes shear locking. Following Ref. In fact.y + θ y.x θ y .4 that an exactly integrated Mindlin plate element would lock even in its rectangular form. A spectral analysis of the element stiffness matrix revealed a rank deficiency .26) The stiffness matrix of a Mindlin plate element will now have terms from the bending strain energy and the shear strain energy. However. It was established in Ref. It was the formation of these mechanisms that led to the deterioration of element performance if the plate was too thick or if it was very loosely constrained.

θx and θy are the section rotations.30a) b2 − a3 → 0 b1 → 0 b3 → 0 (6. γxz. y are Cartesian co-ordinates (see Fig.5 Cartesian and natural coordinate system for a four-node rectangular plate element. The admissible displacement field interpolations required for a 4-node element can be written in terms of the Cartesian co-ordinates itself as. ν is the Poisson's ratio.30d) 74 .28b) admissible shape γ xz = (b0 − a1 ) + (b2 − a3 )y + b1x + b3 xy (6. the shear strains must vanish. where x. w is the transverse displacement.Fig.29) are. in the Cartesian system. 6.30b) (6. The discretized constraints that are seen. field derived from these kinematically (6.5). b0 − a1 → 0 (6. w = a0 + a1x + a2 y + a3 xy θ = b0 + b1x + b2 y + b3 xy The shear strain functions is. The factor k is introduced to compensate for the error in approximating the shear strain as a constant over the thickness direction of a Mindlin plate.28a) (6.29) As the plate thickness is reduced to zero. Let us now examine the field-consistency requirements for one of the shear strains. E is the Young's modulus. to be enforced as γ xz → 0 in Equation (6.30c) (6. k is the shear correction factor and t is the plate thickness. 6.

This element would be the optimal rectangular bi-linear Mindlin plate element.32) Thus. the requirement for consistency of the interpolations for the shear strains in the Cartesian co-ordinate system is easily recognized as the polynomials use only Cartesian co-ordinates. b1. This strategy can also be seen to be variationally correct in this case. Let us now look at the 1-point integration strategy used in Ref. the part of the shear strain energy contributed by γxz.xy ( 2 )0 ] (6. we let the centroid of the element lie at the origin of the Cartesian co-ordinate system).4. We must integrate exactly. only the true constraints are retained and all spurious constraints are removed. 6.30c) and (6. It is clear from Equations (6.29) and (6. (b2-a3)y. By a very similar argument. b1x.30) that the terms b1x and b3xy are the inconsistent terms which will contribute to locking in the form of spurious constraints.x→0 and θx. ò ò γ xzdx dy = 4lh [(θ x − w.x )2 .x )0 + h 2 2 2 2 3 (θ x − w.xy→0 in the element.4 led to zero energy mechanisms. 6. ò ò γ xz dx dy = 4lh [(θ x − w. We now identify terms such as (b0-a1). If a 1×2 Gaussian integration is used. we can show that the part of the shear strain energy from γyz will require a 2×1 Gaussian integration rule. (θx.33) 75 .x)0. and (θx. y0 + l 2 3 (θ x . An exact integration.xy)0 where the subscript ‘0’ denotes the values at the centroid of the element (for simplicity. This will give shear energy terms such as.x)0.x). (θx-w.x )0 + h 2 2 2 3 (θ x − w. terms such as (b0-a1). (b2-a3). we have.30a) and (6. y0 ] (6.these are the spurious or `inconsistent' constraints which lead to shear locking.30d) are the cause for concern here . We shall consider first. Let us now try to derive the optimal element and also understand why the simple 1-point strategy of Ref. Thus. in a rectangular element.x )0 ] 2 2 (6.x )2 .x )0 + h 2l 2 9 θ x . these four quantities act as constraints.The constraints shown in Equations (6. we shall see later that in a quadrilateral case. and b3xy.31) In the penalty limit of a thin plate. that is a 2×2 Gaussian integration of the shear strain energy leads to ò ò γ xzdx dy = 4lh [(θ x − w. The first two reproduce the true Kirchhoff constraints and the remaining two act as spurious constraints that cause shear locking by enforcing θx. and b3 as being equivalent to the quantities (θx-w. Constraints (6. it is not possible to ensure variational correctness exactly.30b) are physically meaningful and represent the Kirchhoff condition in a discretized form.y0. Let us now look for optimal integration strategies for removing shear locking without introducing any zero energy mechanisms.

6. This makes it more difficult to restore consistency in a simple manner. An exact integration of bending and shear strain energies resulted in an element that locked for most practical boundary suppressions even in its rectangular form. This is not so simple in a general quadrilateral where the complication arising from the isoparametric mapping from a natural co-ordinate system to a Cartesian system makes it very difficult to see the consistent form clearly.29) so that they are also variationally correct.3] and which actually pre-dates the 4-node Mindlin element. 6. the reduced and selective integration techniques. etc. It is therefore very important for one to know points of optimal stresses in the Mindlin plate elements.We have now only one true constraint each for the shear energy from γxz and γyz respectively while the other Kirchhoff constraints (θ x − w. it is well known that the 9-node element in its rectangular form is free of shear locking even with exact integration of shear energy terms and that its performance is vastly improved when its shear strain energies are integrated in a selective sense (2×3 and 3×2 rules for γ xz and γ yz terms respectively). Many ad-hoc techniques e. It is in fact analogous to the quadratic Timoshenko beam element. There are now two possibilities. as the nodes are usually the points where the strains and stresses are least accurate. We have seen now that it is a very simple procedure to re-constitute field-consistent assumed strain fields from the kinematically derived fields such as shown in Equation (6. It would seem that an higher order element based on quadratic functions would be far more accurate. (θ y − w. It is possible however to determine points of optimal stress recovery using an interpretation of the displacement method as a procedure that obtains strains over the finite element domain in a least-squares accurate sense.3 Stress recovery from Mindlin plate elements The most important information a structural analyst looks for in a typical finite element static analysis is the state of stress in the structure. y0 → 0 and are lost.2 The quadratic 8-node and 9-node plate elements The 4-node plate element described above is based on bi-linear functions.4. This introduces two zero energy modes and accounts for the consequent deterioration in performance of the element when the plates are thick or are very loosely constrained. It seems possible to attribute this noticeable difference in the performance of the 8. hybrid and mixed methods.g.x0 → 0 6. the fieldinconsistencies not being severe enough to cause locking.y ). an 8-node element based on what are called the serendipity functions and a 9-node element based on the Lagrangian bi-quadratic functions.3. It was therefore regarded for some time as an unreliable element as no quadrature rule seemed to be able to eliminate locking entirely without introducing other deficiencies. We shall see the difficulties associated with this form in a later section. as shown in Ref. failed or succeeded only partially. we saw a basis for this interpretation. There has been a protracted debate on which version is more useful. This is however not true for the 8-node element which was derived from the Ahmad shell element [6. By now. It is known that the stress recovery at nodes from displacement elements is unreliable.3.x ). In Chapter 2.and 9-node elements to the missing central node in the former. We can apply this rule to determine points of accurate stress 76 . both versions having fiercely committed protagonists.

15291543. Int. accurate recovery of stresses from the 8-node element is still a very challenging task because of the difficulty in formulating a robust element.the stress analyst must then exercise care in applying these rules to seek reliable points for recovering stresses. C. Num. that the shear stress resultants are most accurate at the same points at which they must be sampled in a selective integration strategy to remove the fieldinconsistencies! For anisotropic cases. 1951. Meth.recovery in the Mindlin plate elements. 575-586.2 R. However. The most efficient elements known today are variationally incorrect even after being made field-consistent and need special filtering techniques before the shear stress resultants can be reliably sampled. 1977. Num. 1970. M. W.8] (note that in a fieldinconsistent 4-node element. there will be violent linear oscillations in the shear stress resultants corresponding to the inconsistent terms). Appl. Int. 6. Influence of rotary inertia and shear on flexural motion of elastic plates. Int. When the elements are distorted. Irons and O. 1986.5 References 6. R.1 G.4 Concluding remarks We can conclude this section on shear locking by noting that the available understanding was unable to resolve the phenomena convincingly. So far. 19731984. Prathap and C. it is no simple matter to determine the optimal points for stress recovery . B. 6. 18. Hughes.5 S. Num. The proposed improvement. allowed us to derive an error estimate for a case under locking and we could show through numerical (digital) experiments that these estimates were accurate. Field-consistent strain interpolations for the quadratic shear flexible beam element. Taylor and W. together with the functional re-constitution procedure. 6. 6. 2. J. D. Engng. Mindlin. Babu. L. Ref. discussion on stress sampling has been confined to rectangular elements. Num. Improved numerical integration of thick shell finite elements. Thus. 1971. A simple and efficient finite element for plate bending. it is safest to sample all stress resultants (bending and shear) at the centroid. Ahmad. Clough. Kanoknukulchal. Analysis of thick and thin shell structures by curved finite elements. 6. Meth. R. 6. 31-38.4 T. the points are very easy to determine [6. Pawsey and R. Engng. In this way we are convinced that a theory with the consistency paradigm is more successful from the falsifiability point of view than one without. 43. Zienkiewicz. Engng. J. J. Mech. R. and therefore fortuitous. J. Int. 411. which was the consistency paradigm.3 S. Such rules can be extended directly to the 9-node rectangular element.8 shows that bending moments and shear stress resultants Qxz and Qyz are accurate at the centroid and at the 1×2 and 2×1 Gauss points in a rectangular element for isotropic or orthotropic material. The bending moments are now accurate at the 2×2 Gauss points and the shear stress resultants in an isotropic or orthotropic problem are optimal at the same 2×3 and 3×2 Gauss points which were used to remove the inconsistencies from the strain definitions. For a field-consistent rectangular 4node element. Meth. J. 6. 23. 77 . Engng. J. Meth. It is coincidental. 419-451. F.

Accurate force evaluation with a simple bi-linear plate bending element.7 H. Taylor and J.8 G. 25. Belytschko. R.6. 78 . Int. C. 172-178. 259-270. M. 43. 1971. Prathap and C. Comp.6 O. 49. J. Babu. Reduced integration technique in general analysis of plates and shells. L. Meth. 1982. R. 6. 1987. Stolarski and T. Mech. Engng. 275290. 6. Num. Membrane locking and reduced ingression for curved elements. Zienkiewicz. J. Struct. Too. Appl.

etc.these are no longer dependent on the various forms of shell theories proposed and should be simple to use. Flugge. One key issue behind this locking . 7.e. The second approach is called the degenerate shell approach three dimensional solid elements can be reduced (degenerated) into shell elements having only mid-surface nodal variables . There are considerable controversies regarding the relative merits of these theories as each has been obtained by carrying out approximations to different degrees when the 3-dimensional field equations are reduced to the particular class of shell equations. parasitic shear and incompressible locking 7. One popular and efficient form of construction of such thin flexible structures is the shell . we argued that the most part of structural analysis deals with the behavior of thin flexible structures. a detailed analysis is omitted as it would be beyond the scope of the present book and readers should refer to the primary published literature. The field-consistency paradigm now allows all these phenomena to be traced to inconsistent representations of the constrained strain fields. This 79 .2 Membrane locking Earlier. Sanders. Hence. However.). There are two ways in which this could be done.1 Introduction The shear locking phenomenon was the first of the over-stiffening behavior that was classified as a locking problem. Such a model is understandably inaccurate in that with very coarse meshes. they do not capture the bending-stretching coupling of thin shell behavior. accurate and robust difficult to design .whether it is shear locking in a straight beam or flat plate element. Other such pathologies were noticed in the behavior of curved beams and shells.Chapter 7 Membrane locking. It is possible to perform a finite element analysis of a shell by using what is called a facet representation .5). membrane locking in a curved beam or shell element. Vlasov theories. the shell surface is replaced with flat triangular and/or quadrilateral plate elements in which a membrane stiffness (membrane element) is superposed on a bending stiffness (plate bending element). A shell is therefore the curved form of a plate and its structural action is a combination of stretching and bending. in plane stress modeling and in modeling of three dimensional elastic behavior at the incompressible limit (ν→0. They are in fact equivalent to a Mindlin type curved shell element.the very poor behavior of curved shell elements have been extremely challenging researchers for nearly three difficulty is the phenomenon of membrane curved elements. Unlike the chapter on shear locking. the Donnell.a problem decades. as they become thin. parasitic shear in 2D plane stress and 3D elasticity or incompressible locking in 2D plane strain and 3D elasticity as the Poisson's ratio ν→0.enclosing space using one or more curved surfaces. the motivation for designing elements with mid-surface curvature taken into account and which can capture the membrane-bending coupling correctly.i. One is to use elements based on specific shell theories (e.g. This chapter will now briefly summarize this interpretation. A unifying pattern is therefore introduced to the understanding of the locking problems in constrained media elasticity .5.

1 The classical thin curved beam element. We shall now see that a similar problem appears in curved finite elements in which both flexural and membrane deformation take place. i.1 The classical thin curved beam element Figure 7. We saw that simple finite element models for the shear deformable bending action of a beam or plate behaved very poorly in thin beam or plate regimes where the shear strains become vanishingly small when compared with the bending strains.e.1b) required. In this section. Such elements perform very poorly in cases where inextensional bending of the curved structure was predominant.1 describes the simplest curved beam element of length 2l and radius of curvature R based on classical thin beam theory. the classical thin curved beam element and the linear and quadratic shear flexible curved beam elements in a curvilinear co-ordinate system.2. The displacement degrees of freedom required are the circumferential displacement u and the radial displacement w. The co-ordinate s follows the middle line of the curved beam.s + w R χ = u. This is most easily identified by working with the onedimensional curved beam (or arch) elements. 7. we shall apply the field-consistency paradigm to the problem to recognize that a consistent representation of membrane strains is desired to avoid the phenomenon called membrane locking and that it is the need for consistency rather than the requirement for strain-free rigid body motion which is the key factor. In this section. 7. Kinematically 80 . we shall examine in turn. was neither recognized nor understood for a very long time. This phenomenon is known as shear locking. It was believed at first that this was because curved elements could not satisfy the strain-free rigid body motion condition because of their inherent curved geometry.ss A C description for u and a C1 description w is admissible displacement interpolations for u and w are 0 (7. we studied structural problems where both bending and shear deformation was present. In Chapter 6. the membrane strains become vanishingly small when compared to the bending strains. We also saw how the field-consistency paradigm was needed to explain this behavior in a scientifically satisfying way. This phenomenon has now come to be called membrane locking.1a) (7. The membrane strain ε and the bending strain χ are described by the straindisplacement relations ε = u.s R − w.Fig.

3b) If a thin and deep arch (i.s at the two nodes.2) is modeled by the curved beam element.2a) + b3 ξ 3 (7. see Fig. From Equation (7. 81 .3a) 3 5R 3ξ − 5ξ 3 ( ) (7. 7. having a large span to rise ratio so that L/t>>1 and R/H is small.e. It can therefore represent the condition ε = u.4a) has terms participating from both the u and w fields.4d) Following the classification system we introduced earlier.2 Geometry of a circular arch u = a0 + a1ξ w = b0 + b1ξ + b2ξ 2 (7.4b) (7.4c) (7. we can observe that constraint (7.4a) (7.3b) we see that the inextensibility condition leads to the following constraints: a1 l + b0 R + b2 3R → 0 b1 + 3b3 5 → 0 b2 → 0 b3 → 0 (7. The strain field interpolations can be derived as χ = a1 Rl − 2b2 l 2 − 6b3 l 2 ξ ε = (a1 l + b0 R + b2 3R ) + (b1 R + 3b3 5R )ξ − b 2 2 ( ) ( ) 3R (1 − 3ξ )− b (7.s + w R → 0 in a physically meaningful way.Fig. 7.2b) where ξ = s l and a0 to b3 are the generalized degrees of freedom which can be related to the nodal degrees of freedom u. the physical response is one known as inextensional bending such that the membrane strain tends to vanish. w and w.

82 .sss → 0 . For simplicity assume that the cross-section of the beam is rectangular so that if the depth of the beam is t. From the three constraints we have the conditions b1 → 0. For a curved beam of length 2l.i. This element has been studied extensively and the numerical evidence given there confirms that membrane locking is mainly determined by the first of these three constraints.ss → 0 and w. the strain energy can be written as. After discretisation. then I = At2 12 . it is a true constraint. These are the spurious constraints. The loading system acting on it is assumed to induce nearly inextensional bending. w.ss and neglecting the consistently modeled part of the membrane energy) one can show that due to membrane locking. We therefore choose a problem of simple configuration which can reveal the manner in which the membrane locking action takes place.the principal stiffening term is physically equivalent to a spurious inplane or membrane stiffening action due to an axial load N = EAl2 3R 2 .4d) have no participation from the u field. However.4d)) modify the physics of the curved beam problem in the discretization process.6) We can therefore expect membrane locking to depend on the structural parameter 2 of the type (Ll/Rt) . Each in turn implies the conditions w. the strain energy of the discretized arch (again assuming that χ=-w. We consider a shallow circular arch of length L and radius of curvature R.e. (which in Chapter 6 revealed how the discretization process associated with the linear Timoshenko beam element led to large errors known as shear locking) to see how the inconsistent constraints in the membrane strain field interpolation (Equations (7. We can now use the functional re-constitution technique.4b) to (7. moment of inertia I and area of cross-section A.5) where ε and χ are as defined in Equation (7. b2 → 0 and b3 → 0. We also assume that the transverse deflection w can be approximated in the form w=csin(πs/L). the three remaining constraint (7. Let us now examine in turn what these three constraints imply for the physical problem. If the entire arch is discretized by curved beam elements of length 2l each. simply supported at the ends.s → 0.4b) to (7. and reconstitution of the functional. we can show that the inconsistent membrane energy terms disturb the bending energy terms . Next. we shall briefly describe the functional re-constitution analysis that will yield semi-quantitative error estimates for membrane locking in this element. This action is quite complex and it is not as simple to derive useful universally valid error estimates as was possible with the linear Timoshenko beam element. the bending rigidity of the arch has now been altered in the following form: ( ) (EI )′ (EI ) = { + (4 1 π 2 (Ll Rt )2 ) } (7. π e = ò 1 2 EI χ T χ + 1 2 EA ε T ε ds ( ) (7.1). The analysis is a little more difficult than for the shear locking case and would not be described in full here.

We have also seen how we could restore the element into a very accurate one (having the accuracy of the straight beam models) by designing a new membrane strain field which met the consistency requirements. As in the Timoshenko beam element. Of interest to us now is to understand how the membrane locking phenomenon can take place in a general curved shell element.s which describes the shear strain at the section. 7. The strain energy in a Mindlin curved beam is U=UM+UB+US where the respective contributions to U arise from the membrane strain ε.Fig. Fig.3 The Mindlin curved beam element. 7. The belief that the lack of the strain free rigid body motion in the conventional formulation was the cause of these errors cannot be substantiated at all. These are now described by the circumferential (tangential) displacement u. 7. it will be useful to investigate the behavior of the Mindlin curved beam elements. To place the context of membrane locking in general shell elements correctly.2. The curved beam element used to demonstrate this was based on classical thin beam theory and used a curvilinear co-ordinate description. Most general shell elements which have now been accepted into the libraries of general purpose finite element packages are based on what is called the degenerate shell theory which is equivalent to a shear deformable theory and use quadratic interpolations (i.e. this rotation differs from the rotation of the mid-line w. Numerical experiments with various order integration rules applied to evaluate the various components of the stiffness matrix indicated that the inconsistent elements locked exactly as predicted by the pattern describe above. three nodes on each edge to capture the curved geometry in a Cartesian system).s by an amount γ = θ − w. the bending 83 . the radial (normal) displacement w and the rotation of a normal to the midline θ. This classical curved beam element (the cubic-linear element) was discredited for a very long time because it was not possible to understand why the element should perform so poorly.3 shows how the Mindlin curved beam element is specified. The consistent element was entirely free of locking.2 The Mindlin curved beam elements The exercise above showed that the locking behavior of curved beams could be traced to the inconsistent definition of the membrane strain field. We have now found that the explanation for its very poor behavior could be attributed entirely to the inconsistency in the membrane strain field interpolations.

strain χ and the transverse shear strain γ. For an element of length 2l and radius R,

**U M = (1 2 ) (EA ) ò ε 2 ds where ε = u,s + w R U B = (1 2 ) (EI ) ò χ 2 ds where χ = u,s R − θ ,s U S = (1 2 ) (kGA ) ò γ 2 ds where γ = θ − w ,s
**

where EA, EI and kGA represent the respective rigidities.

(7.7a) (7.7b) (7.7c)

There are no difficulties with the bending strain. However, depending on the regime, both shear strains and membrane strains can be constrained to vanish leading to shear and membrane locking. The arguments concerning shear locking are identical to those expressed in Chapter 6, i.e. while the inconsistency in the shear strain field is severe enough to cause locking in the former (a significant stiffening which caused errors that propagated indefinitely as the structural parameter kGl2 Et2 became very large), it is much milder in the latter and is responsible only for a poorer rate of convergence (the performance of the element having reduced to that of the linear field-consistent element). However with membrane locking, the inconsistencies at the quadratic level can cause locking - therefore both the linear and quadratic elements must be examined in turn to get the complete picture. For the linear Mindlin curved beam element, as a C displacement type formulation is sufficient, we can choose linear interpolations of the following form,

0

**u = a0 + a1ξ w = b0 + b1ξ θ = c0 + c1ξ
**

where ξ = s l is the natural co-ordinate along the arch mid-line. interpolations lead to the following membrane strain field interpolation:

(7.8a) (7.8b) (7.8c) These

ε = (a1 l + b0 R ) + (b1 R ) ξ

(7.9)

There is now one single spurious constraint in the limit of inextensional bending of an arch, i.e. b1→0, which implies a constraint of the form w,s→0 at the centroid of the element. The membrane locking action is similar to but simpler than that seen for the classical thin curved beam element above. In a thin beam, in the Kirchhoff limit, θ→w,s, which implies that membrane locking induces the section rotations θ to lock. It can therefore be seen that a constraint of this form leads to an in-plane stiffening action of the form corresponding to the introduction of a spurious energy

(1 2 ) (EAl2

3R 2

) òw ,

2 s

ds

(7.10)

84

The membrane locking action is therefore identical to that predicted for the classical thin curved beam. A variationally correct field-consistent element can easily be derived using the Hu-Washizu theorem. This is a very accurate element. The quadratic Mindlin curved beam element can also be interpreted in the light of our studies of the classical curved beam earlier. Spurious constraints at both the linear (i.e. w,s→0) and quadratic (i.e w,ss→0) levels lead to rate respectively. We errors that propagate at a (Ll Rt )2 rate and l 2 Rt expect now that in a quadratic Mindlin curved beam element, the latter constraint is the one that is present and that these errors are large enough to make the membrane locking effect noticeable. This also explains why the degenerate shell elements, which are based on quadratic interpolations, suffer from membrane locking.

2

(

)

A quadratic (three-node) element having nodes at s=-l, 0 and l (see Fig. 7.3) would have a displacement field specified as,

**u = a0 + a1ξ + a2ξ 2 w = b0 + b1ξ + b2ξ
**

2

(7.11a) (7.11b) (7.11c)

θ = c0 + c1ξ + c2ξ 2

The membrane strain field can be expressed as:

ε = (a1 l + b0 R + b2 3R ) + (2a2 l + b1 R ) ξ − b2 3R 1 − 3ξ 2

(

)

(7.12)

The spurious constraint is now b2→0 a discretized equivalent of imposing the constraint w,ss→0 at the centroid of each element. To understand that this can cause locking, we note that in a thin curved beam, θ ,s → w,ss and therefore a spurious constraint of this form acts to induce a spurious bending energy. A functional re-constitution approach will show that the bending energy for a thin curved beam modifies to,

2ù é 2 U B = (1 2 ) (EI ) ê1 + (4 15 ) l 2 Rt ú w,ss ds ë û

ò

(

)

(7.13)

This can be interpreted as spurious additional stiffening of the actual bending rigidity of the structure by the (4 15 ) l 2 Rt 2 term. It is a factor of this form that caused the very poor performance of the exactly integrated quadratic shell elements. A variationally correct field consistent element is obtained by using the Hu-Washizu theorem or very simply by using a two-point integration rule for the shear and membrane strain energy energies. This is an extremely accurate element.

(

)

i.e.

(l

**It is interesting to compare the error norm for locking for this element,
**

2

Rt2

)

with the corresponding error norm for locking in the inconsistent

85

linear Mindlin curved beam element, which is L l Rt 2 . It is clear that since L>>l when an arch is discretized with several elements, the locking in the quadratic element is much less than in the linear element. One way to look at this is to examine the rate at which locking is relieved as more elements are added to the mesh, i.e. as l is reduced. In the case of the inconsistent linear 2 element, locking is relieved only at an l rate whereas in the case of the 4 quadratic element, locking is relieved at the much faster l rate. However, locking in the quadratic element is significant enough to degrade the performance of the quadratic curved beam and shell elements to a level where it is impractical to use in its inconsistent form. This is clearly seen from the fact that to reduce membrane-locking effects to acceptable limits, one must use element lengths l << Rt . Thus for an arch with R=100 and t=1, one must use about a 100 elements to achieve an accuracy that can be obtained with two to four field-consistent elements.

(

)

**7.2.3 Axial force oscillations
**

Our experience with the linear and quadratic straight beam elements showed that inconsistency in the constrained strain-field is accompanied by spurious stress oscillations corresponding to the inconsistent constraints. We can now understand that the inconsistent constraint b1→0 will induce linear axial force oscillations in the linear curved beam element. Similarly, the inconsistent quadratic constraint will trigger off quadratic oscillations in the quadratic element.

7.2.4 Concluding remarks

So far, we have examined the phenomenon of membrane locking in what were called the simple curved beam elements. These were based on curvilinear geometry and the membrane locking effect was easy to identify and quantify. However, curved beam and shell elements which are routinely used in general purpose finite element packages are based on quadratic formulations in a Cartesian system. We shall call these the general curved beam and shell elements. Membrane locking is not so easy to anticipate in such a formulation although the predictions made in this section have been found to be valid for such elements.

7.3 Parasitic shear

Shear locking and membrane locking are phenomena seen in finite element modeling using structural elements based on specific theories, e.g. beam, plate and shell theories. These are theories obtained by simplifying a more general continuum description, e.g. the three dimensional theory of elasticity or the two dimensional theories of plane stress, plane strain or axisymmetric elasticity. The simplifications are made by introducing restrictive assumptions on strain and/or stress conditions like vanishing of shear strains or vanishing of transverse normal strains, etc. In the beam, plate and shell theories, this allows structural behavior to be described by mid-surface quantities. However, these restrictive assumptions impose constraints which make finite element modeling quite difficult unless the consistency aspects are taken care of. We shall now examine the behavior of continuum elements designed for modeling continuum elasticity directly, i.e. 2D elements for plane strain or stress or axisymmetric elasticity and solid elements for 3D elasticity. It appears that

86

Reduced integration of the shear strain energy or addition of bubble modes for the 4-noded 87 . The finite element modeling of plane stress.4). 7.g. The problems of shear locking. for which the shear stress should be zero. 7. i. membrane locking and parasitic shear obviously are very similar in nature and the unifying factor is the concept of consistency. In the linear elements. the errors progress indefinitely as the element aspect ratio increases. (c) Eight-node rectangle and (d) Nine-node rectangle. axisymmetric and 3D elasticity problems can be made with two-dimensional elements and threedimensional elements (sometimes called continuum elements to distinguish them from structural elements).5 (for nearly incompressible materials or for materials undergoing plastic deformation). However. Parasitic shear occurs in the plane stress and 3D finite element modeling of beam flexure. in some applications. One such problem is parasitic shear.Fig. We shall see the genesis of this phenomenon and show that it can be explained using the consistency concepts. the quadratic 6-noded triangle and 8.or 9-noded quadrilateral elements (Fig. centroids in linear elements).g. considerable difficulties are seen. (b) Rectangular bilinear element.4) are the simplest known in the 2D case. of which the linear 3-noded triangle and bi-linear 4noded rectangle (Fig. 7. the errors progress in the same fashion as for shear locking in the linear Timoshenko beam element.e. the plane stress modeling of beam flexure or in plane strain and axisymmetric applications where the Poisson's ratio approaches 0. plane strain. Under pure bending loads. locking behavior is possible. the elements yield large values of shear stress except at certain points (e. these elements are reliable and accurate for determining general two-dimensional stress distributions and improvements can be obtained by using the higher order elements.4 (a) Constant strain triangle. under constraining physical regimes. e. In most applications.

in their original (i. For an isotropic case. UE and shear strains. 7.Fig. Here. the displacements u and v (Fig. The strain energy functional for this problem consists of energies from the normal strains.14) A one-dimensional beam theory can be obtained by simplifying this problem. These elements. where E is the Young's modulus.y +2ν u. y + Gb 2 u.2 The 4-node rectangular element An element of length 2l and depth 2t is chosen. For a very slender beam. field-inconsistent) form show severe linear and quadratic shear stress oscillations.ν û ë ( )( ) ( ) (7.x + v2 . 7.x v . we shall extend the consistency paradigm to examine this phenomenon in detail. u1-u4 and v1-v4 at the four nodes. Two independent field variables are required to describe the problem.5). The field-variables are now interpolated in the following fashion: 88 .3. 7. UG.e.1 Plane stress model of beam flexure Figure 7. confining attention to the rectangular elements. The 8. shear stresses and shear forces would remain finite. U = U E + UG ù Eb L T é u 2 . G is the shear modulus and ν is Poisson's ratio. the shear strains must vanish and this would then yield the classical Euler-Bernouilli beam theory. depth T and thickness (in the normal to the plane direction) b. rectangular element completely eliminates this problem.x 2 ú dx dy = ò0 ò0 ê 2 ú ê 2 1.and 9-noded elements do not lock but improve in performance with reduced integration. We shall see now that it is this constraint condition that leads to difficulties in plane stress finite element modeling of such problems. y +v .5 shows a two-dimensional plane stress description of the problem of beam flexure.3. The rectangular form is chosen so that the issues involved can be identified clearly. 7.5 Plane stress model of beam flexure. Note that although shear strains vanish. There are now 8 degrees of freedom. The beam is of length L.

γ = u.x = (a2 + b1 ) + a3 x + b3 y and the shear strain energy within the element will be.17). This can enforce the true constraint of vanishing shear strains in a realistic manner.6.17) suggests that in the thin beam limit. In contrast.19) Thus.18c) We can see that Equation (7. which is `field-consistent' and two linear variations which are `field-inconsistent'.15b) We need to examine only the shear strain field that will become constrained (i. This will be interpolated as.16) U e (G ) = ò-l ò Gb 2 a2 + b1 + a3 x + b3 y l -t t ( )2 dx dy = (Gb 2 ) (4tl ) [(a 2 + b1 )2 + (a3l )2 3 + (b3t )2 3 ] (7. let us consider the use of the shear strain along y=0 as a measure of the averaged shear strain across a vertical section. (7. Therefore. the following constraints will emerge: a2 + b1 → 0 a3 → 0 b3 → 0 (7.17) that as l >>t . However.18c) impose constraints on terms from one field function alone in each case. i. along the length of the beam element. we would be using elements which are very long in the x direction. it contributes a finite energy in Equation (7.18a) (7. We see now that the variation along the beam length is the critical term.18b) (7.17) and in the modeling 89 .y +v.e.18b) and (7.17) Equation (7. As in the linear beam element in Section 4. Equations (7. which is related to the spurious constraint in Equation (7. the constraint a3→0 will be enforced more rapidly than the constraint b3→0 as the beam becomes thinner. 2 2 One can then expect from Equation (7. and a linear oscillating term along the length.u = a0 + a1x + a2 y + a3 xy v = b0 + b1x + b2 y + b3 xy (7. this oscillation is selfequilibrating and does not contribute to the force equilibrium over the element.15a) (7.1. We shall now see how these lead to the stiffening effect that is called parasitic shear.18b). This gives γ = (a2 + b1 ) + a3 x (7. These are undesirable constraints. The spurious energies generated from these two terms will also be in a similar proportion. one can interpret the shear strain field to comprise a constant value.18a) is a constraint that comprises constants from both u and v functions. In Equation (7.e. γ have a constant term that reflects the averaged shear strain at the centroid. elements with l>t would be used. In the plane stress modeling of a slender beam. become vanishingly small) while modeling the flexural action of a thin beam.

20) parameter that This factor. The field-consistent and field-inconsistent terms will differ by. 90 . we saw that in the plane stress modeling of thin beam flexure. Earlier. the more severely constrained the shear strain becomes. as in a very slender beam. the value from a solution using energies based on inconsistent terms.e. i. a3 a3 = 1 + Gl 2 Et 2 ( ) stiffening (7.21) The oscillations are proportional to the centroidal moment. We shall denote by a3 . The need for field-consistency becomes important only when a strain field is constrained. there are shear force oscillations. they become more severe for smaller lengths of the plane stress element. e L = 1 + Gl2 Et2 is the additional determines the extent of parasitic shear. and for e L >> 1 . Of the many strategies that are available for removing the spurious constraints on the shear strain field. we can show that the shear strain-field for the 4-node rectangular plane stress element will be.16). the coefficient determined in a solution of the beam problem by using a field-consistent representation of the total energy in the beam. the use of a variationally correct re-distribution is recommended. V = V0 + (3M 0 l ) (x l ) (7. 7. this spurious energy is so large as to completely dominate the model of the beam behavior and cause a locking effect. the shear strain field is constrained and that these constraints are enforced by a penalty multiplier Gl2 Et2 .22) Thus instead of Equation (7.of very slender beams. It is important to note that a field-inconsistent two node beam element based on linear interpolations for w and θ (see Equation (7.3 The field-consistent elements It is well known that the plane stress elements behave reliably in general 2D applications where no constraints are imposed on the normal or shear strain fields. This requires the field-consistent γ to be determined from γ using the orthogonality condition ò δγ T (γ − γ ) dx dy = 0 (7. We shall now use the functional re-constitution technique to determine the magnitude of locking and the extent of the shear stress oscillations triggered off by the inconsistency in the shear strain field. the oscillations are inversely proportional to the element length. and by a3 . We shall simplify the present example and derive the error estimates from the functional re-constitution exercise.19)) produces an identical stress oscillation. The larger this term is. ( ) We can relate a3 to the physical parameters relevant to the problem of a beam in flexure to show that if e L >> 1 .3.

2×2×2 Gaussian integration). i. 7. However. These a priori estimates have been confirmed through numerical experiments. Although this element performs very well in general 3D stress analysis. Solid or three dimensional elements are needed to carry out the finite element modeling of such cases. 7.e. with strain-displacement matrix. It is possible however to improve the hexahedral 8-node and 27-node elements so that they are free of locking and of other ill-effects like stress oscillations. The element is unable to represent the shear strains in a physically meaningful form when the shear strains are to be constrained. we have dealt with the finite element modeling of problems which were simplified to one-dimensional or two-dimensional descriptions. It would appear now that the stiffness matrix for the element can be derived very simply by carrying out the usual finite element operations.5 Solid element model of beam/plate bending So far. it failed to represent cases of pure bending (parasitic shear) and cases of near incompressibility (incompressible locking). The phenomenon is the 3D analogue of the problem we noticed for 91 . These options can be built into a shape function sub-routine package in a simple manner so that where field-consistency requirements are paramount. triangular prism or hexahedral elements. A field-consistent re-distribution strategy allows these elements to be free of locking and spurious stress-oscillations in modeling flexure. In general cases of 3D stress analysis. tetrahedral.3. the appropriate smoothed shape functions will be used.γ = (a2 + b1 ) (7. In fact. In a very similar fashion. in such models. parasitic shear and shear locking merge indistinguishably into each other.3. stress-strain matrix and numerically exact integration (i. There remains a large class of problems which need to be addressed directly as three dimensional states of stress. no problems are encountered with the use of any of these elements as long as a sufficiently large number of elements are used and no constrained media limits are approached.23) It is seen that the same result is achieved if reduced integration using a 1 Point Gaussian integration is applied to shear strain energy evaluation. the consistent shear strain field can be derived for the quadratic elements.g. as in a case of pure bending.4 Concluding remarks on the rectangular plane stress elements The field-consistency principle and the functional re-constitution technique can be applied to make accurate error analyses of conventional 4-node and 8-node plane stress elements.e. One problem encountered is that of parasitic shear or shear locking when solid elements are used to model regions where bending action is predominant. under such limits the tetrahedral and triangular prism elements cannot be easily modified to avoid these difficulties. e. A variety of 3D solid elements exist. We briefly examine this below. The 8-noded brick element is based on the standard tri-linear interpolation functions and is the three dimensional equivalent of the bi-linear plane stress element.

w with respect to the natural co-ordinates ξ. for γxy we have. From Equations (7.25) is constrained. a8→0 and b8→0. These error models have been convincingly verified by numerical computations [7. γ xz and γ yz . we shall restrict our analysis to a rectangular prismatic element so that the (x.24c) where a1 to a8 and b1 to b8 are related to the nodal degrees of freedom u1 to u8 and v1 to v8 respectively. these are to be computed from terms involving the derivatives of u. u = a1 + a2 x + a3 y + a4z + a5 xy + a6 yz + a7 xz + a8 xyz v = b1 + b2 x + b3 y + b4z + b5 xy + b6 yz + b7 xz + b8 xyz w = c1 + c2 x + c3 y + c4z + c5 xy + c6 yz + c7 xz + c8 xyz (7. To facilitate understanding. We can consider the tri-linear interpolations to be expanded in the following form. γ xy = (a3 + b2 ) + (a6 + b7 )z (7. but have been achieved using a judicious mix of a priori and a posteriori knowledge of zero strain and stress conditions to simplify the problem. b5→0.e. Therefore attention is confined to a regular hexahedron so that consistency requirements can be easily examined. ζ) systems can be used interchangeably. y. to assure the consistency of the interpolations for the Cartesian shear strains in a distorted hexahedral. A simple way to achieve this is to retain only the consistent terms from the original interpolations. addition bubble functions and assumed strain hybrid formulations.the bi-linear plane stress element above.1].24b) we have.24b) (7. ζ and the terms from the inverse of the three dimensional Jacobian matrix.24a) and (7. These error models are now much more difficult to construct. Thus. η. A field-consistent representation will be one that ensures that the interpolations for the shear fields will contain only the consistent terms. i. In the isoparametric formulation. v. It is obvious that it will be a very difficult. if not impossible task. The problem of designing a useful node brick element has therefore received much attention and typical attempts alleviate this have included techniques such as reduced integration.26) 92 . 8to of at We can observe here that the shear strain energy is computed from Cartesian shear strains γ xy . the conditions appear: a5→0. z) and (ξ.25) constraint Our arguments projecting these as the cause of parasitic shear would be proved beyond doubt if we can derive a suitable error model. We shall now consider a case of a brick element undergoing a pure bending response requiring a constrained strain field γ xy = u. We shall now look this from the point of view of consistency of shear strain interpolation.24a) (7. η. following spurious (7.x → 0 . γ xy = (a3 + b2 ) + (a6 + b7 )z + a5 x + b5 y + a8 xz + b8 yz When Equation (7.y +v .

incompressible fluids or in plasticity.an incompressible hollow sphere It is instructive to demonstrate how field-consistency operates in the finite element approximation of a problem in incompressible (or nearly-incompressible) elasticity by taking the simplest one-dimensional case possible. A simple problem of a pressurized elastic hollow sphere serves to illustrate this.g. Such situations can arise in modeling of material such as solid rocket propellants. If the shear modulus G is 1.5. Displacement fields lock and highly oscillatory stresses are seen . i.Similarly. 7. the radial displacement field for this case is given by. saturated cohesive soils.drawing an analogy with shear and membrane locking. é 1 (1 − 2ν ) r ù u=ρ ê + ú 2 2 (1 + ν ) û ë 4r where ρ = Pba3 G b 3 − a3 ratio.5) is easily identifiable as enforcing incompressibility. plastics.e.r and the volumetric strain is εθ = u r εφ = u r ε = ε r + ε θ + ε φ = u.r + 2u r The elastic energy stored in the sphere (when G=1) is 1 U = a ò (ε 2 r 2 2 + ε θ + ε φ + 1 2 α ε 2 r 2 dr ) (7. e. These strain fields should now be able to respond to bending modes in all three planes without locking.4 Incompressible locking Conventional displacement formulations of 2D plane strain and 3D elasticity fail when Poisson's ratio ν→0. In fact.1 A simple one-dimensional case . it is easy to argue that a lack of consistency in the definition of the volumetric strain will cause locking and will appear as oscillatory mean stresses or pressures.27) [( )] and r=R/b non-dimensional radius and ν is the Poisson's A finite element approximation approaches the problem from the minimum total potential principle. a penalty term (bulk modulus→∞ as ν→0.4.28) 93 .0. (7. we can describe the phenomenon as incompressible locking. Thus. as the material becomes incompressible. Since this multiplies the volumetric strain. the other shear strain fields can be modified. Consider an elastic hollow sphere of outer radius b and inner radius a with an internal pressure P. elastomers and rubber like materials and in materials that flow. a radial displacement u leads to the following strains: ε r = u. 7.

r +2u ) that will determine whether there will be locking when α→∞.31) Consider now the constraints that are enforced when this discretized field is used to develop the strain energy arising from the volumetric strain as shown in Equation (7.5 (r2 − r1 )ξ u = 0.. this means that ε→0.r +2u ) 2 dr (7. Thus.5.32a) (7.30) It is the discretized representation of this term (ru. (7.and α = 2ν (1 − 2ν ) when G=1 and a = a b is the dimensionless inner radius.4. The displacement type finite element representation of this very simple problem runs into a lot of difficulties when ν→0.29a) (7.e. when α→∞ in the functional represented by Equation (7. However.2 Formulation of a linear element The displacement field u and the radial distance r are interpolated by linear isoparametric functions as r = 0.28).r +2u ) = ê é (r1 + r2 ) ù (u2 − u1 ) + (u1 + u2 )ú + [3 2 (u2 − u1 )] ξ 2 (r2 − r1 ) ë û (7. unlike in previous sections. In a physical situation.31) when α→∞. To see this. (ru.5 (u2 − u1 )ξ where the nodes of the element are at r1 and r2. let us examine what will happen when a linear element is used to model the problem. It is therefore meaningful to define this as a pseudo-strain quantity γ=rε. i. when a finite element discretization is introduced. To see how this is inconsistently represented.28).29b) The troublesome term in the potential energy functional is the volumetric strain and this contribution can be written as. These arguments now point to the volumetric strain field in Equation (7. 7.5 (r1 + r2 ) + 0. When α→∞.5 (u1 + u2 ) + 0. (r1 + r2 ) ( u2 − u1 ) + (u1 + u2 ) → 0 2 (r2 − r1 ) (u2 − u1 ) → 0 (7. this does not happen so simply. and in an infinitesimal interpretation of Equation (7. the constraint is not on the volumetric strain ε but on rε because of the use of spherical coordinates to integrate the total strain energy.32b) 94 . òε 2 2 r dr = ò (ru.28) as the troublesome term. let us now write the interpolation for this derived kinematically from the displacement fields as. this strain must vanish.

Consider the fields given in Equations (7. (7.36c) (7.Condition (7.3 Incompressible 3D elasticity We now go directly to modeling with 3D elasticity.2ν )] . it is convenient to describe the elasticity matrix in terms of the shear modulus.36b) (7. ε n = u. εd is the distortional strain and εn is the volumetric strain. This is therefore the spurious constraint that leads to locking and spurious pressure oscillations. We re-write the strain energy for 3D elasticity in the following form: U =1 2G ò ε d D ε d dV + 1 2 λ ò ε n ε n dV T T (7.36e) (7.34) we have.r → 0 . From Equations (7. y +w.32b) leads to an undesired restriction on the u field as it implies u.x +v .33) In Equation (7. G and the bulk modulus K.37) nearly 95 . 7.36g) Equation (7.35) can be constrained to zero only when the coefficients of each of its terms vanish.36a) (7.36b) to (7.36d) (7.35) Equation (7.32a) represents a meaningful discretized version of the condition of vanishing volumetric strain and is called a true constraint in our field-consistency terminology. ε n = (a2 + b3 + c4 ) + (b5 + c7 )x + (a5 + c6 )y + (a7 + b6 )z + (a8 yz + bz xz + c8 xy ) (7. only the consistent terms should be retained.36) are the inconsistent terms that cause locking.36f) (7.4. In this form. giving ε n = a2 + b3 + c4 This field will now be able to respond to the incompressible or incompressible strain states and the element should be free of locking.24) and (7. K = E [3 (1 . where G = E [2 (1 + ν )] .24) for the 8-noded brick element. giving rise to the constraint conditions. To ensure a field-consistent representation. a2 + b3 + c4 → 0 b5 + c7 → 0 a5 + c6 → 0 a7 + b6 → 0 a8 → 0 b8 → 0 c8 → 0 (7.34) where εn is the volumetric strain. condition (7. An elastic response for incompressible or nearly-incompressible materials will require a constrained strain field.33).2ν )] and λ = (K − 2G 3 ) = E ν [(1 + ν ) (1 .z → 0 (7. However.

4 Concluding remarks It is clear from this section that incompressible locking is analogous to shear locking. Dordrecht. 7.5 References 7. 1993.7.4. etc. The Finite Element Academic Press. Method in Structural Mechanics. Prathap.1 G. A robust formulation must therefore ensure consistent representations of volumetric strain in cases where incompressibility or near incompressibility is expected. Kluwer 96 .

we use a quadratic tapered bar element so that the extraneous oscillations. Again. This allows the constraints that emerge after discretization to remain physically faithful to the continuum problem. which 97 .2. This is a familiar problem in finite element thermal stress analysis. otherwise. To obtain accurate solutions at reasonable levels of discretization. However.2 Variable moduli problems 8. In initial strain problems. In structural regions with varying rigidity the spatial variation of strain-fields and stress or stress-resultant fields will not match. Such situations develop where there are structural regions in which the rigidity varies spatially due to varying elastic moduli or cross-sectional area and also in initial strain problems. 8. Here. the initial strain variation and the total strain variations may be of different order. This is necessary so that a correct interpretation of computed stresses and stress resultants is possible. This was the rule that guided the locking-free design of all elements discussed so far.e. one must carefully identify the conflicting requirements on the order of discretised functions to be used. The criterion governing the relationship between the various terms in the modified strain field interpolation was described as consistency .i.1 A tapered bar element We shall now conduct a simple numerical experiment with a tapered bar element to show that the force field computed directly from strains in a structural element of varying sectional rigidities has extraneous oscillations. we take a look at a class of problems where no constraints are imposed on the strains but there is a need to relax the satisfaction of the constitutive relationship linking discretised stress to discretised strain so that again a certain degree of consistency is maintained. it was necessary to modify these strain fields and use these in computing the stiffness matrix and also in stress recovery. the strain field interpolations must maintain a consistent balance internally of its contributing terms. it is necessary to obtain kinematically equivalent thermal nodal forces from temperature fields and also ensure that the discretised thermal strains corresponding to this are consistent with the discretised description of total strains. of which the thermal strain problem is the most commonly encountered. In this chapter. oscillations will be seen.1 Introduction In Chapters 6 to 7 we examined the difficulties experienced by the displacement approach to the finite element formulation of problems in which some strain fields are constrained. it is necessary to consider a form of external consistency requirement between the discretized strain fields and the discretized stress or stress-resultant fields. In the discretization of such cases.Chapter 8 Stress consistency 8. A linear element would have sufficed to demonstrate the basic principles involved.

We shall consider an isoparametric formulation for a quadratic bar element of length 2l with mid-node exactly at the mid-point of the element. in a displacement type finite element formulation. will not contribute to the energy and therefore will not provide terms to the stiffness matrix! It is clear that. Due to the orthogonal nature of the Legendre polynomials. Young's modulus of elasticity After discretization. as sampling at the centroid of the element gives the correct stress resultant. the kinematically constituted axial force. N should comprise only the terms that will contribute to the stiffness and strain energy and the simplest way to do this is to expand the kinematically determined N in terms of Legendre polynomials. Next.for a general case can be of cubic form. In a linear element. Then the interpolations for the axial displacement u and the cross sectional area A in terms of their respective nodal values are. T 98 . N3 and N4 respectively. the stresses recovered from the displacement vector will have extraneous oscillations if N3 and N4 are not eliminated from the stress field during stress recovery. U =1 2 òN T ε dx From this product the terms of the stiffness matrix emerge. the conventional element using N for stiffness matrix evaluation and recovery of force resultant. we must see how the consistent representation of the force field denoted by N must be made. this exercise becomes very trivial. the potential of external forces. We start with a functional written as. The strain energy of deformation is then expressed as. We shall designate by BAR3. and retain only terms that will meaningfully contribute to the energy in (N ε ) . π = 1 2 ò N T ε dx − W where.0. ε = du dx N = E A (x ) ε W = ò pu dx p E the axial strain. the displacements recovered from such a formulation cannot recognize the presence of the quadratic and cubic terms N3 and N4 in the stress field N as these have not been accounted for when the stiffness matrix was computed. Hence. are not only vividly seen but also need special care to be reduced to its consistent form. terms from N which are linked with the quadratic and cubic Legendre polynomials. distributed axial load. ε will be a linear function of ξ but N will be a cubic function of ξ. u = u2 + (u3 − u1 ) 2 ξ + (u1 − 2u2 + u3 ) 2 ξ 2 A = A 2 + (A 3 − A1 ) 2 ξ + (A1 − 2A 2 + A3 ) 2 ξ2 We shall first examine how the element stiffness is formulated when the minimum total potential principle is used.

The taper is defined by the parameters. To see the variational basis for the procedure adopted so far. Fig. retain only up to linear N = N 1 + N 2ξ (8.0. α = (A 3 − A1 ) 2A 2 and β = (A1 − 2A 2 + A 3 ) 2A 2 Finite element results from a computational exercise using the two versions described above for a bar with cross section tapering linearly from the root to the tip for which β=0 are obtained. the stiffness matrices and computed displacements are identical.01 A1) It can be noted here that the results are accurate at ξ = ±1 3. note that in both cases.2 shows the axial force patterns obtained from the finite element digital computation for a case with α=-0. 8. N terms: must be consistent with ε. Thus α is given by.2 Numerical experiments We shall perform the computational exercises with the two versions of the element. the problem is re-formulated according to the Hu-Washizu principle which allows independent fields for assumed strain and assumed stress functions.1.1 shows a tapered bar clamped at node-1 and subjected to an axial force P at node-3. Fig. 8.e. It uses N for stiffness matrix evaluation and recovery of forces resultant.1 A cantilever bar modeled with a single element. α = (A 3 − A1 ) (A1 + A3 ) Thus α can vary from 0 to –1. in this case.2. i. 8.9802 (with A3=0.1) Such an element is denoted by BAR3. 99 . Figure 8.Thus.

0.0) with A3=0.3 Reconstitution of the stress-resultant field using the Hu-Washizu principle In forming the Hu-Washizu functional for the total potential. 8. Fig.2.Fig. 8. 8. and the area ratios are α=-4/3 and β=4/9.2 Axial force pattern for linearly tapered bar (α=0.3 shows the results from the finite element computations. an as yet undetermined assumed force function N is introduced but the assumed strain field ε can be safely retained as ε (note that in a constrained media problem it will be required to introduce a field-consistent ε that will be different from the kinematically derived and therefore usually field-inconsistent ε) . there are no points which can be easily identified for accurate force recovery! Therefore it is necessary to perform a re-constitution of the force resultant fields on a consistency basis using the orthogonality principle as done here before reliable force recovery can be made.9802 and β=0. A general case of taper is examined next where A1=1.01 A1. Due to the presence of both quadratic and cubic oscillations.the functional now becomes π = ò {1 2 (EAε ) T ε + N T (ε − ε ) dx − W } 100 . A2=0.36 and A3=0.04.

T ò δu {dN dx − p} dx = 0 Variation with respect constitutive relation to the assumed strain field ε gives rise to a ò δε T {.3) are the orthogonality conditions required for reconstituting the assumed fields for the stress resultant and the strain.3) and variation with respect to the assumed force field condition ò δN T {ε . A variation of the Hu-Washizu energy functional with respect to the kinematically admissible degree of freedom u.N + EA ε } dx = 0 N (8.ε } dx = 0 Now Equations (8. 8.2) and (8.2) gives rise to the (8.3 Axial force pattern for bar with combined linear and quadratic taper (α=-4/3 and β=4/9). The consistency paradigm suggests that the assumed stress resultant field N 101 . gives the equilibrium equation.Fig.

Earlier in this chapter. the stress to strain relationship has to be written as 102 . we show that much the same behavior carries over to the problem of thermal stress computations. This was traced to the fact that these stress resultant fields were of higher interpolation order than the strain fields and that the higher degree stress resultant terms did not participate in the stiffness matrix computations.N } dx = 0 (8. we saw that stress resultant fields computed from strain fields in a displacement type finite element description of a domain with varying sectional rigidities showed extraneous oscillations. the thermal stresses should also be based on these Gauss point values.3. Thus the procedure adopted in the previous section has variational legitimacy. ò δε T {N .2) can be used to from the kinematically reconstitute the assumed stress-resultant field N derived field N. 8.3 Initial strain/stress problems Finite element thermal stress analysis requires the formulation of what is called an initial strain problem. 8. orthogonality condition (8. The usual practice in general purpose codes is to use the same shape functions to interpolate the temperature fields and the displacement fields. This strategy emerged from the understanding that the unreliable stress predictions originate from the mismatch between the element strain ε and the initial strain due to temperature ε0. Now. if N is expanded in terms of Legendre polynomials.4) is obtained very simply by retaining all the Legendre polynomial terms that are consistent with ε . In this section. as shown in Equation (8.2) can be written as. Thermal stresses computed directly from stress-strain and strain-displacement matrices after the finite element analysis is performed thus show large oscillating errors.1 Description of the problem With the introduction of initial strains ε0 due to thermal loading. The temperature fields which are imposed must be converted to discretised thermal (initial) strains and from this kinematically equivalent thermal nodal forces must be computed.4) Thus. Some useful rules that are adopted to overcome this difficulty are that the temperature field used for thermal stress analysis should have the same consistency as the element strain fields and that if element stresses are based on Gauss points. Equation (8.3) gives the orthogonality condition for strain field-redistribution.should be of the same order as the assumed strain field ε . This can be traced to the fact that the total strains (which are derived from displacement fields) are one order higher than the thermal strains (derived from temperature fields).e.1). On the other hand. We shall now show that this is due to the lack of consistency of their respective interpolations within the element. Then Equation (8. it can be proved that N which is consistent and orthogonally satisfies Equation (8. i.

The calculation of the total strains {ε} involves the differentiation of the displacement fields and the strain field functions will therefore be of lower order than the shape functions. ò δ {ε } T D {ε0 } dV (8. {ε} is the total strain and {εm} is the mechanical or elastic strain. a thermal load vector is created which corresponds to a initial strain (and stress) vector that is `consistent' with the total strain vector.σ = D (ε . In a finite element description.2 The linear bar thermal load vector element - Derivation of stiffness matrix and Consider a linear bar element of length 2l. the total strain ε. the initial strain ε0 and stress σ are interpolated as follows: u = (u1 + u2 ) 2 + ξ (u2 − u1 ) 2 (8.6) where T is the temperature relative to a reference value at which the body is free of stress and α is the coefficient of thermal expansion. The free expansion of material produces initial strains ε0 = α T (8. It is this lack of consistency that leads to the difficulties seen in thermal stress prediction. the problem is that the `higher order' components of the thermal (or initial) stress vector are not sensed by the total strain interpolations in the integral shown above. Thus.9a) 103 . the displacements and temperatures are interpolated within the domain of the element using the same interpolation functions. The axial displacement u. The initial strain fields (see Equation (8. the inclusion of the inconsistent part.7) where {d} is the vector of nodal displacements. 8. the total strain terms "do work" only on the consistent part of the thermal stress terms in the energy or virtual work integral. as was done earlier. We demonstrate these concepts using a simple bar element.8) Again.ε0 ) = Dε m (8. results in thermal stress oscillations.e.6)) involve the temperature fields directly and this is seen to result in an interpolation field based on the full shape functions. The initial strain matrix is of higher degree of approximation than the kinematically derived strain fields if the temperature fields can vary significantly over the domain and are interpolated by the same isoparametric functions as the displacement fields. {ε}=[B]{d} (8.5) The strain terms now need to be carefully identified.3. In other words. The finite element displacement and total strain fields which are obtained in the finite element computation then reflect only this consistent part of the thermal loading. The total strains (i. only the consistent part of the thermal stress should be computed when stress recovery is made from the nodal displacements. This originates from the fact that the thermal load vector is derived from a part of the functional of the form. the kinematically derived strains) are defined by the strain-displacement matrix. Therefore.

9b) to (8. ε = u. Thus the equilibrium equations that result from the assembly of the element equilibrium equations represented by Equation (8. A the area of cross-section of the bar and α the coefficient of expansion.e.4). constant and linear terms) and tracing the way they participate in the `work' integral in Equation (8.9c) (8.11) where F1 and F2 are the consistently distributed nodal loads arising from the distributed external loading. To form the element stiffness and thermal load vector. EA 2l é 1 ê ë− 1 − 1ù ú 1û ìu1 ü (T1 + T2 ) ì − 1ü ìF1 ü í ý=í ý í ý − EAα u2 þ 2 î 1 þ îF2 þ î (8.x = (u2 − u1 ) 2l ε 0 = α {(T1 + T2 ) 2 + ξ (T2 − T1 ) 2} σ = E (u2 − u1 ) 2l − Eα {(T1 + T2 ) 2 + ξ (T2 − T1 ) 2} where (8. see Equation (8.9b) (8. 8.11).10). an integral of the form ò δε T σ dx (8.8. ξ) term in σ (originating from ε0) cannot do work on the constant term in δεT and therefore vanishes from the thermal load vector.9d) ξ =x l (see Fig. E is the modulus of elasticity. This leads to a matrix equation for each element of the form. only the consistent part of these fields should be used.9d) carefully (i. We shall work these out by 104 . The use of the original initial strain or stress fields will result in oscillations corresponding to the inconsistent part. If these computed displacements are to be used to recover the initial strains or thermal stresses. it is clear that the (T2 − T1 ) 2 term associated with the linear (i. By observing the components of the interpolation fields in Equations (8.Fig.10) is to be evaluated.4 Linear bar element.e.11) will only respond to the consistent part of the initial strain and will give displacements corresponding to this part only.

é 1 −1 EA ê −1 2 2l ê ê −1 ë ù ú −1 ú 1ú û ì 0 ü ìF1 ü ï ï ï ï EAα íu2 ý = í 0 ý + 2 ï 0 ï ïF ï î þ î 3þ ì − (T1 + T2 )ü ï ï í (T1 − T3 ) ý ï (T + T )ï 2 3 þ î (8. The nodal reactions are F1 and F3 (corresponding to the clamped conditions u1=u3=0). Problems however appear when the nodal displacement computed from Equations (8.13) and these are the correct answers one can expect with such an idealization. one gets a constant stress σ = −Eα (T1 + 2T2 + T3 ) 4 . we can compute the displacements and nodal reactions as. in both elements. which is again the correct answer one can expect for this idealization.5 Clamped bar subject to varying temperature hand using a simple example below and compare it with the analytical solution. so that we require the nodal temperatures T1.12) From this.14a) (8. T2 and T3 as input.9d) to compute the initial strains and stresses in each element. u2 = αl (T1 − T3 ) 2 and F1 = −F3 = EAα (T1 + 2T2 + T3 ) 4 (8. We would obtain now.Fig. the stresses as (subscripts 12 and 23 denote the two elements) σ 12 = −Eα (T1 + 2T2 + T3 ) 4 − ξ Eα (T2 − T1 ) 2 σ 23 = −Eα (T1 + 2T2 + T3 ) 4 − ξ Eα (T3 − T2 ) 2 (8. 8.retain only that part of the stress field that will do work on the strain term in the 105 . This offers us a straightforward definition of what consistency is in this problem . 8.9b) to (8. If the stress in the bar is computed from the nodal reactions. We shall consider a case where two conventionally derived linear elements are used. which the above model recovers exactly. It is a very trivial exercise to show analytically that a problem in which the temperature varies linearly from 0 to T at both ends will give a constant stress field σ = − EαT 2 . We have the assembled equations as.5 shows a bar of length L=4l clamped at both ends and subjected to a varying temperature field.3 Example problem Figure 8.3.14b) It is now very clear that a linear oscillation is introduced into each element and this corresponds to that inconsistent part of the initial strain or stress interpolation which was not sensed by the total strain term.12) is used in Equations (8.

3. we must proceed to the Hu-Washizu theorem. It is therefore necessary to reconstitute the discretized stress field into a consistent stress field σ without violating any variational norm.e. σ. ( ) 3 .7) and the displacement and strain fields are interpolated from the nodal displacements using the element shape functions and their derivatives and P is the potential energy of the prescribed loads. thus thermal stresses derived using a formal theoretical basis will show these quadratic oscillations which will vanish to give correct answers at the points corresponding to ξ = ± 1 integration rule.5 Re-constitution of the thermal strain/stress field using the HuWashizu principle We now seek to find a variational basis for the use of the re-constituted consistent initial strain and stress interpolations in the Hu-Washizu principle.e. The minimum total potential principle states the present problem as. strain field ε and stress field σ (we project that the strain field ε is derived from the displacement field through the strain-displacement gradient operators 106 .e. introducing ε0 and σ . 8. To see how we progress from the inconsistent discretized domain (i. involving ε0 and σ) to the consistent discretized domain (i. is inconsistent to the extent that the initial strain field ε0 is not of the same order as the total strain field ε. the initial strain and thermal stress field will now have a quadratic representation (provided the temperature field has a quadratic variation) and the inconsistency will now be of the 1 − 3ξ 2 type. To see how this part can be derived in a variationally correct manner. we had seen a simple way in which this was effected. i. We now know that the discretized stress field thus derived. the total strains would have been interpolated to a linear order. it is again convenient to develop the theory from the generalized HuWashizu mixed theorem. find the minimum of the functional. We proceed thus: Let the continuum linear elastic problem have a discretized solution based on the minimum total potential principle described by the displacement field u.3. the points corresponding to the 2-point Gauss 8. Π MTP = ò {σ T ε m 2 + P dV } (8. In the examples above. We shall present the Hu-Washizu theorem from the point of view of the need to re-constitute the inconsistent ε0 to a consistent ε0 without violating any variational norms.functional.4 A note on the quadratic bar element We may note now that if a quadratic bar element had been the basis for the finite element idealization.15) where σ and εm are as defined in (8.5) to (8.

17) where ε m = (ε − ε0 ) . The three-field Hu-Washizu theorem can be stated as. is subject to variation. is to introduce a "dislocation potential" to augment the usual total potential. all three fields are subjected to variation and leads to three sets of equations which can be grouped and classified as follows: Variation on Nature Equilibrium Orthogonality (Compatibility) Equation. i. Note that we also argue that we can use ε = ε as there is no need to introduce such a distinction here (in a constrained media elasticity problem it is paramount that ε be derived as the consistent substitute for ε. At this stage we do not know what σ they are to be of consistent order with ε. u ∇ σ + terms from P=0 (8. following the interpretation given by de Veubeke. the displacement field u). However. σ = D (ε − ε0 ) . or ε0 are except that In the simpler minimum total potential principle.16) Π HW = ìσ T ε m 2 + σ í î ò T (ε m − ε m ) + P ü dV ý þ (8.e. of the theory of elasticity and that the stress field σ is derived from the strain field ε through the constitutive laws as shown in (8.2).are computed using the constitutive relationships.18b) σ ε0 òδσ ò δε0 T (ε0 − ε0 ) dV =0 T Orthogonality (Equilibrium) æ σ − σ ö dV = 0 ç ÷ è ø (8. Let us now replace the discretized domain by another discretized domain corresponding to the application of the Hu-Washizu principle and describe the computed state to be defined by the quantities ε.18c) 107 . which is the basis for the derivation of the displacement type finite element formulation in most textbooks.) fields σ What the Hu-Washizu theorem does. It is clear that ε0 is an approximation of the strain field ε0.e. ε0 and σ . This dislocation potential is based on a third independent stress field σ which can be considered to be the Lagrange multiplier removing the lack of compatibility appearing between the kinematically derived strain field ε0 and the independent strain field ε0 . only one field (i. we take that the stress δΠ HW = 0 where (8.18a) (8. in this more general three field approach. where again.

18c). (c) Plane stress model. In this instance this condition reduces to σ = σ . this condition shows us how to smooth ε0 to ε0 to maintain the same level of consistency as ε. we now see how the consistent initial strain field ε0 can be derived from the inconsistent initial strain field ε0 without violating any variational norm. Note that in a problem where the rigidity modulus D can vary significantly over the element volume. as done in the previous section for the consistent stress-resultants.for the bar element above. 108 . Therefore. σ to be reconstituted from σ in a The orthogonality condition in (8. This equation therefore provides the variationally correct rule or procedure to determine consistent thermal strains and stresses from the inconsistent definitions. Let us first examine the orthogonality condition in (8.18b) is now very easily interpreted. We can interpret this as a variational condition to restore the equilibrium imbalance between σ and σ . Since we have σ = σ consistent with ε. We thus have a variationally correct procedure for reconstituting the consistent initial strain field . 8. this can be very trivially done by using an expansion of the strain fields in terms of the orthogonal Legendre polynomials. this condition allows consistent way.6 (a) Clamped bar under linear temperature variation and its (b) Bar element model.Fig.

1 represent the bar element versions with inconsistent and consistent thermal stress fields. depth 2 units subjected to a linear temperature distribution as shown in Fig 8.6b).7 Consistent and inconsistent thermal stress recovery for a clamped bar problem. Fig.0 shows linear oscillations (see Equations (8. As already predicted.13)). BAR.1 gives exact stress throughout while BAR.6 Numerical examples In this section we take up the same example of a bar with fixed ends subjected to linearly varying temperature along its axis and model it with the bar element and a plane stress element for demonstrating the extraneous stress oscillations resulting from the lack of consistency in initial strain definition.002 units at the mid point). Let BAR. 109 . see Equation (9.3.14b)) as shown in Fig 8. 18. A bar of length 8 units.8.0 and BAR. both give accurate displacements (-0.8.14a) and (8.7.6a is modeled first with two bar elements (see Fig 4.

which again lies outside the scope of the variational correctness paradigm. Note that the consistency paradigm. Similarly. It is shown that such stresses must be computed in a consistent way. Note that Equation (9. it is necessary to derive a consistent initial strain-field for purposes of recovery of stresses from the computed displacement field.7 Concluding remarks Before we close this section. the potential energy formulation and the Hu-Washizu formulation describe the continuum physics in exactly the same way. This would result in the Hu-Washizu formulation yielding the same stress predictions as the minimum potential principle. The stress field-consistency paradigm introduced here also applies to finite element applications where certain terms in either the stress or strain fields do not participate in the displacement recovery due to their inconsistent representation e. lies clearly outside the variational correctness paradigm. in the earlier chapters of this book. problems with varying material properties within an element etc. It is very important to understand therefore that both consistency and correctness paradigms are mutually exclusive but are needed together to ensure that the finite element formulations are correctly and consistently done. when approximations are introduced in the finite element method.3. we saw another variation of the consistency paradigm . Thus.2) and (9. In the next section.g. it will be worthwhile to discuss the "correctness" of the various approaches from the variational point of view. It follows that from the theoretical point of view of the variational or virtual work methods. the Hu-Washizu formulation gives decidedly better results than the potential energy formulation because of the flexibility it provides to modify stress fields and strain fields to satisfy the consistency requirements in situations where they play a prominent role.in applications to evaluation of stresses and stress resultants in finite element thermal stress computations.3) would have been fulfilled if an assumed stress-resultant field of the same order as N had been used in place of the N of consistent order. but only N is "correct" with the minimum potential energy formulation. initial stress/strain problems. Thus. both N and N are equally "correct" in the Hu-Washizu sense. we shall extend this paradigm to extract consistent thermal stress and/or strains from a displacement type formulation. A simple and variationally correct way to do this has been established from the Hu-Washizu theorem. Where they differ is in the consistency aspect i.8. In this section.the need to ensure a proper balance in the discretized representation of constrained strain fields. N is consistent with ε whereas N isn't. 110 . in an element where the temperature field interpolations do not have the same consistency as the strain fields. However. recognizing that the displacement type finite element procedure can sense only stresses which have the same consistency as the total strain field interpolations used within each element. we have demonstrated another variation of the consistency paradigm . Both the potential energy and HuWashizu formulations are equally valid as far as variational correctness is concerned. in this case that used to justify the truncation of N (obtained directly from the displacement approach) to a consistent N on the argument that the truncated terms are not sensed in the discretized strain energy computations.e.

Chapter 9 Conclusion 9. etc.1 Continuity conflict and the consistency paradigm We saw in chapters 5 to 7 that in a class of problems where some strain fields need to be constrained. In fact. Two strain energy components are important.x . The FEM is also an excellent example of a body of knowledge that began as Art and Technology. This chapter will therefore sum up the ideas that constitute the science of the finite element method and also point to the future course of the technology. namely continuity and completeness. Let us recapitulate our fresh insight in an epistemological framework below. i. This was not so in the beginning.x and w. continuity of θ and w at nodes (or across edges in a 2D problem.2 The C-concepts: An epistemological summing-up One would ideally like to find a set of cardinal or first principles that govern the entire finite element discretisation procedure. it became necessary to treat the two energy components on separate terms. science came later.2. Science was identified more slowly. as there was Technology much before there was Science. long time was that the existence of the However.e. The simplistic understanding that prevailed for a θ . each fertilizing the other and providing an explosive synergy. ensuring the quality of accuracy and efficiency of the method. Let us again take up the simple problem of a Timoshenko beam element. provided a necessary and sufficient basis for the choice of trial functions to initiate the discretisation process. This required that the shear strains must vanish without 111 . it was believed that two rules. Chapter 3 of this book reviewed the understanding on these two aspects.) was necessary and sufficient. The shear strain energy was constrained in the regime of thin beam bending to become vanishingly small. this led to locking situations. At one time. To resolve this conflict.x and the shear strain energy which is based on the shear strain γ = θ − w. This was the locking problem which we examined carefully in chapters 5 to 7. the implementation of the continuity rules on the displacement trial functions blindly led to a conflict whereby excessive constraining of some strain terms caused a pathological condition called locking.x terms in the energy expression demanded that the trial functions selected for the θ and w fields 0 must be C continuous.1 Introduction Technology and Science go hand in hand now. finite element practitioners were to be confronted with a class of problems where the strict implementation of the continuity condition led to conflicts where multiple strain fields were required. 9. there was technology before there were humans and it was technology which enabled the human race to evolve. However. 9. the bending energy which is constituted from the bending strain as κ = θ .

It also became very apparent that if the internal working of the discretisation procedure is examined more carefully. Note that although the bending strain requires θ to be 0 0 C continuous.this is the consistency paradigm. being a single field variational principle. The idea of field consistent formulation was to anticipate these conditions on the constrained strain field. We can also interpret this as a multiple continuity requirement. such a conflict is resolved by operating with consistently represented constrained strain fields. it is phenomena which first confronts us and is systematically recognized and classified. we require θ not to have C -1 continuity.2 The correctness requirement The consistency requirement now demands that when trial functions are chosen for displacement fields. This brings us to the stress correspondence paradigm. 9. we can therefore have a choice from two courses of action: 112 .e. it turns out that it is strain and stress fields which are being approximated in a "best-fit" sense and not the displacement fields as was thought earlier.3 The correspondence principle The study so far with consistency and correctness showed the primacy of the HuWashizu theorem in explaining how the finite element method worked. as shown in chapter 2. we have come to understand that a multi-field variational principle like the generalized Hu-Washizu theorem provided exactly that flexibility to resolve the conflicting demands between consistency and continuity. often.2.x ) which need to be constrained must have a reconstituted consistent definition. then the logical which one teaches a subject will be to start with first principles the various facets of the phenomena from these principles.that the reconstitution of the strain field which needed a reduced-continuity satisfying displacement field from an inconsistent strain field kinematically derived from the original continuous displacement fields must be done strictly according to the orthogonality condition arising from the HW theorem.x to vanish without producing gratuitous constraints. we called the correctness requirement .3 How do we learn? How must we teach? In a chronological sense or in a logical sense? In an epistemological sense. The underlying first principles that explain the phenomena are identified very much later. i.producing spurious constraints. these fields (such as θ in γ = θ − w.2. 9. In practice. Thus for θ − w. does not provide a clue as to how this reconstitution of the strain fields can be performed without violating the variational rules. Thus. sequence in and derive pedagogical if and when the first principles are available. 9. This. if not at all. is. in the shear strain component. Thus. the trial functions used for θ and w in the shear strain field alone must be complete to exactly the same order . In the course of this book. and the discovery that this paradigm can be axiomatised from the Hu-Washizu theorem. The clear understanding of first principles can also lead to the unraveling of new phenomena that was often overlooked. say a C continuity. The minimum total potential principle.

Modern software packages. leading to the emergence of this new discipline we call computational structural mechanics (CSM).teach in a chronological sense. Very large structural calculations can be performed to account for complex geometry.1 sums up the first principles. chapter 12 of Reference 9. numerical analysis.the HW and correspondence rules are taken up first and then the various phenomena like locking. CSM has virtually grown out of the finite element method (FEM). explaining how the ideas unfolded themselves to us. functional analysis.4 Finite element technology: Motivations and future needs Advances in computer science and technology have had a profound influence on structural engineering. From these. In my first book [9. or teach in the logical sense. automotive.1]. reliability and economy.solid and structural mechanics. The use of these packages in the Computer-AidedDesign/Computer-Aided-Manufacturing cycle forms a key element in new manufacturing technologies such as the Flexible Manufacturing Systems.2]. I choose the chronological order.that the fem approach manipulates stresses and strains directly in a best-approximation sense. Already artificial intelligence in the form of knowledge based expert systems and expert advisers and optimization procedures are being coupled to FEM packages to reduce human intervention in structural design to a bare minimum. Such calculations are now routinely performed in aerospace. The conflicts created by the locking phenomena were resolved by inventing the consistency paradigm. CSM has brought together ideas and practices from several disciplines . Along with it a huge software industry has grown. These allow for unprecedented opportunities for increase in productivity and quality of engineering by automating the use of structural analysis techniques to check designs quickly for safety. integrity. In the present book. Algorithms for the use of the finite element method in a wide variety of structural and thermomechanical applications are now incorporated in powerful general purpose software packages. computer science. 9. It is not difficult to delineate many compelling reasons for the vigorous development of CSM [9. and approximation theory. I have preferred a logical sequence . called general purpose programmes. oil and nuclear industries. one could axiomatise the correspondence rule . These are: 113 . This led to the correctness principle and the recognition that the HuWashizu principle formed a coherent basis for understanding the fem procedure. civil engineering. etc. from first principles to observable facts. couple FEM software with powerful graphics software and the complete cycle of operations involving pre-processing (automatic description of geometry and subsequent subdivision of the structure) and post-processing (projecting derived information from FEM analysis on to the geometry for color coded displays to simplify interpretation and make decision making that much easier). loading history and material behavior. Thus. mechanical engineering. It is not clear that one is superior to the other.

in certain mission critical areas in space. thermal/control/structural coupling in space exploration.2 are: simulation of response of transportation vehicles to multidirectional crash impact forces. the space station).g. Computer simulation is often required to reduce the dependence on extensive and expensive testing. and numerical algorithms that exploit the capabilities of these new machines (e. such as stochastic mechanisms of fatigue. 2. c) study of structure/media interaction phenomena (e. it is necessary to develop new formulations. To realize this potential fully. damage tolerance of structural components made of new material systems). Emerging and future computer systems are expected to provide enormous power and potential to solve very large scale structural problems. dynamics of large flexible structures taking into account joint nonlinearities and nonproportional damping. and vehicle crash-worthiness). d) use of stochastic models to account for associated with loads. structural penetration. 3. their low natural frequencies. and material variability e) development of efficient high-frequency nonlinear dynamic modeling capabilities (with applications to impulsive loading. electromagnetic/thermal/structural coupling in microelectronic devices). the fundamental mechanics concepts are still being studied (e. study of thermoviscoelatic response of structural components used in advanced propulsion systems etc. large solar arrays. artificial intelligence). computer modeling may have to replace tests. such as: a) examination of more complex phenomena (e. f) improved representation of structural details such as damping and flexible hysteritic joints: g) development of reliable life-prediction methodology for structures made of new materials. Noor and Atluri [9. Expansion of the scope of engineering problems modeled.g.g. Some of these demand large computational power. high-energy impact. such as metalmatrix composites and high-temperature ceramic composites. 114 .g. adequate characterization of finite strain inelasticity is still needed). Thus. in metal forming. b) study of the mechanics of high-performance modern materials. In many structural problems.g. it may not be possible for ground-test technology in 1-g environment to permit confident testing in view of the large size of the structures. environment. parallelism. for large space structures (e. There are a large number of unsolved practical problems of current interest which still await experimental and/or numerical solutions. etc. large antennas.1.2] also expect that high-performance structures will demand the following technical breakthroughs: 1. computational strategies. Some of the examples described in reference 9. light weight and the presence of many joints. vectorization. hydrodynamic/structural coupling in deep sea mining. material/aerodynamic/structural coupling in composite wing design.

These can be easily performed by using predata and postdata processing packages using high resolution. Continued reduction of cost and/or engineering design/analysis problems. Distributed computing environment having high-performance computers for large scale calculations. to aid in the modeling of the structure. 2.g. 3. These include: 1.h) analysis and design of intelligent structures with active and/or passive adaptive control of dynamic deformations. Special and general purpose application software systems that have advanced modeling and analysis capabilities and are easy to learn and use. earthquake-resistant structures i) Computer simulation of manufacturing processes interface mechanics. Special hardware and software requirements must become available to meet the needs described above. Computerized symbolic manipulation capability calculations and increase their reliability. 115 . in flight vehicles. The accurate analysis of a complex structure requires the proper selection of mathematical and computational models. Three areas which are expected to receive increasing attention are: 1) modeling of complex structures. a wide range of intelligent engineering workstations for interactive user interface/control and moderate scale calculations. high speed long distance communication etc. Complex structures will require an enormous amount of data to be prepared. time for obtaining solutions to 3. Development of practical measures for assessing the reliability of the computational models and estimating the errors in the predictions of the major response quantities. superplastic forming.5 Future directions It is expected that CSM will continue to grow in importance. such as solidification. 2. incorporating the experience and expertise of practitioners. to automate analytic 5. 4. The generation of three dimensional color movies can help to visualize the dynamic behavior of complex structural systems. the adaptive refinement of the model and the selection of the appropriate algorithm and procedure used in the solution. large space structures. Artificial intelligence-based expert systems. 2) predata and postdata processing and 3) integration of analysis programs into CAD/CAM systems. 9. User-friendly engineering workstations with high-resolution and high speed graphics. e. high throughput graphic devices. There is therefore a need to develop automatic model generation facilities.

7 References 9.2 A. 9.6 Concluding remarks In this concluding chapter. Dordrecht. Advances and trends in computational structural mechanics.1 G.9. 116 . N. Prathap. K. It has the potential not only to be a tool for research into the basic behavior of materials and structural systems but also as a means for design of engineering structures. we have summarized the ideas that should inform the rational development of robust finite elements and we have also described the current trends in the use of structural modeling and analysis software in engineering design. 9. 977-995 (1987). Kluwer Academic Press. It has been seen that CSM has greatly improved our capabilities for accurate structural modeling of complex systems. The Finite Element Method in Structural Mechanics. Atluri. 25. 1993. Noor and S. AIAA J.

Sign up to vote on this title

UsefulNot useful- earthquake analysis [Compatibility Mode]
- Finite Element Analysis - Theory & Progr. - C.S. Krishnamoorthy
- FEA for Non Linear
- FEA-MECHANICAL 2010 Student Version
- AFEM.Ch27
- Ballistic Impact Simulation Using DEM
- Finite Element Methods for Fluids
- 184816579XBoundary
- Full M.E. Thesis
- Ls Dyna Theory Manual 2005 Beta
- MD Nastran R3 Quick Reference Guide
- Finite Element Analysis Thermomechanics of Solids Second Edition
- Structural Motion Control in MSC NASTRAN
- Weigted Residual Methods
- Tech Review
- FEM
- Introduction
- Introduction to the Finite Element Method in Electromagnetics
- Stiffness
- Standard Structural Designs for Low Rise Residential Buildings in RC Construction
- Crisfield M a Vol 2 Non-Linear Finite Element Analysis of Solids and Structures Advanced Topics
- Non-Linear Finite Element Analysis of Solids and Structures Vol.1 - M.a. Crisfield
- ansys
- Finite Element Mesh Generation
- Concepts of Computational Finite Elements and Methods of Static and Dynamic Analyses in MSC.NASTRAN and LS/DYNA
- Understanding and Implementing the Finite Elementh Method
- FEM Book of Gangan Prathap