Space Structures: Principles and Practice Subramanian Narayanan Chief Executive, Computer Design Consultants, Chennai, India. unrverstratea TeHNICA BIBLIOTECA CENTRALA ‘Uavenararea TEAMEA Do cLUsaAroCA | BIBLIOTECA CENTRALA©2006 ‘Subramanian Narayanzn ISBN 0 908822 420 Multi-Science Publishing Co. Ltd, '5 Wates Way, Brentwood, Essex CM15 9TB, UK This book is respectfully dedicated to my beloved guide Prof. Dr. C. Ganapathy and Prof. Dr. Z. S. Makowski from whom I learnt much about this exciting topicTable of Contents Foreword Preface Acknowledgement 1, Introduction... 1.1 General 1.2 Definition of Space Structures 1.3 Historical Developments ..... 1.4 Types of Space Structures 1.5 Materials Used in Space Frames 1.6 Advantages and Disadvantages -...... 1.7 Practical Difficulties feceeeses References 2. Single and multi-layer grids ...... 2.1 Introduction 2.2 Single Layer Grid Systems . 23 Double Layer Grids 2.4 Two-Way Space Grids 25 Three-Way Space Grids ...... 2.6 Double Layer Grids for Multistorey Buildines 2.7 Advantages of Double Layer Grids 28 Cladding 2.9 Water Drainage in Double Layer Grids 2.10 Triple Layer Grids : 2:11 Double Layer Grids for Walls 2.12 Multi-Layer Gridslable ot Contents 2.13 Progressive Collapse and d Composite Space Trusses References 78 3. Braced domes os... 0eeeeeeeeeeee 7 3.1 Introduction ...... 97 3.2 Materials for Braced Domes - 98 ‘3.3 Funetional Applications ot Domes 88 3.4 Basic Geometries of Domes 3.5 Types of Braced Domes 3.6 Ribbed Domes 3.7 Schwedler Domes 53.8 Stiff-Jointed Framed Domes -.0.0.......0.00. TV 3.9 Plate Type Domes ........ fess M2 3.10 Network Domes 2020202202021 12 3.11 Zimmermann Domes m4 3.12 Lamella Domes : : m6 3.13 Geodesic Domes... 2.2.2... sees 120 3.14 Grid Domes . 125 3.15 Improving the Buch 8 Behaviour of Sine Layer Domes : 129 3.16 Double Layer Domes 130 3.17 Search for the Optimum Type of Braced Dome. 135 3.18 Timber Domes ..... cece 186 3.19 Stressed Skin Domes 138 3.20 Some Outstanding Domes _ 138 3.21 lee Domes .... 141 3.22 Freedome® Structures . 0... ceeseee 142 3.23 Wind Loading on Domes .... 2 153 3.24 Low Cost Housing 187 3.25 Cladding 158 3.26 Erection . veeteeeeetteeee 189 References... ceeeees 165 4, Braced folded structures. . seeeeeeeeee 173 4.1 Introduction 173 42 Steel Frame Folded Plate Roofs cies 178 4.3 Baer’s Approximate Method .... o 178 4.4 Behaviour of Steel Frame Folded Plate Roofs... 179 45 Braced Barrel Vaults 183 4.6 Behaviour of Braced Barrel Vaults . fees 185 47 Practical Considerations ........ 193 48 Practical Examples feo . 197 References ........2. feveeseeeeses 208 Table of Contents 5. Connectors « 5.1 Introduction 5.2 Classification of Connectors . 5.3 Ball Joint Systems . 54 Socket Joint Systems 55 Plate Joint Systems 56 Slot Joint Systems . '5.7 Shell Joint Systems 58 Modular Systems ...... 5.9 Compositive Systems . 5.10 Comparison of Prefabricated systems 5.17 Design of Spherical Joints . 5.12 Other Patented Systems .... 5.13 Newly Developed MERO Joints 5.14 Some Simple Connectors References 6. Stressed skin systems 6.1 Introduction 62 Stressed Skin Stee! Buildings 6.3 Stressed Skin Grids 64 Stressed Skin Domes and Folded Plate Stuctures References ceeetseceeerseee 7. Cable suspended roof structures . 7.1 Introduction 7.2 Components of Suspended Roofs 7.3 Types of Cable Network Systems 7.4 Shapes of Cable Suspended Systems 7.5 Examples of Cabie Suspended Roofs 7.6 Analysis of Suspension Structures . .. 77 Construction Aspects of Cable Roofs . 7.8 Design of Cable Roofs 7.9 Ancillary Problems : 7.40 Erection of Cable Roofs... 7.11 Economy : 7.12 New Trends References 8, Tensile membrane structures ceeeeeeees General Part | Pneumatic Structures 8.1 Introduction 213 213 214 216 225 223 237 240 245 254 256 260 266 268 273 276 281 281 281 286 296 304 309 310 ant 35 320 337 340 343, 347 349 350 351 359 365 365 367 3678.2 Characteristics of Pneumatic Structures 8.3 Types of Pneumatic Structures .... 84 Structural Materials and Coatings 8.5 Fans and Pressure Control 86 Doors . . 87 Lighting .... 88 Other Aspects 839 Erection 8.10 Anchorage Design ... 8.17 Maintenance . 8.12 Temperature and Sound Control 8.13 Analysis of Pneumatic Structures 8.14 Trends in Pneumatic Constructions 8.15 Failures of Air-Supported Structures Part lI Membrane Structures ....... 8.16 Introduction 8.17 Analysis of Net and Tent Structures - 8.18 Factors to be considered in the design 8.19 Examples of Membrane Structures ..... 8.20 Examples of Large Scale membrane Structures 8.21 Membrane Materials .. 8.22 Typical Joints and Edge Details 8.23 Design Standards for Membrane Structures 8.24 Other Practical Considerations . References... 9. Tensegritie structures 2... 4. 9.1 Introduction... 9.2 Tensegritic Nets... 9.3 Tensegritic Structures 9.4 Tensegritic Structures and Maxwell's Rules 9.5 Morphological Studies 96 Characteristics of Tensegritic Structures. 9.7 Stability of Tensegritic Structures 98 Application of Tensegritic Nets ....... 9.9 Analysis and Design of Tensepie Structures 9.10 Other Aspects ... 9.11 Cable (Tenstar) Dome 9.12 Tension Strut Dome 9.13 Other types of Tensegrity Domes... 9.14 Truss Structures stabilized by Cable Tension 50) 9.15 Suspen - Dome Structure ....... 9.16 Flying Mast Fabric Roof # Stem References . 468 369 370 377 373 380 381 381 382 382 384 384 385, 386 396 404 408 409 412 430 438 439 446 454 - 461 461 463, 464 465 467 487 468 470 472 472 477 480 481 484, 486 489 ‘axe of Contents 10. nh 12. Analysis of skeletal space frames. - 495 10.1 Introduction . o - 495 10.2 Available Anal 495 10.3 Matrix Methods of Structural Analysis 512 10.4 Basic Principles of the Stiffness Method 513 10.5 Development of Stiffness Matrix for Plane Frame Members... cose 514 10.6 Assembly ofthe Stifiness Matrix for a Structure cee 817 10,7 Banding of the K-Matrix oe s+ 520 10.8 Solution to the Stiffness Equations 522 109 Stiffness Matrices for a Straight Uniform ‘Member in a Space Structure .... 523 10.10 Stiffness Matrices which include Secondary Effects cee 530 10.11 Stiffness Matrices which Take into ‘Account the Offset Members... 532 10.12 Determination ofthe Elastic Critical Loads of Space Structures coven 534 10.13 Software Packages ....... 536 References cocteeeeteeees 544 Formex data generation . - $55 11.1 Introduction 555 11.2 Formex Algebra cee 555 11.3 Formex Approach to Data Generation + 556 1.4 Formian .... : 572 115 Concluding Remarks 573 References 574 ‘Approximate methods of analysis . sees 579 12.1 General .. : 579 122 Double Layer Grids 579 123 Curved Space Frames. 617 12:4 Steel Frame Folded Plate Roofs ............ 649 12.5 Stressed Skin Folded Plate Roof 646 12.6 Design of Suspended Roof Structures 651 12.7 Pneumatic Structures 658 References... eeeeveeecee 663 Some important aspects of space structures ..... 667 13.1 Introduction... 667 13.2 Design of Skeletal Space Structures «10.11.11. 667Table of Contents Tle oF contents 13.3 Construction 679 134 Supports and Support Conditions... 690 13.5 Cladding of Space Frames ..... sees 693 13.6 Aesthetics, feteteeeeenens 696 13.7 Failures of Skeletal Space Frames ci. 697, 13.8 Other Types of Skeletal Space Frames... 702 13.9 Some Outstanding Space Structures n7 13.10 Current and Future Trends in in Spece Struct es. 722 References ....... : 2 735 14, Space structures with changing geometries . . 747 141 Introduction... 2... .... cee TAT 142 Deployable Structures... settee 87 143 Adaptive Trusses ...... +. 752 14.4 Pantadome System, A Deployable System for the Erection of domes . . os 783 1455 Retractable Roofs...) 76 References 778 ‘Appendix | Polyhedra and Polygon Patterns . 779 ALT General... . 773 1.2 Definition of Polyhedra sees 779 13 Types of Polyhedra fee 779 1.4 Properties of Regular Polyhedra . sees 782 15 Polygon Patterns 783 References 784 Appendix Il Computer Program for the Buckling Anais of Skeletal Space Frames .... se 785, A2.1 General 785 A2.2 Description of the Program 785 A23 Input/Output Data 787 2.4 Dimensions . 791 AZ.5 Examples . 791 A256 Listing of the Computer Program cise 782 A27 Sample Input of the Program... : 816 A28 Output of the Program .....2....... 817 Appendix Il Bending of an Axially Loaded Member ... 825 Index ++ 829 “Tensegitic Structures 41 Tensegritic Structures 9.4 Introduction ‘The famous American Engineer, R. Buckminster Filler is responsible for coining the word “Tensegrity’ and for doing important early work on this topic. He noticed that tension and compression always coexisted in Nature. He also utilized the concept that members carrying tensile loads are much lighter than those carrying compressive forces. This concept hhas been fully utilized in our human body which has heavy bones and slender sinews. The invisible skin of water is an example of a tensile ‘component of high structural performance. Any structural system needs continuity between its components to permit forces to pass through it to the foundation. In many structures the continuity is usually achieved through a series of compression elements, with isolated tensile elements. Faller advocated that if this Pater @ viany (o) Teepe pra tal Fig. 9.1 Tensegrity dome systemSpace Structures: Prineples and Practice —________ Spore Structures: Principe and Practice situation is reversed, then very lightweight structural systems can be built. To these structures he gave the name ‘Tensegrity— which was derived from the words ‘Tension-integrity’ Hence, according to Filler the Tensegrity system can be defined as a system which is established when a set of discontinuous compression components interact with a set of continuous tensile components to create a stable volume in space, Tensegritic structures may be composed of bars and a cable net, ‘The bars may be arranged in such a way that no bar is connected to another. Tensegritic shells may be constructed and a stable configuration is obtained by pre-stressing the bars against the cable net. When the shell gains its final shape no bar touches the other. An example of a tensegritic shell in the shape of a spherical dome is shown in Fig. 9.1 In many ways tensegritic shells are similar to pneumatic shells. In the case of tensegritic shells, the envelope, the cable net, is prestressed by the bars. In pneumatic shell, the envelope is pre-stressed by the compressed air. However, unlike pneumatic shells, no constant pumping of air is required in tensegritic shells. Hence they are attractive to the designer than pneumatic shells. ‘Tensegritic structures may be constructed from simple modules. Elementary modules comprise three struts, four struts and six struts and can be derived respectively from a triangular prism having 3 struts and 9 cables (Fig. 9.2), a half-cuboctahedron having four struts and 12 cables (Fig. 9.3), and an octahedron having 6 struts and 24 cables Fig. 9.4), For these three modules, all the cables have the same length (o) and all the struts have the same length{s). Hence they are called “regular” ‘modules. During many years, only these regular systems were examined, Pugh? gave a good description of these modules, which are completely defined by the ratio s/c. This ratio is equal to 1.47 for the 3 struts, 1.55 for the 4 struts and 1.63 for the 6 struts. Inegular shapes have also been subsequently developed. Some authors developed tensegrity system modules, not with straight compression elements, but with interlaced polygons, whose elements are in compression. Since their geometry must satisfy self-stress equilibrium, design of LO ai Fig. 9.2 3 struts system Fig. 9.3 4 struts system Figu 9.4 6 struts Triangular (Halt system prism) ccuboctahedron) (Gctahedron) Tenzerte Sutures tensegrity system requires a form finding process. For regular systems, designed by one parameter (the ratio s/o), form finding is mono parametered, Multi-parametered form finding processes have been recently developed for irregular modules and assemblies (see Ref. 27 fer the details) It has to be noted that in the assemblies of modules, more than one self-stress state exists and hence local collapse of a cable will not lead to the total collapse of the whole system. Playing on the ratio r= s/c, allows us to develop folding tersegrity systems?” by keeping the length of one component (cable or strut) constant and varying the length of the other component. 9.2 Tensegritic nets ‘Vinay'?223947 developed a different concept for tensegrity domes based on tensegrity nets. The main difference between Filler’s dome and that of Vilnay is that in Fuller's dome the bars are relatively short. Asa result, as the span increases, the curvature is reduced causing bars, to interfere with one another. This problem is avoided in Vilnay's concept; but the problem is that of longer bars, which, being in Fig. 9.5 Net No.2 Fig. 9.6 Net No.3 T Fig. 9.7 Net No.4 Fig. 9.8 Net No.5Space Structures: Principles and Practice compression, are subject to buckling (see Fig. 9.1) The net is considered regular when in every node the bar is connected to the same number of cables. The net is infinite when there is no limit to the number of nodes which can be added to a given net. Usually the nets are numbered according to the number of cables at a node. Figure 9.5 shows a net with two cables in a node and hence called Net No. 2. Similarly, Figs. 9.6, 9.7, 9.8 and 9.9 show net Nos. 3, 4,5 and 6 respectively. Space nets can also be built using these nets. 9.3 Tensegritie structures ‘The general arrangement of cables and bars in the various nets can be used to construct various structures. In Fig. 9.10, the Net no. 2 is used AAAS CSRS RRRORER PSR DR RKARY Fig. 9.9 Net No. 6 Fig. 9.10 Barrel vault using Net No2 Fig, 9.11 Spherical dome using net no.6 Tensegrte Structures to construct a barrel vault. Similarly, Fig. 9.11 shows a spherical dome constructed out of Net No. 6, ‘Tensegritic structures may also be classified according to their shape. Thus, if each strut is surrounded by four tendons defining the edges of a diamond shape, itis called a diamond pattern system. Similarly systems having interconnected struts which do not touch one another are called circuit pattern systems. Fig. 9.12 shows a larger circuit pattem system which has twelve decagonal circuits of struts which interweave without touching one another. In another relationship between struts and tendons, three tendons Join the opposite ends of each strut and define a mirror image along it. Systems which have their struts and tendons arranged in this way are called zigzag pattern systems, though in many cases the zigzags of tendons will almost be straight lines. Figure 9.13 shows a larger system with 90 struts and 270 tendons based on a pentagonal dodecahedron ‘whose faces have been divided into 45 triangles. It can be seen that the struts in these zigzag system do not touch one another, continuity being achieved through their tendon networks. 9.4 Tensegritic structures and maxwell’s rules ‘An interesting point about these zigzag systems is that three tendons and one strut meet and form a simple pinned joint at each strut end. But if a framework is designed with pinned joints, according to Maxwell's rule, it should have a minimum of 3J-6 members if itis to Fig. 9.12 Circuit pattern system of a tensegrtic structure, (Courtesy: A J. Pugh)66 Space Suvetuts: be simply stiff, where J is the number of its joints. But inthis case only cone strut and three tendons meet at each strut end. A zigzag pattern figure can only have 2J members, three-quarters of which can take tersion only. According to Maxwell's rule the 90 strut figure in Fig. 9.13 should have 534 members. But it has only 360 members and it can be checked experimentally that this frame is stiff. The frame thus constitutes @ paradoxical exception to Maxwell’ rule. But Maxwell does in fact anticipate such exceptions to his rule, forhe states (ref. (1), p-599, collected papers, Vol. 1) ‘In those cases where stiffness can be produced with a smaller number of lines (bars), certain conditions must be fulfilled, rendering the case one of a maximum or ‘minimum value of one or more of its lines (bars). The stiffness of the frame is of an inferior order, as a small disturbing force may produce a displacement infinite in comparison with itself! Maxwell might have intended to refer to a maximum or minimum value ofthe length of one or more ofits bars. Fler’ Tensegrity structures hhave tumbuckles in some members and they cease to be stiff if the turabuckles are relaxed, There is a limit to the length of interior bars within the net formed by the outer members, and it was at this limit that the frame in fact became stiff. This is presumably precisely the kind of maximum which Maxwell hac in mind. Thus it seems clear that Fille’s invention corresponds to an exceptional special case anticipated by Maxwell. “Tensei Structures 9.5 Morphological studies Several morphological (morphology simply means the Study of Form) studies have been conducted on tensegrity structures. Most of them have the feature that bars are not in contact with one another. Only a few examples of these studies, representing milestones in the development of tensegrity structure morphology are highlighted here. ‘Snelson's work *”2” is of artistic nature, Many of his ‘sculptures’ are of more or less free form or of low degree of regularity. From the structural point of view they are all characterized by having a single state of prestress i.e, the tensioning of one cable presiresses the whole structure. Filler’s tensegrity dome? is a multifaceted, single-layer tensegrity polyhedron, in which the cables form the outside skin and the bars form an internal layer (as already shown in Fig. 9.12). Emmerich!9! of France was the first to concrive double-layer tensegrity networks. In this configuration, bars are confined between ‘two layers of cables or tendons. This configuration is obtained by Joining together tensegrity prisms or truncated pyramids, or "simplexes” as Emmerich terms these objects. Flat and curved surfaces are also possible. Emmerich also investigated extensively and systematically tensegrity polyhedra, which can produce double-layer domes? Vilnay introduced the concept of infinite tensegrity networks which have already been explained in sections 9.2 and 9.3, ‘Motro?*“° produced double-layer tensegrity grids by joining together tensegrity prisms at their nodes. As a result, his grids contained bars Joined together at nodes, unlike other configurations studied, but bars are still confined between two parallel layers of cables. Hanaor” carried out extensive investigations on Double-layer Tensegrity Grids (DLTGs) with non contacting bars. He also investigated the geometry of double-layer tensegrity domes by gecdesic subdivision of a pyramid, employing DLTGs made of tensegritic pyramids. Grip** extended the inventory of tensegrity polyhedra and hyper polyhedra, through the concepts of duality and space filling polyhedral networks to produce a virtually unlimited range of intricate and complex networks, 9.6 Characteristics of tensegritic structures As already mentioned, tensegritic structures are similar to bubbles and hence to air-supported structures and balloons (see chapter 8). In these structures, the struts are on the inside of the system and the tendons on the outside. The struts push outwards like the pressurized air inside the balloon but are restrained by the tendons which are equivalent to the tensile skin of the balloon. Another characteristic which is common to balloon and tensegrity structures is that they are already stressed. Though these structures 467‘Space Structures: Pricinles and Practice need only to support the weight of the system, the prestress may be increased to improve the load bearing capacity of the system. If the structure can be prestressed and if it is stable under prestressing, the geometry of the structure is uniquely defined. Most of the tensegrity structures, like balloons are very sensitive to vibrations. However the vibration effects can be improved by the addition of extra tendons. The whole system can be deformed under load only to return to its original shape when that load is removed. ‘The main feature that distinguishes tensegrity structures from conventional prestressed cable networks is that they are free standing and do not require external anchorages or continuous stiff ring beams. This feature, combined with the simplicity of the connection of the bars to the cables, makes this structural system particularly suitable for applications requiring deployability or demountability An important point that should be remembered is that each tensegrity system is enantimorphic, each system having two versions which are mirror images of one another, so care should be taken when this could be critical. 9.7 Stability of tensegritic structures ‘To fulfil equilibrium conditions at every node, every bar has to be connected at least to three cables not in one plane or two cables where bar and cables are all in one plane. Nets with two cables at a node can be prestressed only in that plane. Only nets with three and more cables at a node can be prestressed for any desired geometry. ‘The structure presented in Fig. 9.10 cannot be prestressed and hence is not stable. The structures given in Figs. 9.11, 9.12 and 9.13 can be prestressed and will be stable when the length of bars and cables is appropriate, ‘The stability of the prestressed structures in most cases is self evident; the prestressed net No. 2 in the plane is not stable since it will collapse at any movement of any node perpendicular to this plane. The prestressed structures given in Fig. 9.11, 9.12 and 9.13 are stable because these have features common to a pneumatic structure where the cables act as the skin and the bars transfer the inner pressure, ‘The most convenient way of prestressing is by lengthening the bars ‘until equilibrium is reached. In most of the cases the geometry of the prestressed structure will differ largely from the pre-prestressed structure. A geometric nonlinear analysis must be used to calculate the geometry of the net under prestressing. 9.8 Application of tensegritic nets Since tensegrity can be defined in such general terms, there is a chance that it has very important implications outside the realm of large scale engineering. The basic concept of establishing a structure through an interaction of forces, rather than through a preconceived arrangement Tensei Structures Fig. 9.14 — Masts built out of tensegritic nets of components could be very important. “Masts can be formed by joining several figures together as illustrated in Fig. 9.14 (The thick lines represent the compression members ad the thin lines represent tension members). Many lightweight lattices can be formed from tensegrity systems, especially the less complex ones. Since the struts and tendons of the models represent forces in the system, they can be replaced with other components, provided those components carry those forces. For example domes can be formed ‘using compression elements and tensile skins where the tensile skin470 Space Structures: Principles and Practice Fig. 9.15 Model of Tensegrtic dome with tensile skin. (Courtesy: A 1. Pugh), acts in a way similar to that of the network of tendons (see Fig. 9.15). 9.9 Analysis and design of tensegritic structures Tensesrity structures are, from the structural analysis point of view, prestressed pin jointed networks similar to cable networks. Except Jn some special configurations, they are geometrically fexible-they contain intemal mechanisms. As in the case of cable network, the analysis consists typically of two major phases: A shape finding phase, aimed at obtaining the equilibrium pre-stressed geometry, and the static (or dynamic) analysis which determines member forces and nodal displacements under applied loads. An intermediate phase of investigation of stability, mechanisms and states of prestress is sometimes employed, mainly for research purposes”. Motro and his co-investigators (Ref. 18, 27, 34 and 35) applied the technique of dynamic relaxation to the shape finding problem. They also performed analysis of tensegrty prisms" and simple double-layer networks. Vilnay, in his book ®, outlines the principles ofthe investigation fof mechanisms and states of prestress and the shape finding, static and dynamic analysis phases, but his procedure is not very suitable for computer implementations. Hanoar* presented a flexibility based model for investigating mechanisms and states of prestress and for the static analysss employed a stiffness based procedure, which is computationally efficient, for the actual analysis of sizable networks. Vinay” #” used Tensei Strvclunes linear algebra to determine the nodal displacements as well as the forces induced by the prestressing ‘The intemal forces and reactions P due to the extemal forces Q acting at the nodes of these structures can be calculated using the following, equilibrium conditions at every node: (Al (P} = (9) (1) where [A] is the statics matrix as defined by the geometry of the prestressed structure. In Eqn (9.1), the cables are treated as bars, when the prestressing action ensures that the forces in the cables will remain tensile under the given extemal forces. In each node there are three equilibrium equations and the number of unknowns around each node is half the number of cables and bars ‘which meets there, In solving Eqn (9.1) three situations arise: (1) when ‘tke number of unknowns is larger than the number of equations, (2) when the number of unknowns is equal to the number of equations and (3) when the number of unknowns is less than the number of equations. Under condition (1) the structure is indeterminate, under condition (Q) the structure is determinate and under condition (3) the structure is unstable. Hanaor?™' carried out parametric analytical investigations of double-layer tensegrity grids of varying geometries, including domes and geometrically rigid configurations (statically indeterminate, with no internal mechanisms). From these studies, he found that domes are stiffer than plane grids over a given span and geometrically rigid configurations are stiffer than geometrically flexible ones. There is a direct correlation between stiffness and load carrying capacity for a given span and topology. Hanaor®®! investigated the various configurations of double-layer tensegrity grids (DLTG) covering a circular span of diameter 27 m and cempared them with double-layer grids. The results indicate that, so Jong as the principle of non-contacting bars is maintained, all DLTG ‘configurations are heavier than the double-layer grids, due to the long bars. From the tests conducted by him’, he concluded that it is desirable to design DLTGs for bar buckling as the governing failure mechanism, High bar slendemess ratio can provide ductility and ethance force redistribution capabilities. Experimental tests were carried out by Hanaor*! and by Motro, et al on small models. Hanaor performed static tests on small DLTG models of various configurations. Motro et al. performed dynamic tests on single triangular prism (simplex) models. All tests indicate good agreement with theoretical predictions. However Hanaor and Liao”? suggest that the design of tensegrity grids must be based on a nonlinear analytical model accounting for large deflections. A linear anan Space Structures: Principles and Practice ‘model was found to produce an unsafe design. 9.10 Other aspects Structures built of infinite regular tensegritic nets having appropriate boundary conditions can be pre-stressed if atleast three cables meet at a joint. If the net has less than five cables at a join: the pre-stressed structure can carry in its prestressed geometry only a limited number of families of external loads. For other external loads the structures change their geometry a great deal. Structures having more than five cables at a node can carry external forces in any direction. These structures will deform only because of their elasticity under external forces. It is clear that the tensegrity structures display Faller’s ingenuity in designing large but easily packed and portable structures: an obvious saving in weight results from the fact that about three-quarters of the members are wires rather than rods On the other hand, if the aim is to design economical but stiff engineering structures, itis not clear that there is much point in making the outer network so sparse that the resulting frame has a number of infinitesimal nodes whose stiffness is necessarily low. Hence there is a need to study these structures experimentally, with and without additional wires which will make the tensegrity net satisfy Maxwell's rule. One of the main technical aspects in the implementation of tensegrity structures is the roof covering, Most configurations are geometrically flexible. Hence roof covering has to be a flexible membrane. However, not many studies have been carried out on tensegrity structures with surface membranes. It is important that the membrane forms an integral part of the design, as the problem of ensuring a creast-free, flutter-free surface, is by no means trivial. For deployable applications, itis highly desirable that the membrane serves as a structural, as well as functional role, reducing the size of cables or even eliminating taem’®™. 9.14 Cable (tenstar) dome In the search for solutions which retain the advantages of air supported structures while avoiding their disadvantages, severa! tensile structure systems have been developed. First practical example of application of the tensegrity dome was provided by a Polish Engineer, Waclaw Zalewski in 1961, three years before B. Filler patented his idea in 1964. Zalewski covered an auditorium in Katowice, Poland, using the first version of tensegrity dome’?, David Geiger, who designed the famous cable restrained air supported roof for the U.S. Pavilion at Expo “70, modified the principle of Filler's tensegrty structure and developed @ radial cable dome called the tenstar dome. . Cable domes were first adopted for the Gymnastic Arena (dia. 120 m) Tensegrte Suuctures an Ridge coon ew Diego be oe Hop cca ener plano Tet Done Fig. 9.16 Schematic diagram of Tenstar Dome wan VZes PN SOLIS SSS / i Valley cable Fig. 9.17 Cable (Tensegrity) dome48 Space Structures: Pincinles and Proctce Crist Te ge ce cond sein ¥- Je a eu Lowercase — Lo Samim Fig. 9:18 Cable dome detals; casting, pot svand assembly and the Fencing Arenas (dia. 90 m) of the Korean Olympic Games at Seoul, in 1986, The schematic diagram of the tension roof is shown in Fig. 9.16 and the details of the roof over the Gymnastic arena are shown in Fig. 9.17. These cable domes span the space using continuous tension cables and discontinuous compression posts. Loads are carried from a central tension ting Unrough a series of radial ridge cables, tension hoops, and. intermediate diagonals until they are resolved and anchored) in a perimeter compression ring*. The tension hcops are 14.5 m apart. Three tension hoops were used for the Gymnastic Arena and two for the Fencing ‘Arena. The ridge and diagonal cables separate the roof into sixteen equal “Tensegitie Steuctures segments. For roofs with equal segments, equal hoop spacing, equal loading on tributary areas, and corresponding vertical geometry, the corresponding members of cable domes of different diameters cary the same load as one moves from the centre of the dome outward. By analogy, the dome behaves as two cantilever trusses not quite touching at the centre. This has resulted in significant repetition of details and allowed the use of same castings (see Fig. 9.18) at the connection points where posts and cable meet. As a consequence, the two rings of the Fencing ‘Arena share the same castings at the top and bottom of the posts and the Gymnastics Arena required another set of castings for the top and bottom of the posts of the outermost hoop. ‘The components of the roof consisted primarily of continuous, parallel 15 mm diameter prestressing strand (ultimate capacity 266 kN) and metal castings. Bridge rope with pinned sockets run between castings, controls the geometry and provides equilibrium at the castings. The elements were assembled at ground and hoisted into a hanging position where they were jacked into final position using two hand held jacks at the base of the posts of successive hoops. After the cables ‘were erected, the fabric was deployed and clamped to the top of the ridge cables. Then the valley cables were prestressed, thus inducing the required prestress into the fabric. ‘The structural behaviour of these domes was found to be highly nonlinear. The cable, hoop, post structure, however, is statically determinate for the snow load case; if the valley cable has zero load (this may be assumed as the initial configuration for the nonlinear analysis program). Use of the parallel cables builds redundancy into the cable dome system. These low rise cable domes permit the construction of very economical and lightweight domes. The domes in Korea, translucent and insulated to R10, weigh only 14.65 kg/m? and were constructed at a cost of approximately US $215 per sq.m. ‘A cable dome over an elliptical plan (78 m dia. x 92.65 m dia) for the Ilinois State University (with only one hoop cables) and a 210 m diameter Florida Sun coast dome (with four hoop cables) at St Petersburg, Florida (Total weight of the structure = 8,258 KN, i.e, 0.24 knN/m’) have been constructed by Geiger and Associates. The details of these domes may be found in Ref. 45. It has to be noted that neither the weight nor the cost of these cable domes increase significantly with the increase in span or change from circular to super elliptical plan configurations‘ Cable dome made its debut in Japan in Izu Shizuoka Prefecture. Weidlinger Associates designed the largest hyperbolic tensegrity dome over the Atlanta Stadium in Georgia, which hhas drastically changed the existing concepts regarding construction methods used for stadia 5 *. The Georgia Dome roof (built in 1994) also has concentric rings. But (reverting back to Buckminster Filler's a8476 Space Surverutes: rincples ane Practice Fig. 9.19 Support system of HB cable dome invention) they are supported by triangulated cables thus adding to the complexity. ‘There is no doubt at all that tensegrity domes offer many advantages over more conventional skeletal types and will be used in an increasing ‘way during the years to come. Geiger has also presented cost comparison, of double-layer grids (the Takenaka Truss), air supported roofs and the cable domes", His comparison includes energy considerations, roof, ‘membrane comparisons and insurance costs. He has illustrated the impact produced by the introduction of tensegrity domes upon lightweight ow cost space structures for large span buildings. Because of the economy and ease of erection, cable domes have generated widespread interest and resulted in investigations and extensive testing by several other researchers*®*"., Berger has suggested a system called composite cable dome consisting primarily of intersecting cables whereby each cable supports two struts and each strut rests on two intersecting cables as shown in Fig. 9.19. This was originally developed for the St. Petersburg Sun dome (though finally Geiger’s system was adopted for that dome). This system is simple to install and provides a considerable degree of redundancy. The advantage of this system (and any cable dome) is the predominant use of tensile members. The disadvantage is the very long and complex force flow though high strength members which results in large forces and deformations (and hence not suitable in areas with large snow loads). To avoid these problems, Berger suggests a composite system in which the predominant uniform loads are carried to the supports in a direct way by the use of arches (see Fig. 9.20). The cables and struts in this system permit an erection process without Fensegrite Structures Fig. 9.20. Section of arch reinforced cable dome interior temporary supports and provide stability for the arches and the reinforcement for unbalanced loading. This composite cable dome system has been adopted for the replacement of the Unidome roof at Iowa, U.S.A. and the stadium with the new roof was opened in 1998°° 9.42 Tension strut dome The first tension strut dome (TSD) (Amagi dome, with a diameter of 43 m) was built in Japan in 1991 to cover the multipurpose gymnasium at Amagi Yugashima Town. The TSD dome has similarities to the cable dome, which was originated by Geiger in the U.S.A. There are, however, major differences®, (1) Cable domes use membrane material. They are constructed so that the rise of the domes is secured by a combination of diagonal cables and hoops as shown in Fig. 9.16 and 9.17. Whereas, the TSD dome is supported by a central lens-shaped sable girder (see Fig. 9.21 b). This is suspended by a self supporting tension truss which is anchored in cables below and 1 circular tension ring above that stabilizes the cable girder. In other words, it is a new system that integrates two different structural systems. ; (2) The Cable dome system introduces tension to cable members in ‘a continuous way through various pulleys which are situated at the top of each of the posts. The TSD dome system utilizes tension in all of the upper and lower chords. These chords are a7a8 6) Space Stuctres: Principles and Rootpan | Root ning pine Fig. 9.21 Details of Amagi (TS.0) dome, Japans3 situnted independently and do not make use of any slip metal fittings. ‘The tension to a cable dome is introduced through numerous diagonal cables in gradual stages. It is a step by step system that uses wedges to increase the tension gradually. Tensior. can be introduced to a TSD dome by pulling the ends of the upper chord cables in basically one movement and at one ime. Tensegriie Stuctures Fig, 9.22 Interior view of the Amagi Dome, Japan. (Courtesy: Pro. K. Ish ‘The original configuration is therefore easily monitored and ‘maintained, and it remains more stable as a whole. ‘TheTSD dome system can be appreciated as a realistic and practical design in the field of tension dome structures. Complexities regarding slip fitting technology and construction procedures used by cable domes. have been improved with the development of this TSD system. ‘The self-supporting tension strut Amagi dome is connected to ‘ts substructure through a complex of V-shaped steel beams. These beams. are fixed to a reinforced concrete compression ring (see Fig. 9.21) It is difficult to solve the problem of supporting the compression. reaction force created from the large roof tension dome at the tips of he ‘V-shaped beams. These compressive reaction forces may be supported by a ring truss consisting of two rings: (1) an upper compression ring placed at the tips of the V-shaped beams where all the compression reaction forces. concentrate, (2) a reinforced concrete compression ring placed on the substructure. However by adopting this method the continuous wave shape of the roof is distorted. Hence in the Amagi dome, the tops of folded plete ‘wall and the tops of V-shaped beams were connected directly with stays. ‘which continued along the line which extended to form the upper chords. (see Fig 9.21) a0‘Space Structures: Pincines and Practice Using this method, the compression reaction force can be directly transferred from the chords to the V-shaped beams-then to the stays and to the folded plate wall. With this method it was thus possible to concentrate the compression reaction force onto the reinforced concrete compression ring which has been placed on the tip of the folded plate walls. ‘The grandness of a TSD dome lies in its astonishing lightweight roof generally less than one-fifth the weight of any dome made from ordinary metal. Using the advantages of its lightness, a lift-up construction, method has been adopted for this dome. This method involves the lifting of the completely fabricated roof structure from the ground floor in a single, unbroken movement. More details about this dome may be found in Ref. 53. The interior view of the completed dome is shown in Fig. 9.22. 9.13 Other types of tensegrity domes Based on their work on flat double-layer tensegrty grids, Kono, Kunteda?® presented a numerical form finding method, a shape formation process. and the static response of a new type of pretensioned double-layer tensegrity grid dome. The double-layer domes investigated by them are formed by tetrahedron-like tensegrity modules as shown in Fig, 9.23 ‘Two adjacent modules were connected by a single compression member Gee Fig. 9.24). Each module can be independently prestressed, and the prestressing is used for the shape formation of the cable dome. ‘The number of infinitesimal mechanisms the grid dome contains relatively small. These domes feature deployment type construction methods and found to be greatly stiffer than inital flat grids Wang” discusses a new type of cable dome, called Wang's dome, which is a cable dome strengthened by additional cables to make the conventional kinematically indeterminate cable dome into a kinematically determinate one. It is capable of sustaining all loading conditions without relying on prestress. The additional cables work and slacken under varied loading conditions, removing all possible mechanisms effectively. Unlike cable domes, in which the structural shape has to be adjusted during erection by proper prestress distribution, Wang's dome does not pose any problem during erection. It can be applied to rectangular plan layout also. Wang’ also discusses about tensegrity ring beams, which can interact with a cable dome to form a truly self stressed, self equilibrium tensegrity system. The tensegrity ring beam can be constructed of tensegrity prismatic or pyramidal simplexes and by radially linking them. Besides, boundary struts must be added to reinforce the ring afterall simplexes are prestressed and connected, Wang” also discusses new concepts such as tensegrity frames and Linear Complementary Equation (LCE) method for solving cable- Tensegte Structures Fig. 9.23 Tetrahedron like tensegrity modules used by Kono and Kunieda®® Fig, 9.24 Flat double-layer grid formed by using Tetrahedron modules slackening problem of cable-strut systems. 9.14 Truss structures stabilized by cable tension (TSC) ‘A four-bar link system as shown in Fig. 9.25a, forming a lozenge configuration isa kinematically indeteminate frame. It has inextensional displacement modes and usually cannot be used for structural skeletons.Space Srvtuts: Principles and Practice Cable Truss member “ruse member Post (2) Kinematically indeterminate frame (®) Unit structure Fig. 9.25 Unit structure Addition of a post and eight cable members, connecting each node of the truss to the two ends of a post (as shown in Fig. 9.25b), alters the framework kinematically determinate having one self equilibrated stress state, Introduction of this self-equilibrium stress keeps cable members in pretension and cables can act as if they were compressible. This prestressed framework was named as “Unit Structure’ by Kawaguchi and his associates®, Many unit structures may be assembled to form structures called "truss structure stabilized by cable tensions (T.S.C)" Kawaguchi and his associates have investigated the structural ‘behaviour of this truss system by means of numerical analysis. In order to grasp the actual performance of the frame they carried out two series of tests using full scale unit structures and a partial model of vaulted TS.C. dome%. The vaulted TS.C, having a dimension of 13.35 m x 22.5 m in plan, constructed and tested by them is shown in Fig. 926. A set up of tum buckles was arranged on each tension member for the introduction of prestress. Mero type steel ball joints were employed at the connection of truss members. PVC coated polyester membrane was employed as the roofing to cover the TS.C. skeleton of the dome. From the tests it was found that there exists difference in the behaviour between the nonlinear analytical model and the real model, especially after the slackening of the short members. Tey also concluded that the inital stress of a unit structure ‘could be introduced through the operations of just one tum-buckle ‘mounted on the longest tension member. Wang? has also discussed about these truss structures stabilized by cable tension, but calls them as Reciprocal prism (RP) and Crystal- cell pyramids (CP). According to him, RP system is formed by linking self stressed equilibrium reciprocal prisms (see Fig. 9.27) at their middle Joints directly and at their upper and bottom joints by additional connecting cables respectively. A rectangular RP simplex, for example, is made of a vertical strut, four enclosed horizontal struts and eight edge cables connected by (frictionless) hinged joints (as already shown in Fig. 9.25). For each simplex, the elongation of the vertical strut, either by telescoping or screw turning, will prestress the system conveniently. In the formed planar grid, most upper connecting cables TensegrticSuuctures The details of the TSC dome tested by Kawaguchi & Associates®® Fig. 9.26‘Space Structures: Principles and Practice oe (© Squre RP (0) Penagoas RP (e) Henaagote RP Fig 927 RPSimplenee YM (2) ange CP (@) Resangsar (6) Ponape. Fig. 9.28 CP. Simplexes always slacken under downward load, but they can be used as ridge cables for the membrane roof or under upward load response, Wang's CP system is formed by connecting self-stressed equilibrium crystal-cell pyramids (see Fig. 9.28) at their upper joints directly and at their bottom joints by additional cables. A rectangular CP simplex, for ‘example, is composed of five inner struts: one vertical strut and four inclined struts and joined by eight outer cables by frictionless hinged joints. Like RP simplexes, there is only one state of prestress in a CP simplex, and if no cable slackens, there will be no inner mechanisms. Case studies conducted by Wang show that in comparison with a space truss of similar layout, cable strut grids can save about 2/5 of the self ‘weight and further reductions are possible using optimal design employing small grid lengths. Also, each joint is connected with five bars at the most and hence joint design can be simplified and consequently joint be lowered. The reduced grid length increases the stability of struts with reduced strut forces. However, the self weight of the system is increased only slightly”. 9.15 Suspen-dome structure ‘A Suspen Dome developed by Prof. Kawaguchi and his team, is a composite system consisting of a single-layer truss dome, and a tensegric system (struts and cables) as shown in Fig, 9.29. Connected to the struts suspended from the single-layer truss dome are radial cables, which radiate from the centre of the dome, and ring shaped hoop cables. Thrust of the entire dome, and stresses of single-layer truss members can be reduced by the tensegric system with appropriate prestress. In the case of additional loads such as Tensegiie Structures ‘Singlelayer truss dome Hoop cable ‘Strut Radial cable Fig. 9.28 Suspen-dome (structural system) snow loads, increase in the stresses of truss members or the deformations can be controlled because of the tensile forces produced in the cables. As described above, the suspen-dome structural system is characterized by stress and deformation control by the effective use of the tensegrie system. In addition, excessive structural fexibility, as experienced in cable domes, can be prevented and the strength of the single-layer dome against buckling can be increased. The suspen-dome system, therefore, enables the construction of safe and economical large~ scale domes, which was impracticable with conventional methods®. ‘Thus the suspen-dome has the following advantages: * reduced burden on boundary structures © increased strength against buckling and © control of structural flexibility The behaviour of suspen-dome has been studied extensively by Kawaguchi and his associates at the Hosei University, Japan and its safety against buckling was verified with structural model tests™. Based on these studies, the Hikarigaoka dome was constructed in Nerima Ward, Tokyo in 1994. This dome is a 35 m span single-layer truss dome consisting primarily of H-beams, as shown in Fig. 9.30. It has a height of 14 m and the total weight of the roof was about 1274 kN. Only the outermost ring of the trussed dome is stiffened with a tensegrity system. Cable prestresses were applied so that the axial stress in the outer ring girder under long-term loads may be close to zero. In this structure, prestressing steel bars are used as radial cables. The dome is supported by V-shaped steel columns jointed continuously, with both column capitals and bases being pii-jointed in the radial direction so that movements of the roof in the radial direction are permitted. The roof was. finished with steel plates and the central part of the roof was glazed to obtain sunlight. The tensegric system consists of three components: hoop486 Space Structures: Principles and Practice Pri se oN ie Radial od 61.246) Se 323) amas Pe (©) Cable st sytem ig. 9.30 Principal components of the Hikarigaoka dome"? cables, radial cables and struts. A method was developed and adopted by Kawaguchi by which prestress was applied by extending the struts®, Static load tests and vibration tests were conducted on this Full-size dome. From these tests it was concluded that (1) the design stiffness duly rellects the actual conditions, (2) no significant difference exists between the natural frequency characteristics of the suspen-dome and the single-layer truss dome and (3) the eigen value analysis showed good agreement with the test results in frequency and vibration mode®®®, Thus, these tests verified the effectiveness of the Suspen-dome system. Ancther dome, Fureat Dome, has been constructed using the Suspen- dome system in Nagano Prefecture, where there is heavy snow fall. Icis proposed to measure the changes in cable tension forces due to the snow fall and temperature changes. 9.46 Fiying mast fabric roof system Another type of tensegritic structure has been adopted over the Shanghai Sports Centre, China®!. When this fabric roof was planned, Tensegii Structures ‘the American team prepared two schemes: (1) A cantilever steel trussed structure with secondary bow string arches to support a fabric roof, (2) A cable supported fabric roof anchored to the perimeter concrete compression ring beam (Tenstar dome). However, a third scheme with flying mast fabric roof structure based on tensegrity concept was developed and accepted as the final solution®. The flying mast fabric roof structure layout was governed by the arrangement of the cantilever steel trassed roof structure and consists of 32 bays and 17 different types of fabric roof panels. Like large umbrellas over the heads of 80,000 spectators, 57 flying mast cable and fabric panel systems hover over the Shanghai Stadium. Completed in October 1997, the stadium has a floor area of 150,000 m?. ‘The roof is a saddle shaped ring that is a cantilevered spatial structure composed of radial and circular trusses. It is made up of 57 umbrella-like cable structures covered with an advanced fabric membrane. The panels appear to be flying because they are up in the air. The long cantilevered ‘russes make the effect even more striking, Birdair Inc., Buffalo, NY. manufactured the panels and designed them with assistance from Weidlinger Associates Inc, New York. The Shanghai Institute of ‘Architectural Design and Research designed the stadium, Design wind loads were established through two wind tunnel tests in China and one in Canada. A 1:35 model was made and tested under four different load cases. ‘The concrete superstructure consists of an underground athletic centre, a parking garage, an event area, seating boxes and a hovel. The building and roof were finished in about 20 months, with as many as 3,000 workers on-site at one time. ‘Tae plane projection of the roof is in the form of two ellipses, one inside the other. The outer ellipse measures 288.4 m in the East-West direction and 274.4 m North-South; while the inner ellipse is 150 x 213 1m, The highest point of the roof is 62.5 m above ground on the Westen part of the roof, 41.2 m on the Eastern side; while the highest points on both the Northern and Souther sections is 31.8 m. The maximum overhang of the roof is 73.5 m. ‘The 28,000 m? fabric roof rests on 32 steel cantilever steel trusses that span 21 to 73.5 m, some of the longest in the world. The trapezoidal fabric panels inserted between the trusses are supported by a mast and cables (see Fig. 9.31) to provide a translucent, lightweight roof. To erect the parels, the telescopic masts were pushed up into the middle of the fabric, ike an umbrella The trusses are interconnected and stabilized by three rings of steel space trusses. Depending on the length of the trusses, the bays between them consist of one to three fabric panels. The panels, made of Teflon (PTFE)-coated fibre glass fabric, were shipped to Shanghai from Birdair's manufacturing plant in the U.S.A and installed by the Chinese288 Space Structures: Principles and Practice Tensegrte Structures crew. The installation took only eight weeks-six weeks less than the original schedule, _ Each cantilever steel truss consists of a pair of parallel welded pipe trusses spaced 1.3 m apart with horizontal lacing tie members provided at each truss node joint along both top and bottom chords. Each bay formed by two adjacent cantilever trusses is trapezoidal but roughly rectangular in plan. To accept the fabric roof, a perimeter steel pipe framing was introduced 1.28 m above the top chord of the cantilever ‘russes at each bay and short posts supporting this framing were connected to the truss top chord at the node joints. The maximum diameter of the steel tubes in the spatial roof is 508 mm. Specially designed paired K-shaped joints were evolved, tested and used. This raised pipe framing fulfils three functions: (1) it acts as the boundary supporting structure for the flying mast fabric roof with the fabric panels and suspension cables of the flying masts anchored to this framing (Fig. 9.31), (2) it provides the roof drainage gutter system in the radial direction of the fabric roof by utilizing the vertical space between this framing and the cantilever steel trusses, 4 and (3) it transfers the fabric roof loads to the main supporting cantilever trussed roof structure. =. Following the tensegrity concept, each flying mast unit consists of 2 four upper ridge cables, four bottom suspension cables and a flying vertical compression post. Although all cantilever trusses are slightly tilted upward from their back supports, the decision was made to keep ‘the flying posts vertical in the global direction with post heights varying from 4.5 m to 5.5 m at each flying mast location. More details about this roof may be found in Ref. 61 and 62. References 64 to 95 provide more information on latest developments. REFERENCES 1. Calladine, CR., Buckminster Foller’s Tensegrity Structures and Clerk Maxwells Rules for the Construction of Stiff Frames, Intemational Journal of Solids and Structures, Vol. 14, No. 2, 1978, pp. 161-172. 2. Filler, R.8., Tensile-Integrity Structures, US. Patent 3, 063, 521, 1962. 3. Filler, RB., Synergetics: Exploration’ in the Geometry of ‘Thinking, ‘MacMillan, New York 1975. 4. Kavanagh, T.C., Some Aesthetic Considerations in Steel Design, Journal of the Structural Div., ASCE, Vol. 101, No. ST 11, Nov. 1975, pp. 2257-2275, 5. Kuznetsov, E.N., Statical Kinematic Analysis of Spatial Systems, Proc. of the 2" International Conference on Space Structures, University of Surey, England, Sept. 1975, pp. 123-127. 6 Marks, RW. The Dymaxion World of Buckminster Faller, Reinhold, New York, 1960. 7. Metal, J., Buckminster Faller, R.. Prentice-Hall International, London and George Braziller nc., New York, 1962, p.127. 8 Pugh, A.t, An Introduction of Tensegrity, University of California Press, Berkeley, 1976, p. 122. Fiying mast fabric over stadium in Shanghai, China al sot I Convent Fig. 9.31‘pace Structures: Principles and Pratice 9, Pugh, AJ., A Brief Introduction to Tensegrity, Proc. of the 2" International Conference on Space Structures, University of Surrey, England, Sept. 1975, pp. 707-711. 10. Srinivasan, CN, Human Skeleton - An Amazing Load Carrying Framework, Joumal of the Institution of Engincers (India), IDGE Div., ‘Vol. 56, pt-3, May 1976, pp. 69-71. 11, Subramanian, N, What is Tensegrity, Science Reporter, Vol. 16, No. 2, Feb. 1975, pp. 120- 121. 12, Viluay, 0,, Structures Made of Infinite Regular Tensegritic Nets, Bulletin of the IASS, Spain, Vol. XVIll-1, April 1977, No. 63, pp. 51-57 13, Subramanian, N, and Radhakrishnan, R., Tensegritic Structures, Journal of the Institution of Engineers (India), Civil Engg. Div. Vol .60, pt. C15, ‘March 1980, pp. 276-279. 14, Vilnay, 0., Detemminate Tensegritic Shells, Architectural Science Review, Vol. 22, No. 2, June 1973, pp. 39-43. 15, Viinay, 0, Determinate Tensegritic Shells, Journal of Structural Div, ASCE, Vol. 107, No. ST10, Oct. 1981, pp. 2029-2033. 16. Pugh, A., Tensegrity Technique, Building Design, April 20, 1979, pp. 34-35. 17. Viinay, Oy and Sek, S.S., Tensegritic Shell Behaviour, Journal of the ‘Structural Div., ASCE, Vol. 108, No. ST, Aug. 1982, pp. 1831-1845. 18, Motro, R., Tensegrity Systems and Geodesic Domes, International Journal of Space Structures, Vol. 5, No. 3 & 4, 1990, pp. 341-351. 19. Emmerich, D., Self Tensioning Spherical Structures, Single and Double- layer Spheroids, International Journal of Space Structures, Vol. 5, No. 3 and 4, 1990, pp. 353-374. 20, Hanaor, A., and Liao, M.K., Double-layer Tensegrity Grids, Static Load Response I, Analytical Study, Journal of Structural Engincering, ASCE, Vol. 117, No. 6, ‘une 1981, pp. 1660-1674, Hanaor, A., Double-layer Tensegrity Grids, Static Load Response Tl: Experimental Study, Journal of Structural Engineering, ASCE, Vol. 117, No, 6, June 1991, pp. 1675-1684. 22. Vilnay, 0., Cable Nets and Tensegrity Shells: Analysis and Design Applications, Ellis Horwood Ltd., Chichester, 1990, p.130. 23, Hanaor, A., Aspects of Design of Double-layer Tensegritic Domes, International Journal of Space Structures, Vol. 7, No. 2, 1992, pp. 101-113. 24. Grip, R., The Correspondence between Convex Polyhedra and Tensegrity Systems: A Classification System, International Jounal of Space ‘Structures, Vol. 1, No. 2, 1992, pp. 115-125. 25, Pallegrino, S., A Class of Tensegrity Domes, International Journal of Space ‘Structures, Vol. 7, No. 2, 1992, pp. 127-142. 26. Connelly, R., ant Whiteley, W, The Stability of Tensegrity Frameworks, International Journal of Space Structures, Vol. 7, No. 2, 1992, pp. 153-163. 27, Motro, R., Tensegtity Systems - Past and Future, Intemnationcl Symposium. on Shell & Spatial Structures, Nov. 97, Singapore, pp. 69-86, 28, Kono,¥., and Kunieda, H., A Class of Double-layer Tensegrity Grid Domes, International Symposium on Shell & Spatial Strurtures, Nov. 87, Singapore pp. 455-464. 2 Tensegie Structures 29, Wang, B.B., New Concepts in Space Structures: I, Tensegrity and Cable Dome, International Symposium on Shell & Spatial Structures, Nov. 97, Singapore, pp. 849-856. 30. Wang, B.B., New Concepts in Space Structures: I, Cable-strut and Space ‘Truss, International Symposium on Shell & Spatial Structures, Nov. 97, Singapore, pp. 857-862. 31, Hanaor, A., Developments in Tensegrity Systems: An Overview, Proceedings of the Fourth International Conference on Space Structures, University of Surrey, Sept. 1993, pp. 987-997. 32. Nestorovic, M., Analysis of the Properties of Integrally Tensioned ‘Tensegrity Domes, Proceedings of the Fourth International Conference on Space Structures, University of Surrey, Sept. 1993, pp. 1008-1016. 33, Special Issue on Tensegrity Systems, Guest’ Editor: Rene Motro, International Journal of Space Structures, Vol. 7, No. 2, 1992. 34. Morro, R., Forms and Forces in Tensegrity Systems, Proceedings of the Third International Conference on Space Structures, Univ. of Surrey, Sept. 1984, pp. 283-288. 35. Motro, R, Structural Morphology of Tensegrity Systems, International ‘Journal of Space Structures, Special Issue on "Morphology", Lalvani, H. Guest Editor, Vol. 11, No. 18 2, pp. 233-240. 36, Hanaor, A., Prestressed Pin-jointed Structures - Flexibility Analysis and Prestress Design, Computers and Structures, Vol. 28, No. 6, 1988, pp 757-169. 37. Snelson, K.D., Continuous Tension-Discontinuous Compression Structures, US Patent 3,169,611, Feb. 1965, 38. Morro, R., Tensegrity Systems for Double-layer Space Structures, Topping, BILN, ed,, Proc. International Conference on the Design and Construction of Non-Conventional Structures, Civil Comp. Press, London, 1987, Vol. 2, pp. 43-52. 39, Vinay, 0,, Design of Tensegric shells, Journal of Structural Engineering, ASCE. Vol. 117, No. 7, July 1991, pp. 1885-1895, 40. Motro, R., Tensegrity Systems: The State of the Art, International Journal Of Space Structures. Vol. 7, No, 2, 1992, 75-83. 41. Geiger, D.H., A Cost Comparison of Roof Systems for Sports Halls, Bulletin of IASS, Vol. XXIX-1, No. 96, April 1988, pp. £-18. 42. Geiger, 'D.H., A Cost Evaluation of Space Trusses of Large Span, Engineering Journal, American Institute of Steel Construction, New York, ‘April 1968. 43. Geiger, D.HL, Roof Structure, U.S. Patent No. 4, 736, 553. 44, Geiger, D.H., Stefaniuk, A., and Chen, D., The Design and Construction of ‘Two Cable Domes for the Korean Olympics, JASS Symposium on Shells, Membranes and space frames, Osaka, 1986, Vol. 2, pp. 265-272, 45. Geiger, D.HL, etal, Design Details of an Elliptical Cable Dome and a Large Scale Cable Dome’ 210 m Under Construction in the US.A., International ‘Symposium on Innovative Applications of Shell and Spatial Forms, Bangalore, Nov. 1988, pp. 919-930. 46, Kenner, H., Geodesic Math and How to use it, Univ. of California Press, Berkeley, 1976.a7, 49. 51 52, 53, 54, 55. 56. 57. 58. 59, 60. 61 62, 6. 64, Space Structures: Principles and Prastice Vilnay, 0., Characteristics of Cable Nets, Journal of Structural Engineering, ASCE, Vol. 113, No. 7, July 1987, pp. 1586-1599. . Pellegrino, S., and Calladine, C.R., Matrix Analysis of Statically and Kinematically Indeterminate Frameworks, International Journal of Solids and Structures, Vol. 22 No. 4, April 1986. pp. 409-428. Vilnay, O., Tensegric Shells, Shell Structure-Stability and Strength, R. Narayanan, ed., Applied Science Publishers, London, UX., 1985. Taniguchi, T, Ishi, K., Toda, 1, and Komatsu, K., Report on Experiments concerning Tension Dome, Proc, IASS Int. Colloquium on Space Structures for Sports Buildings, Beijing, China, 1987, Yamaguchi, et al., A Study on the Mechanism and Structural Behaviour of Cable Dome, Proc,, IASS International Colloguium on Space Structures for Sports Buildings, Beijing, China. 1987. Makowski, Z.S., Three Dimensional Structures, in view of their ‘Achievements in the Past, can we forecast their Influence upon Modern Engineering and Lightweight Large Span Structures?, LASS International Symposium on Shell and Spatial Structures, Singapore, Nov. 97, pp. 55-68. Ishii, K., Membrane Structures in Japan, SPS Publishing Company, Tokyo, Japan, 1995, Berger, H., New Developments in Lightweight Dome Structures, Bulletin of JASS, Vol. 30. No. 101, Nov., 1989, pp. 135-145, Berger, H., Shaping Lightweight Surface Structures, IASS International ‘Symposium on Shell and Spatial Structures, Singapore, Nov. 97, pp. 1-14, Hanaor, A., Double-layer Tensegrity Grids as Deployable Structures, Int. ‘journal of Space Structures, Vol. 8, No. 1 & 2, pp. 135-143, Hanaor, A, Geometrically Rigid’ Double-layer Tensegrity Grids, Int Journal of Space Structures, Vol. 9, No. 4, 1994, pp. 227-238. Kawaguchi, K., Oda, K., and Hangai, Y., Experiments and Construction of Truss Structure Stabilized by Cable Tension, IASS International ‘Symposium on Sheil and Spatial Structures, Singapore, Nov. 97, pp. 421-428. Hangai, ¥., Kewagueki, M., and Oda, K., Self Equilibrated Stress system and structural behaviour of Truss Structures Stabilized by Cable Tension, Int. Journal of Space Structures, 1992, Vel. 7, No. 2, pp. 91-99. Fawaguchi, etal, Structural Tests on a Full-size Suspen-Dome Structure, JASS Int. Symposium on Shell and Spatial Structures, Singapore, Nov. 97, rp. 431-438. Jing, TE, and Levy, M.P, Shanghai Stadium - China's First Fabric Roof, JASS Int. Symposium on Shell and Spatial Structures, Singapore. Nov. 97, Ep. 355-360. Zuyan, S,, etal, Experimental Study on the Spatial Roof Structure of 80,000, Seat Stadium in Shanghai, Proceedings, Advances in Steel Structures, Hong Kong, Dec. 1996, Pergamon Press, pp. 601-606, Wang, B.B., RP Systems - Conventional Application, Proceedings, Advances in Steel Structures, Hong Kong, Dec. 1996, Pergamon Press, pp. 303-307. Mamoru, K, Masaru, A, amd Tkuo, T, Design, Test and Realization of ‘Suspend-dome” System, Journal of the International Association for Shell ‘nd Spatial Structures, No. 131, Vol. 40, 1999, pp. 179-192. Tersegitie Suctures 65. Obre Bsk, J.B, Fahema, A.H., Advantages in Shape and Form Finding for ‘Wide Class of Space Bar Structures, Sructual Engineering World Congress (SEWC), Structural Engincering World Wide, San Francisco, 18-23 July, 1998, Elsevier, Amsterdam. 66, Obre Bski J.B, On Procedures for Shape and Form Finding on Example of Some Families of Space Bar Structures. Int. TASS WG 15% Structural ‘Morphology Conf. Delf, Aug. 2000, pp. 123-130. 67. Snelson, K, tp wwvegrunch.net/snelson{trigonal html 68. Vassert, N., Morro, R, Multiparamered Form Finding Method: Application to Tensegrity Systems, International Journal of Space Structures, Vol. 14, No, 2, 1999, pp. 147-154 69, Wang, B.B., Simplexes in Tensegrty Systems, Journal of IASS, Vol. 40, No. 1, 1999, pp. 57-64. 70, Wang, B.B, Definition and Feasibility Studies of Tensegrity Systems, Int. Journal of Space Structures, England, Vol. 13, No. 1, 1998, pp. 43-49, 1, Wang, BB., Cable-strut Systems, Part 1: Tensegrity, Journal of Constructional Steel Research, England, Vol. 45, No. 3, 1998, pp. 281-209, ‘72. Wang, B.B., and Li ¥.¥., From Tensegrity Grids to Cable-Strut Gris, Int. Journal Space Structures, Vol. 16, No. 4, 2001, pp. 279-314 73, Wang, B.B., and Li ¥.Y, RP System-Properties, Design and Applications, IASS Symposium, 2001, Poland 74, Kawaguchs, M, Abe, M, and Tatemichi, I, Design, Tests and Realization of Suspend-dome System, Journal of the International Association for Shell and Spatial Structures, Vol. 40, No.3, December 1999, pp 179-192, 75, Buchholdt, H.A., An Introduction to Cable Roof Structures, Second cation, Thomas Telford, London, 1998, 265pp. 16, Zhang, L., Wardenier, I, and Fekhout, A.CIM., Analysis of Steel and Steel / Glas Tensegrity Steel Roof Structures, tevin Report 601.3, Delft University of Technology, 2001 77. Kawaguchi, M., etal, On a Rational Space Structural System Suspend Dome-Design, Construction and Test, First Int. Conf. On Structural Engineering (ICSE1), Oc. 18-20, 1998, pp. 197-200. 7. Campell, D.M, Chen, D., Effects of Spatial Triangulation on the Behavior of "Fensegrity” Domes, Spatial, Latice and Tension Structures, Proceedings of the IASS-ASCE International Symposium, 1994, pp 652-663. 79, Bourdebala, M., and Motro, M., Folding Tensegrity Systems, Proceedings af the International Colloguium on Structural Morphology-Towards the ‘New Millennium, August 1997, Nottingham, pp. 115-122, 80, Djouadi, ., Motre, R., Pons, J. C, and Crosnir B., Active Control of Tensegrity Systems, Journal of Aerospace Division (ASCE), Vol. 11, No. 2, 1998, pp. 37-44. 81. Kono, ¥, Choong, K-K., Shimada, T., and Kunieda, H., An Experimental Investigation ofa Type of Double Layer Tensegrity Grids, Journal of IASS, Vol. 41, No. 131, 2000. 82, Tatemichi, 1, Hatato, T, Anma ¥,, and Fujiworo, S., Vibration Tests on a Full-size Suspen-dome Structure, International Journal of Space Structures, Vol. 12, Nos.3 & 4, 1997, pp 217-224404 Space Sutures: Pinples nd Practice 61, Fest, E, Shea, K, and Smith, LEC, Active Tensegrity Structure, Journal of Structural Engineering, ASCE, Vol. 130, No. 14, Oct. 2004, Bp. 1454-1465, 84, Fest, E, Shea, K,, Domer, B., and Smith, LEC, Adjustable Tensegirty Structures, Journal of Structural Engineering, ASCE, Vol. 129, No. 4 April 2003, pp. 515-526. 85, Wang, B.B., Super Bar Theory and BP System in Super-span Flat Space Frames, Journal of Structural Engineering, ASCE, Vol. 126, No. 11, Nov. 2000, pp. 1365-1370. 86, Moussa, B., Kahla, N, B., and Pons, 1., Evolution of Natural Frequencies in Tensegrty Systems: A Case Study, International Journal of Space Structures, Vol. 16, No. 1, Jan. 2001, 87. Bin-Bing, W, and Yan-Yun, L., From Tensegrity Grids to Cable Strut Grids, “International Journal of Space Structures, Vol. 16, No. 4, Oct. 2001. 88, Selfridge, R.G., About Some Tensegrity Structures, International Journal of| Space Structures, Vol. 16, No. 4, Oct. 2001 89 Kakla, NB, and Mousse, B., Effect of Cable Rupture on Tensegrity Systems, International Journal of Space Structures, Vol. 17, No. 1, Jan. 2002. 90. Fu, F, Structural Behaviour and Design Methods of Tensegrity Domes, Journal of Constructional Steel Research, Vol. 61, No} 2005, pp. 23-35, 91, Kebiche, K, Kazi-Aoual, M.M., and Motro, A.. Geometrical Non-linear Analysis of Tensegrity Systems, Engineering Siructures, Vol. 21, No. 9, 1999, pp. 864-876. 92 Williamson, D., Skelton, RE, and Han, J, Equilibrium Conditions of a Tensegrity ‘Stricture, International Journal of Solids. and Structures, Vol. 40, 2003, pp. 6347-6367. 93, Tibert, A.G., and Pellegrino, S, Review of Form Finding Methods for Tensegrity Structures, International Journal of Space Structures, Vol. 18, No, 4, Dec. 2003, pp. 209-223. 94, Motro, Rand Reducanu, V., Tensegrity Systems, International Journal of Space Structures, Vol. 18, No. 2, May 2003, pp. 77-4 95. Kitiportchai, S., Kang, W., Lam, H-F, and Albermani, F, Factors ASfecting the Design and Construction of Lamella Super-Dome Systems, Journal of Constructional Steel Research, Vol. 61, No. 6, June 2005, pp. 764-785, Analysis of Skeletal space frames 10.4 Introduction ‘The analysis of the three types of space structures, viz., skeletal, stressed skin and suspended structures are different from one another It is difficult to explain the analysis of all these structures in a single volume. Hence here the analysis of skeletal space frames, which are widely used than the other two systems, is only considered. The details of the analysis of stressed skin systems and suspended roofs can be had from references 8 and 30 respectively. ‘A number of analytical tools are currently available for the analysis of plane skeletal structures. These methods apply equally well to the specific area of skeletal space systems. However, the three dimensional nature of these structures complicated by the large number of structural elements and high degree of indeterminacy have made ‘he use of their analysis tedious and time consuming. Hence special analytical techniques hhave been developed in the past which are particularly useful for the analysis of lattice space frames. These analytical techniques can be subdivided into eleven general categories, viz., discrete field analysis, space frame analysis, equivalent continuum analysis, norlinear analysis, experimental analysis, dynamic analysis, stability analysis, limit states or plastic analysis, optimal analysis, thermal analysis and progressive collapse. We shall now discuss these eleven types briefly. 10.2 Available analytical methods 10.2.1 Discrete fleld analysis ‘This approach is closely analogous to the closed form solutions employed in classical plate and shell theories for continuous surfaces. However, the mathematical model consists of difference equations as opposed to differential equations, which are associated with the continua. There areSpace Suructutes: Principles and Practice currently two distinct procedures for a field analysis of lattice structures ~ ‘the ‘micro’ and the ‘macro’ approach. While using these approaches, care rust be taken to see that the boundary conditions are exactly satisfied. ‘The description and comparison of these two procedures have been presented by Dean and Avent (Ref. 11) and are outlined briefly in this section. ‘The mathematical model for a micro analysis is derived by analyzing the basic lattice element and relating its behaviour to that of adjoining elements. The resulting recurrence equations can be expressed as one or ‘more partial difference equations having constant coefficients. Various techniques are available for solving these equations in closed form. ‘The macro analysis requires the solution to only the large or overall members of the structural system. This approach is characterized by the generation of mathematical model in the form of summation equations. The solution to the set of summation equations is determined through the use of orthogonality properties of the series functions found in the overall member analysis. ‘The micro or difference approach requires the general boundary solution to the element with the given element load, if any. An open form. solution results in a sparse coefficient matrix. The macro or summation approach, on the other hand, requires a general load solution to the ‘major components with the given extemal boundary condition and the ‘open form solution results in a full coefficient matrix. ‘The state-of-the-art of discrete field analysis of latticed structures may be found in references 31, 118, 119 and the illustrated use of micro and macro field methods in reference 11. 102.2 space frame analysis, space frame analysis is most commonly used for the analysis of latticed frameworks. The earliest papers associated with space frame analysis predate computers and deal with relaxation methods applicable to hand solutions. Moment distribution, reaction distribution and the method of relaxation of constraints were also used to solve relatively small systems. ‘The background of currently available computer oriented methods lies in the basic relationships required for either a stiffness or flexibility approach, Although not dealing exclusively with skeletal space structures, a number of references are available which provide a sound treatment for the basic topics of force-deformation relations, equilibrium equations and compatibility requirements for interconnected members in space {see Ref. 2,10}. These fundamental relations have been utilized by many investigators to develop stiffness or flexibility analyses for pin connected spece trusses and rigid jointed space frames with straight or curved members. ‘The stiffness method of analysis has got definite advantages over flexibility method, especially when computers are used for obtaining ‘alysis of Skeletal space ames the solution. The usual procedure for stiffness analysis consists of first formulating the member stiffness matrices with respect to local member coordinates. Then these local stiffness matrices are transformed to system or global coordinates and the stiffness matrix for the entire system is generated. Joint deformations are obtained by solving the resulting set of simultaneous equations. The details of this type of analysis are explained in sections 10.4-10.9. Since a very large number of simultaneous equations are often generated with space frame analysis techniques, efficient equation solving procedures can be of great use. Methods of handling large matrices Cfficiently, such az utilizing the sparseness characteristics associated with the stiffness matrix, or various manipulations involving core and secondary storage in the computer have been developed (for instance see Refs. 10, 62 and 68). A recent review of the methods based on finite clement discretization (matrix methods) is provided in Ref. 119. For linear analysis, the maximum possible load in a member is defined by its buckling load in compression or its tensile yield load, Eccentric loading, temperature increases (and decreases to some extent), lack of fit during erection, taper in the members and member end configurations at joints may also influence the response of a member. In the linear-elastic analysis, the initial lack of fit or temperature effects may be approximated by treating them as “pseudo loads” using basic structural mechanics. 10.2.3. Equivalent continuum analysis The idea of approximating the behaviour of latticed structures by considering a similar continuum structure has also gained popularity. ‘This method of analysis has been considered as an alternative to space frame analysis especially for analyzing large systems which require costly computer time. Using this method, certain types of reticulated shells and single and double-layer grids have been analyzed by treating them as continua and applying the shell or plate analogies. Moments, shears, forces and deflections obtained have been considered to form an equivalent system to the stress resultants in the real structure and have been used to obtain approximate values of them. There are many limitations in these techniques, eg. plate analogies cannot take into account the variations of the edge members which can introduce significant modifications and stress disturbances in the real structure. This is an area of great interest to practical designers and numerous papers have been published on this topic; in some cases the results are produced in the form of tables or graphs. Hendry and Jaeger™ have developed a model to replace the cross beams of grid type structures to a continuous medium. Ray*” approximated a grid to an orthotropic plate. Space trusses have been analyzed by treating them as beams or plates™?, 497Space Structures: rnp and Practice ‘These approximations have also been applied to spherical domes, hyperbolic paraboloids and other complex space frames" *. Some of the techniques are explained in chapter 12. A recent review of the methods: based on analogous continuums is provided in Ref. 118 and 119. 10.2.4 Nonlinear analysis Nonlinear analysis techniques for skeletal space frames have received relatively little attention. Two types of nonlinear behaviour, iz, geometric and material, have been studied to some degree by various authors. ‘The main lattice type considered for nonlinear analysis, however, has been the grid. Several authors have presented numerical and iterative techniques for the analysis of grid structures. In the case of a single layer space frame, geometrical nonlinearity Flays an important role as shown in Fig. 10.1. The snap through of a single joint is a typical example. In the case of a plate-like double-layer space frame, collapse lines usually form in the upper chord members and then failure occurs, at which time a rapid reduction in the load carrying ‘capacity takes place. In such cases, limit analysis can be applied. Geometric nonlinear effects such as large displacements and rotations can significantly alter the equilibrium of the double-layer grids resulting in a non-functional structure and possible failure. Geometric imperfections such as, out of straightness of members and eccentricity of loading tend to amplify nonlinear behaviour. In general, nonlinear problems may be treated as a series of linear problems using a variety of methods which may be categorized as follows: (1) incremental; (2) iterative; and (3) a combination of incremental and iterative procedures. These three methods require a definition of the stiffness of the structure in terms of elastic, inelastic (material nonlinea), and geometric nonlinear ‘components (in some cases nonlinear effects may be treated as loads). Materia nonlinear Geometrically noone Fig, 10.1 Nonlinearity and space frame behaviour I space frames ‘Analysis of Stl Techniques such as the Euler method, Secant method, Newton-Raphson. or modified Newion-Raphson methods, and the Broyden-Fletcher- Goldfarb-Shanno (BFGS) method may be used to solve the modelled. ‘numerical problem", Many authors have also developed solution techniques for space frames having both geometric and material nonlinearities. Oden'® used. a network-topological formulation of the matrix force method, whereas Gesund'™! addressed the same problem from a different view point. ‘The author has analyzed steel frame folded plates using geometric nonlinear analysis which is based on the stiffness method*. Jagannathan ct al. employed a nonlinear finite element method based on a Lagrangian formulation to study the geometric nonlinear behaviour and associated snap-through buckling of reticulated structures#*. Using this technique they solved large Schwedler and Lamella domes. A recent review of the various methods available for the nonlinear analysis of space frames is, given in Ref. 118, 10.2.5 Experimental analysis, Owing to the uncertain accuracy associated with various numerical and approximate analytical methods, experimental methods are often used to verify the theoretical results. Descriptions in the literature of experimental work have almost been related to a specific design project. A summary of experimental research is tabulated in Table 10.1. Experimental models have been compared to theoretical formulations with varying degrees of success and the interested reader is requested to refer to the appropriate reference for specific details. A recent review of the experimental methods is provided in Ref. 119. 102.6 Dynamic analysis The main differences between static and dynamic analysis of structures are the time-varying nature of the load and the structural response under dynamic loading. In addition to checking the safety, strength and stability of space structure under static loads, the response of the structures must also be checked under possible dynamic loads throughout their service life. An understanding of the actual behaviour of individual ‘members forming a truss or framed structural system is absolutely vital to obtaining overall system response and assessing the system integrity Jn a service environment. The effects of dynamic loads can be observed in response amplification either through interaction with the natural dynamic characteristics of the structure or in structural fatigue due to repetitive loads. The former does not appear to be significant in collapse failures of latticed structures but the latter may contribute to serviceability falures in roof structures. Loads: Dynamic loads produce a time-dependent response which can adversely affect the integrity ofthe structure. Dynamic loads, which 499Table 10.1 Summary of experimental analysis to skeletal space frames Space Structures: Principles and Practice ‘Analysis of Skeletal space frames ’ should be given attention while design alyzing double g & layer structures, include wind, seismic excitation, machine movement 2 28 and vibrations, impacts and blasts. The action of dynamic loads may g g s2 ¢ eetet BF be represented as concentrated, distributed, or partially distributed. = soe eee ee Earthquake and wind, the two most important naturally occurring dynamic loads, are known only in a probabilistic sense. Typical wind B§ Toad effects include horizontal drag, upliR, and vortex shedding and 3: |2 2 direct pressure (positive and negative). In the case of the seismic ef le 2 excitation, horizontal, as well as vertical (differential) ground motion, 2Ee/2 Fe 2 gy ee gy must be taken into account. £2s)2 2 £ #2 £2 bE Dynamic loads vary not only with time, but also with location within Sea/2 22 &£ EE £2 EE the structure, thus giving rise to a structural response which varies eealS Se 2 eysee ge along and across the structure. Furthermore, since the structure may be eis|e F2 & SS25e Be ‘under the simultaneous action of several loads (both static and dynamic), the response to different probable combinations of load must be o — investigated, Design codes provide appropriate combinations of loads 3 a3 to consider in analysis and design. g Bs Dynamic Analysis: Space Structures, especially double-layer grid 2 os 2.448 #8 structures, can be very large in plan, therefore, relatively flexible and may & 22 EEE EE have a wide range of relatively low natural frequencies. Thus, their a Ss AGEaG ae natural periods may be close to those of extemal dynamic excitation, 3 € ¢2 2 g¢dd ¢€ making the structure even more sensitive to dynamic effects. The dynamic 2s 88 3 s5383 88 response of a structure depends on the nature of the extemal dynamic fi |@ BS ge SEVSE Fe characteristics of the structure. The sophistication of the analysis used is a function of how accurately these parameters are known. Various techniques are available to determine the stifiness and mass distributions z » of a structure, In several cases, a complex system can be represented : edevseeeese by a simple model with the help of appropriate assumptions! E gegelegeseeis ‘The damping properties of the structural system, including the S BeSze ot eG geo eks support medium, affect the dynamic response greatly. Damping usually k |syodskesgessegs results from material hysteretic behaviour, joints and connection fiction, = SRSR EK RSET ES and other energy dissipation schemes which may be specific to different proprietary systems. The energy disipation and input motion due to soil g structure interaction, especially for a structure supported by a deep a 3 Eg foundation, should also be considered. In general, it is difficult to quantify. 2 Es e damping exactly in a structural system. Simple models for damping, 8 EB 22 3 bee such as the dash-pots model (viscous damping) and the proportional ek IE = Sf 2 25a damping model are usually used. Good summaries along with the Be |e 2 222 282 extensive reference lta ven by Sherman? Tool and Male &E | 2 #2 & 22s and Serette!®, “ e - 7 ane Analysis Methods: As in static analysis, the basic mathematical . ‘model for dynamic analysis must include relations for (1) dynamic 3 equilibrium equations (or equations of motion which include the G = a ome wo on inertia effects due to motion in the structure); (2) strain-displacement‘Space Strctures Principles and Practice (compatibility) relations; and (3) stress-strain (constitutive) relations. Thus, the main difference between the dynamic and static analyses is ‘that the dynamic analysis takes into account the inertia effects leading. to equations of motion. For small displacements and low loads (service conditions), the ‘material will exhibit an essentially elastic behaviour. In such cases, a linear analysis suffices to predict the behaviour of the structure. On the other hand, if the displacements are finite, or ifthe stress-strain relations of the material are not linear, or the material is subject to a large number of cycles, a linear analysis will tend to under-estimate displacement and over-estimate strength. In this case, an appropriate dynamic nonlinear analysis will be necessary. Dynamic Analysis for Wind Loads: Wind loads are a function of wind velocity, density of air, area of structure exposed to the wind, height and roughness of the ground surface, topography ofthe site, and configuration of the structure. The design of space structures, due to their large size and light-weight characteristics, is often governed by wind loads. The wind loads on the structures may cause flutter of components and vibration due to vortex shedding. Dynamic analysis is used to determine whether resonance will occur and how to reduce resonance effects due to wind loads. For tubular members subjected to wind pressure, aeroelasticity and aerodynamic aspects must be considered. An additional bending stress combined with axial stress must be taken into consideration in the analysis and design of the members. The existence of a pressure differential between intemal and extemal surfaces should also be considered. Dynamic Analysis for Earthquake Loads: Analysis of structures subjected to ground motion (earthquake excitation) is very complex in general. The intensity, magnitude and duration of the earthquake excitation for a particular structure are not known exactly and are ‘generally based on past experience. As a result, the amplitudes of ground accelerations, velocities and displacements ‘can not be determined precisely and the characteristics of the earthquake excitation are mostly represented by approximate response spectra or Fourier amplitude spectra. For these reasons, several simplified analytical methods have bbeen devised to analyze a structure under seismic excitation. One method, which is commonly used for simple and regular structures, is based on the quasi-static model, in which the dynamic action of the ground motion is represented by equivalent static loads (see also chapter 12). For more complex structures, detailed dynamic analysis is warranted and one simple technique that may be used is the response spectrum method. For the response spectrum method, the design earthquake is determined from site-specific seismic activities (location of the site on earth and properties of bedrock and soil). “Analysis of Skeletal space frames For space structures, consideration should be given to incorporate the effects of the differential ground displacements in the structure. The effects of the soil-structure interaction and possibility of soil liquefaction. (where appropriate) should also be considered. The structural response also depends on the direction of the ground motion input. Although the horizontal component of ground acceleration is usually given more consideration, in several roof structures the vertical component can also be significant'®*: Some other useful references on the analysis of earthquake response are Tsuboi"* and Karamanos and Karamanos"®. It is speculated that the dynamic stability/instability can play a significant role in the collapse of truss and frame-like structures, the extent of which is not yet known": ‘Dynamfe Nonlinear Analysis: For structures which undergo large deformations or for which the material behaviour goes beyond the linear-clastic region, the mass, stiffness and damping characteristics of the structure become functions of displacements, velocities and accelerations, which, in turn, are time-dependent. Thus, in nonlinear analysis these quantities must be updated continuously with time as the structure deforms. This makes the nonlinear analysis tedious and time-consuming. Except for a simple structure, nonlinear analysis requires special numerical solution techniques. Some useful references on this topic may be found in Ref. 119. 10.2.7 instability analysis Instability of a single compression member can result in a progressive collapse of the structure through load redistribution and successive buckling or yielding of other members. Thus detailed attention must be given to (i) factors affecting the buckling (member slenderness, imperfections, joint rotation and axial load eccentricity, flexural stiffness at the member end and member end fixity), and (2) redistribution of load after the first member buckles. In some designs, the dimensions of a member may be reduced at its ends to facilitate a particular type of joint, thus leading to a decrease in the buckling load of the member. In space frames with welded joints, the designer should be aware of possible high residual stresses that can result from the welding process. Preloading of members due to misalignment or lack of fit during erections may induce large stresses in members thereby reducing the loal required ‘o initiate buckling, Significant reductions in load carrying can result from these effects!" ‘The models used for the Instability analysis may be clessified as ‘Continuous mathematical models (Conventional shell theory, Equivalent shell theory, Equivalent stiffness method and Split rigidity method) and Discrete mathematical models (finite difference formulation, matrix formulation and physical model). The descriptions of these methodsSpace Structures: Principles ard Practice “Teddi wih sine ‘Sihtetced ene Fig. 10.2 Effective length of members are outside the scope of this book and an interested reader may refer to Refs. 31, 118, 119 and also to the references given at the end of this chapter for further details. However the discrete mathematical model based on matrix formulation is explained in section 10.12. Several researchers have also proposed methods to predict the post buckling strength of space frames"®, Effective Lengths: The member end restraints are fundamental to determine the buckling strength of a member. Ideally, the member end restraint will be anywhere between being fixed and pinned. Members with pinched section ends at the nodes have the largest effective length factors, k (see Fig. 10.2). Members with full sections at the node typically hhave the smallest effective length factor (particularly members with ‘welded ends), and members with reduced ends have effective length factors which fall between the pinched and full section end values'?, As an overall guideline for members with full section, reduced section, and pinched ends, the following effective length factors have been suggested for use in elastic stability analysis: Full section en reduced section ends: k= 0.90; pinched section ends: k= 0.95. In addition, Madi and El-Tayem'? have suggested that the effective length of double-layer grid compression members in the top chord also depends on whether or not buckling is in or out of the plane of the chord. For members where both ends are assumed to have a high degree of restraint, Madi and El-Tayem suggest using k = 0.76 for in-plane chord buckling and k = 0.67 for out-of-plane chord buckling. For members where one end is assumed to have a high degree of restraint and the other end is pinned, the authors suggest using k = 0.81 for in-plane chord buckling and k = 0.72 for out-of-plane chord buckling. The lower value for out-of-plane buckling results because the web Analysis of Skeletal space frames Table 102 Effective length factor (k) for compression chord members in double-layer grids!®, Location Pec IPey Fixity at sh Ty Mle Pp lPe end of Plane of Action of buckling member double- layer grid 077 FES 0.72 FES 0.86 FES 0.69 FES 10 10 10 10 0.81 IST 10 10 0.0 =1.0 FFS FES FES 0.87 0.64 0.64 10 10 10 14l4 1414 1414 'FFS-far from supports; [SI-Lower support joint h-height of space fame module; +-sze(aquare diension) of base module; Ip-moment of Not force in the chord member inthe y-direction mer; [-~moment of ineta ofthe chord member; Py axial force inthe diagonal web member; axial force in the chord axial force in the chord member in the x-direction; and ., am Inertia ofthe diagonal ms rember; P, 50Space Souctures: rincples and Practice members are more effective in providing restraint to out-of-plane deflection. ‘Murtha-Smith and Bean"? have suggested that the effective length of compression members in double-layer grids depends on the relative ‘moments of inertia of the chord and diagonal (web) members, the ratios of the magnitudes of the axial forces in the chord and diagonal members, the angles at which members frame into the joint, and the flexibility of the joint itself. For elastic stability analyses of concentricelly Joaded members in ‘square-on-offset-square’ double-layer grids, Murtha~ Smith and Bean recommend the effective length factors as given in Table 10.2. ‘The effective length factors given in Table 10.2 do not include che torsional restraint of members, which may have a tendency to reduce the effective length factor. When estimating the effective length of the ‘member, itis recommended that the actual length of the member (not the distance between work points) be used with the effective length factor". However, for a conservative estimate of capacity, it is appropriate to use the work-point distance. ‘Various investigators have used various degrees of approximation of the compressive member behaviour in the actual analysis of truss and frame structures and Murtha-Smith presents a review of the availasle methods! Joint Behaviour, Lack of Fit, and Member Imperfection: In practice, structural members always have some initial geometric imperfection, end eccentricities, or crookedness, and joints have some flexibility. These factors can have a significant effect on the behaviour of space frame systems. In general, imperfections in structures cannot be determined accurately, but their effect can be significant, Rotational joint instability may be caused by reduced member stiffness at the joint, eccentricity of lozds ‘ing on the joint, or by unbalanced loads at the joint. For a typical joint in the double-layer grid structure having many framing members, ‘excessive joint deformation or instability will result in a significant reduction in capacity of every member connected to that joint. ‘When compression members possess ‘brittle-type’ characteristics, test results have indicated that the load-carrying capacities of double- layer grids are 25-40% below the theoretical elastic limit. The reduction {in capacity has been attributed to imperfections such as joint rotations, Joint slippage and member lack of fit. A review of the methods available to induce these factors is discussed in Ref. 119. Italso includes discussion fon the methods for limit state design and for predicting the collapse mechanism. Buckling and Collapse of a Joint: The buckling and collapse of a joint, such as the dimple buckling of a ball joint under element compression forces, may cause a difficulty in the transfer of forces Analysis of Skeletal space frames through the joint as well as a change in the geometry of the overall structure. For this reason, especially when the degree of statical indeterminacy is low, it may be necessary to make a careful and ‘complete study of this problem"®, Local Buckling: The local buckling of space frames is also termed ‘joint buckling’ and, as indicated in Fig. 10.3a, consists of a snap ‘through buckling which takes place at one joint. Snap through buckling. is characterized by a strong geometrical nonlinearity. Local buckling {s apt to occur when the ratio of t/R (where t is the shell thickness, and R is its radius of curvature) is small. Similarly, local buckling of a space frame is likely to occur in single-layered space frames. Local buckling is greatly affected by the stiffness and the loads on the adjacent elements and the flexural strength of the joint. ‘As shown in Fig, 10.3b, there may exist some rare occasions when the determination of the allowable displacement D, is more important than obtaining an estimate of the buckling load. This condition is more likely to occur if the mesh size of the space frame is fine. ‘An approximate formula for the local buckling load was proposed d, applicable to triangular networks with all the elements of the same cross section", For the uniform load shown in Fig. 10.4a, the critical oad is 2., E AP, 4B alee) (o«745-+32t) (on) where o = P/(rR), r is the radius of gyration, A is the area of an clement, B is a non-dimensional bending stiffness of the grid near a joint given in Table 10.3, I is the moment of inertia, R is the radius of ‘curvature of the mid-surface of the framework, | is the length of an clement, his the rise and E, is the tangent modulus. For the concentrated load shown in Fig. 10.4b, the following two formulae are presented. BEAR? /8B BB jon HAP (82 coan(s-is9s88)) on valid as eater han 9 appoint and W,,= 0.0905 EA P/R? (10.3) valid for a regular pin-jointed structure. sorSpace Structures: Principles and Pacce aan { RR DOLLS OF o » Fig. 10.3 Locel buckling and its behaviour Table 10.3 Equivalent bending stiffness B o B 1/32 0.868 ans 0.873 1/8 0.886 14 0.95 a2 1.176 1 185 2 315 4 4.83 8 6.48 16 735 32 7.80 64 7.90 o Q @ \ tof IS » ® | Fig. 10.4 Two kinds of loading 10.2.8 Limit state or plastic analysis ‘There is an evident trend towards application of yield line theory in the design of flat grillages and limit state design (or plastic design) is receiving an increasing amount of attention. Researchers during 1980s realized the importance of the pioneering work done by Heyman at Cambridge University, who demonstrated the general methods leading to the determination of the upper and lower bounds of the collapse load of transversely loaded single-layer grids. Heki, Saka, Schmidt, Cogan, Reiss and Gregorian (see Ref. 61) have extended these concepts to the limit analysis of double-layer grids. There are five alternative approaches for the limit analysis of grids: (0 ‘step by step’ clastic plastic analysis carried out by computers; (i) linear programming; (ii) moment balancing coupled with the finite difference analysis; (iv) continuum transformation (e. yield line theory or strip method) and (v) load sharing method. The details of these methods can be had from Ref. 61 10.2.9 Optimal analysis Numerous attempts have been made to obtain optimum configuration of space structures. Attempts have also been made to find the minimum ‘weight of these frames as a function of certain pre-assigned parameters such as span, support conditions, external loading and structural topology. To obtain an optimum structure in terms of minimum cost is much more difficult, as in addition to the above parameters, one has to include in the optimization approach, the influence of the cost of joints and erection techniques which may have an influence on the final cost of the finished structure. Several important contributions have also been,s10 Space Structures: Principles and Practice ‘made to split up the optimization problem into a number of independent problems.” 10.2.10 Thermal analysis Uniform temperature changes anc temperature gradients can affect the structural response of the space structures significantly. This is especially true for a light weight structure built in extreme climatic environments. Extreme temperature differentials, due to fire, should also be given attention. Some work reported on the effects of fire load and large ‘temperature differential on double-layer grids and truss roofs is described briefly in Ref. 119, 10.2.14 Progressive collapse ‘Several situations, such as excessive vibration, initial lack of fit, impact, accident, material and connection defects and buckling phenomenon, may inflict physical damage or may cause large deformation or overstressing in a member or a portion of the structure. Although the member damage (local damage) affects a small portion of a structure initially, it has potential for propagating to other parts of the structure and may ultimately cause total collapse of the structure. This type of failure mechanism has been termed ‘progressive collapse’ (Also see section 13.7). The collapse mechanism results from the redistribution of Toad when one member fails causing a subsequent overstress condition in another member. Initial lack of fit may cause overstressing in ‘members, particularly in 3D grids, with subsequent unexpected failure at relatively low loads. In double-layer grids where the member sizes are varied to allow as many members as possible to be fully stressed under the worst loading case (optimized design), progressive collapse may not be possible. In this case, several members may fail simultaneously, thereby causing an apparent britle failure of the structure. In practice, truss-type 3D structures have a high number of redundant members, but, collapse of such structures is not uncommon and can have dire safety and economic consequences as exemplified by numerous lattice roof collapses (see section 13.7). There are four modes of single member failure that may trigger a progressive collapse of the double-layer grids: (1) overall buckling in a compression member (with or without postbuckling capacity); (2) yield in a tension member; (3) tension member or connector fracture; and (4) joint rotation instability (due to eccentricities and low member stifiness at the joint ‘Two practical methods for analyzing double-layer grids for progressive collapse include: (1) The member removal method; and (2) the member residual strength method. Both methods follow essentially the same procedure as outlined below: is of Skeletal space frames 1, Analyze the structure (using linear-elastic procedures) to determine which member will fail first (that is which member has the highest stress-compared to its design capacity) 2. Depending on the method being used, either (1) completely remove the member that has reached its maximum capacity and will fail due to overstress - member removal method; or {2} limit the load capacity of the member after it reaches its ‘maximum capacity - member residual strength method. 3. Reanalyze the structure with the removed (zero strength) or reduced strength member and identify the next member most likely to fail 4, Return to the second step and iterate until the structure's capacity is significantly reduced (that is, the structure is about to collapse or can no longer carry the design load). Im applying the residual strength method. the designer/engineer rust devise what constitutes the residual strength of a member. For major structures, itis recommended that full scale member tests be performed, in tension and compression, to estimate the post maximum load behaviour of the double-layer grid members. 10.2.12 Dynamic effects of member failure ‘As mentioned earlier, in a truss-type structural system, where members are primarily carrying axial loads, member failure may take place by ‘yielding (tension) or buckling of compression members. Sudden failure may produce additional structural vibration. During dynamic excitation, a tension yielded member may take up additional load due to strain hardening whereas compressive buckled members lose strength and shed load on to other members. Two major types of sudden member failure which have potential to cause dynamic effects in the structure are: (a) brittle failure, which takes place within the linear elastic material regime, and (b) member snap/dynamic jump phenomenon in post buckling regime. The failure or rupture of members in a lattice truss, which may lead to progressive collapse, is realized to be dynamic in nature. There is very little published literature dealing with the effects of dynamic member failure on the response of the total structure. Recent efforts to develop a systematic approach to determine the response of truss structures due to the dynamic effects of member failure are summarized in Ref. 119. From the above discussions it is seen that the matrix methods of structural analysis provide an extremely efficient means for rapid and accurate treatment of space structures. Matrix algebra is ideally suited for automatic computation and several general purpose computer si512 Space Stuctues: Principles and Practice programs have been developed based on matrix methods of structural analysis. Hence we shall now consider the details of this method in the following section. 10.3 Matrix methods of structural analysis Matrix methods of structural analysis itself is a vast topic and to consider the details of all the aspects of this type of analysis is a difficult problem. Hence only a brief introduction is given here about the general principles of matrix methods. For more detailed descriptions, see References 2, 20, 34, 35, 68, 133 and 134, Matrix methods of analysis are usually subdivided into stiffness or displacement method (in which the basic unknowns are displacements) and flexibility or force method (in which the basic unknowns are forces). In the stiffness method the compatibility conditions are used which give rise to equations of joint equilibrium. Whereas in the flexibility method the equilibrium conditions are fist satisfied which lead to equations of aisplacement compatibility. The basic unknowns of the stiffness method—the joint displacements, are obvious physical quantities associated with the structure. The basic unknowns of the flexibility method, or the other hand, are a certain number of forces ‘and moments which must be chosen according to certain rules. Thus any general computer program based on the flexibility method of structural analysis must include a systematic process for determining the number and location of these forces and moments. Hence the stiffness method has been preferred by many researchers for the analysis of space frames. Another distinctive advantage of the displacement (stiffness) method is that it can be applied with equal ease to statically indeterminate and statically determinate structures. We shall now see: how the equations are formed using these two methods. 10.3.1 The flexibility method ‘As discussed earlier, in this method of analysis the redundants are assumed and removed from the structure so that a stable and statically determinate structure remains. Equations are written for the deflection. condition at the joints for each redundant that has been removed. These are written in terms of the redundants and the resulting simultaneous equations are solved. After the redundants are known, the desired stresses, deflections, etc, are determined by statics, The method of consistent distortions, Castigliano’s second theorem and the ‘three-moment theorem are classical examples of the flexibility method. Taking for example, a structure which is statically indeterminate to ‘the third degree, the flexibility equations can be written as follows: Xion * Xoo + Yoho e Xidoa + Xadboe * Xa Soe analysis of Skeletal space frames X, feat Xabos * Xa fee = 9 ‘These equations can be written in matrix form as below: Soa fat fee XY] [a Soo foo fl %|=|5e (00.4) fie fer feelL%) (8c. or simply as = FK (10.5) In the preceding expression, F is called the flexibilixy matrix; X is the column vector of the correction forces or moments and 6 is the column vector of the errors in geometry (due to deflections and rotations). Inversion of the flexibility matrix will provide a solution to ‘the equations: Xa PS (10.6) 10.3.2 The stiffness method In this method, the displacement of the joints necessary to describe fully the deformed shape of the structure are used in the simultaneous equations instead of the redundant actions as used in the flexibility method. Thus, using this method, the equations for any structure can bbe written as, F=KD (10.7) where F is the matrix containing the complete set of applied joint loads, D is the matrix containing the corresponding set of unknown displacements and K is the stiffness matrix of the structure (It is to be noted that the ‘flexibility’ of a member isthe reciprocal ofthe ‘stiffness’ of the same member). ‘The unknown displacements are determined when the above equations are solved. Its of interest to note that the slope deflection method is a displacement method and the moment distribution method is a successive approximation technique based on the same general theory as the displacement method. Since the stiffness method is well suited (due to its advantages over the flexibility method) for the highly indeterminate space structures, ‘we shall confine ourselves to this method of analysis. 10.4 Basic principles of the stiffness method ‘The displacement method considers a structure as an assemblage of individual elements. The connecting points of these components are designated as nodes. Then the member forces are related to member deformations by a matrix defined as the clement stiffness matrix, designated by the small letter &. The nodal point forces are related tooe Space Stuctues:Prncinies and Practice the nodal point displacements (as shown in equation 10.7) by a matrix defined as the stiffness matrix of the structure, designated by the capital letter K. In solving problems by using the stiffness method the following steps are followed: 1. Forming the element stiffness matrices of the members in the global co-ordinates using the stiffness matrices in the lecal co-ordinates and by using the transformation matrices as shown below: (e1 = (71 04 7 (10.8) where [7] is the transformation matrix 2. Forming the stiffness matrix of the structure from the individual member matrices depend on the way in which the members are connected. Solution of the overall stiffness equation 4F} = [KI {D} (20.9) to get the deformation {D} where {F} is the external load vector, 4. Substitution of the values in the equation {P} = 181 (71 (D} (20.10) to get member force {P} In the above analysis, it is usually assumed that the external loads are in the elastic range. That is, the load-deflection relation is linear and thus obeys Hooke's law. It is also assumed that the warping rigidity of the cross-section and the shear distortion are neglected. The initial stresses such as residual stresses and initial deformations are ignored’. 10.5 Development of stiffness matrix for plane frame members ‘To explain the principles involved in the derivation of stiffness matrices, let us first consider the example of a rigid jointed plane frame. For simplicity we consider the case where all the members are straight and have a uniform cross-section. First, let us consider the load displacement cquations for a single member which is having a length L, area of cross-section A, Young's modulus of rigidity E, and moment of inertia I Let fry Jar fn and fi be the member end forces and m, and m, be the member end moments. Also let &,,, 5, dj, 52 are the member end “if desired these parameters can also be Included in the analysis with minor modification. for details of these see Ref, 3 and 20, ‘Analyse of Shaletl space eames fate the K;, matrices of the members connecting Joint to joints J, K, L... Ifa joint is not directly connected to T by a member, then the associated element is zero (Due to the symmetry of the complete matrix it is only necessary to consider the members which have end 1 at joint I). ‘Taking the plane frame shown in Fig. 10.7 and applying the above rules, we get Fig. 10.7 Example of rigid-jointed frame S19Space Structures: Principles and Practice ike)a . . ty [ler tian Be Oe 0) ay ee yg okey, fe (Kade + (Kivde ie . : ae ey lly fe sino, Kinds | [de Ks). 7 . : : to fs +), In the above stiffness matrix the elements of the upper triangle only are shown. A term on the leading diagonal of the final stiffness matrix represents the ‘direct’ stiffness of the joint, ie., the load required at the joint to produce unit displacement of that joint, al the other Joints being held. fixed. The off-diagonal terms correspond to connections between joints and since there will usually be one member, at most, connecting any: given pair of joints, the off diagonal elements will normally be either single K’,, or K’,, matrices or zero. 10.7 Banding of the K matrix In most of the structures, joints are connected only to their nearer neighbours, and there are rafely more than six to eight members meeting: at any one joint. Hence if a systematic method of numbering the joints is adopted, the difference in the joint numbers at the ends of the ‘members will be small compared with the total number of joints. As a result most of the elements of the stiffness matrix of the structure will lie near the leading diagonal. Such a matrix is called a banded matriz. ‘There are distinct advantages of having a banded K matrix. Less computer space is required to store only the elements of the bard instead of the entire K matrix. Various efficient techniques have been developed to solve the governing simultaneous equations using only the elements of K that are in the band*. To see how a banded matrix is obtained, let us consider the frame of Fig. 10.82. Considering axial and bending deformations in the members ‘we would have three external displacements at each nodal point. For the nodal point numbers as shown in Fig. 10.8a, we would obtain a stiffness relationship of the structure as shown in Fig. 10.6c. In this figure a cross indicates a 3 x 3 submatrix of non-zero elements. If we have numbered the same frame in the way shown in Fig. 10.8b, we might have obtained a tanded matrix as shown in Fig. 10.64. Thus, from the above example it is seen that to obtain good banding, “For the discussion cf these methods see Ref. 10, 34, 84 and 132. nals of Skeletal space frames sot 1 2 1 16 a ‘ 2 dy 5 ‘ 3 3 1 2 © nd wih — 5 a ; abet shel 9 CPBr Eh a She 8 Tee ect eRe P234s678910 Ta34$678910 © o Fig. 10.8 Effect of node sequence on the band width of stiffness matrix Mand — JM Band — (@) Fae toes mate (R] (0) Campa norge of] Fig. 10.9 Efficient storage of frame stiffness matrixSpace Structures: Principles and Practice it is necessary to keep the nodal point numbers as ‘near’ each other as. possible. A general guideline for obtaining good banding for a structure is to number the nodal points across the dimension of the structure that has the least nodal points. ‘The band width MBAND may be obtained by using the relation, MBAND = ID (MB + 1) (10.20) where ID is the maximum number of unknowns at a joint (for a plane frame ID = 3, for a space frame ID = 6 and so on) and MB is the maximum difference between the back and fore node numbers. Thus, for the frame shown in Fig. 10.8a, the band width is 9 and for the frame shown in Fig. 10.8, i is 18 (ti interesting to note that the null (3 x3) arrays within the band of nonzero terms of Fig. 10.8¢ would not occur if the frame were diagonally braced, reflecting the enhanced stiffness of a braced frame in comparison with an unbraced one). ‘A way in which advantage may be taken of the symmetry of the stiffness matrix and the predictability of the maximum band width in order to conserve storage is illustrated in Fig. 10.9. This facility has been made use of in the program listed in Appendix IL Its to be noted that the band nature ofthe stiffness matrix not only reduces the amount of computer storage required but also the number of numerical operations to be performed to obtain the solution. 10.8 Solution to the stiffness equations 40.8.1 Difficulties encountered The solution to large systems of simultaneous equations in a computer may involve the following difficulties: (1) The given coefficients of the system and the results of arithmetic operations are rounded to a limited number of significant figures. The initial inaccuracies and the accumulation of rounding errors, during computation, can seriously affect the solution. This effect may even be such that the results are nothing but a set of irrelevant figures. The annoying feature is that this pseudo solution may satisfy the equations very well. A system of equations susceptible to this kind of trouble is said to be ‘ill-conditioned”. (2) The solution to large systems of equations, in general, calls for ‘vast amounts of storage space and if the computer cannot meet the storage requirements the operations become impossible (Matrix sub-stracturing techniques and the secondary storage of the computer may be used to solve these systems). (3) The solution to large systems is very time-consuming. In some cases, the required computing time may be so long as to make the solution economically or even physically impractical. “A detailed examination of this matter Is found in Ref. 34. ‘naliss of Skeletal space frames ‘The extent to which the solution to a system may involve the above problems depends on the method of solution, properties of the system and the computer used. The solution technique must be so chosen, that for a given system and a given computer the results are obtained in the most economical way and have no less than the required precision. Due to the developments that have taken place in computer hardware, it is now possible to solve very large systems of equations without any difficulty. 10.8.2 Methods of solution ‘A number of methods are available for the solution to these equations and they may be divided into two groups: (@) Direct Methods - Typical examples are: Method of determinants, Gauss elimination method, Gauss Jordan method, Crout method, Doolittle method, Cholesky method (for symmetric coefficient matrix only) and matrix inversion method, (b) Iterative Methods - Typical examples of this method are: Conjugate Gradient method, Gauss Seidel iteration method, Gauss or Jacobi iteration method and Relaxation method. A thorough exposition of these two families of techniques is beyond the scope of this book (The details of all these methods may be found in Refs. 6 and 69). However, the description and comparison of their main features are given in Table 10.4. One of the most widely used ‘direct’ methods of solving linear systems of equations is the Gauss elimination method. This method is, capable of handling ill-conditioned systems of equations, and there is, no need for the coefficient matrix to be either symmetric or positive definite. This method also provides a convenient way of checking the stability of the structural system. It may be shown that if all the leading diagonal elements of the triangular matrix formed by the elimination process are strictly positive, then the original matrix is positive definite. ‘The product of these leading diagonal elements is, in fact, equals to the value of the determinant of the original matrix, and the elimination method is probably the easiest way of computing this quantity if itis required. Hence this method has been adopted in the computer program listed in Appendix Il. Choleski's method is found to be more suitable for processing banded and symmetric matrices and hence has been used by many investigators. 10.9 Stiffness matrices for a straight uniform member in a space structure ‘We shall now consider the more general case of a member in a rigid sa‘A first approximation to the exact solution is improved successively until the required precision is obtained. Is usually used unmodified (or effectively unmodified) throughout the computation Iterative methods Table 10.4 Comparison of solution techniques approximation to the exact solution, Is modified at each step of the AA prescribed sequence of operations computation, (performed only once) yields a solution which is (normally) an Direct methods Procedure of the ‘The stiffness matrix method Description S.No, 2 Space Structures: Principles and Practice ‘Analysis of Skeletal space frames {jointed space structure. We shall limit our analysis to the derivation of the ‘appropriate K matrices since the technique for assembling the complete stiffness matrix of a structure described in section 10.6 is the same whether the structure is a plane-frame or a space frame. ‘As in the previous sections, we shall assume that the member is straight, and has a uniform cross-section. We will also assume that the ‘material of the member obeys Hooke’s law, and that the deformations are le right hand side times the number of right hand sides. if any) of the process of Considerable reduction in the storage Amount of computation will (normally) system and the required precision of requirements. Depends on the conditioning of the the solution, however, do not accumulate. & a END2 Fig. 10.10 Member co-ordinate system for a member of three-dimensional frame sufficiently small for ordinary linear theory to apply. The co-ordinate system which we shall use is shown in Fig. 10.10. We shall assume that the y and 2 axes are the principal axes of the cross-section and denote the flexural rigidity of the beam in the xz and zy planes by EI, and ET, respectively. Since the y and z directions correspond to the principal axes, the flexural behaviour of the member in the 2y and 2z planes may be considered quite independently; the analysis given in section 10.5 being applied to each plane in tum. The equation (10.11) for axial strain will aso be unchanged, so that the only new factor to be considered is the torsional stiffness of the member about its axis. If we consider Depends on the number of operations. However, itis independent of the conditioning of the system or the precision of the results, 2 e 3 z = B E 8 2 requirements may be obtained in certain direct methods, Extra amount of computation is Some reduction in the storage relatively smal § gé s the simple case in which warping effects are ignored, and assume the z zz B28 eee torsion axis of the member to coincide with the centre line (ie, the x S22 28 £2 axis) then the equation for twisting of the beam is simply z ee Eg = B Ses #3 $s My =~ Ma = (GID) (8p, ~ 8) (10.21) 2 83 g8 age where GJjL is the torsional stiffness of the member. Thus, the load~ 5 2:3 £2 gS: displacement equations of the space frame member can be written in = 282 58 324 the usual form Se=Ku dst Kd p= Ky, d,+K,d, (10.22)ne Suuetures: Principles ana Pracbee $$ Spas svostres rinepies ana practice where te 5, h 8, pelle gal & ms o my o, mn, 6 and Lo o 0 ° 0 0 RE 0 eK 8 Ke 0 0 oO GHL 0 oO ° 0 -6EI/ 0 AEE 0 6EI 0 ° 0 4en rE -ENL 0 o 0 0 0 Oo ze 00 went Keemee| ° 0 DEE 0 -6EK/ 0 we 0 0 oO ~GIIL oO o ° 0 ee 0 EKIL 0 ~6EI/Z 0 ° 0 2EL IL. and K,, is equal to K,, with the signs of the off-diagonal clements ~6E1, /L and GEL, |? reversed, We shall now apply the usual technique for transforming (10.22) into global co-ordinates. This transformation may be thought of as the product of three rotation matrices T,, T, and T,, For developing the matrices 7, and T,, we shall consider a’ member having a length L, which is having its orientation as shown in Fig. 10.11 ‘The first rotation is through an angle B about Z axis. This rotation places the X axis in the position denoted by Xp which is the Intersection of X-¥ plane and the 2-Z plane. Also, this rotation places the y axis in its final position at the angle B with the Y axis. The second transformation consists of a rotation through an angle y about y axis. ‘This rotation places the x and z axes in their final position as shown in Fig. 10.11, Consider the first rotation about Z axis through the angle fi The 3 x 3 rotation matrix T, for this transformation consists of the direction cosines of the f axes, (ie. the axes Xp, Y; Z with respect to the structure axes X,Y,Z). Thus we get, ‘Analysis o Skeletal space frames wr Fig. 10.11 Rotation of axes for a space frame member cosB sin B 0 =sinB cosB 0 (10.23) ooo 4 The functions cos and sing may be expressed in terms of the direction cosines of the member by referring to the geometry of Fig.10.11 as, cos sin We +m) where != DELX/L and m= DELY/L. DELX and DELY are the projections of the member length in Xand Y directions respectively. Thus the rotation matrix T, can be rewritten as: mg YC +m) Yu +m?y = 1 Tele ar Very Le amy remy (10.24)Space Sutures: Principles and Practice Fig. 10.12 Rotation of space frame member about X axis Similarly, the second rotation of the axis about the angle ymay be written as: cosy 0 siny m=| 0 1 0 {10.25) =siny 0 cosy, Here again, the functions cosy and siny may be expressed in terms. of the direction cosines of the member as follows (see Fig. 10.11) cos y= VI? +m?) and sin y=n ‘Thus T, may be rewritten as: (u? +m?) 0 n o 1 _o (10.26) 0 Vu? +m?) ‘The third transformation consists of a rotation through the angle a about the x axis, causing the y and 2 axes to coincide with the principal axis of the ctoss-section. This last rotation is indicated in Fig. 10.12. This rotation of axes through an angle o requires the introduction of a rotation matrix T, in which the elements are the direction cosines of the final axes (r, 3, 2) with respect to yaxes. Thus, 1 0 0 0 cosa -sina (10.27) 0 sina cosa Ty Premultiplication of 7, T, by T, yields the rotation matrix T,, for the three successive rotations shown in Figs. 10.11 and 10.12 as analysis of Skeletal space fames T= 0,0, (10.28) The rotation matrix T can be shown to take the form T 0 0 7, (10.29) ‘Thus the equations relating the components of a load vector in member and global co-ordinates can be written in the usual manner as f”= Tf, and since Tis orthogonal it follows that f= T'f’. The vector 4 transforms in a similar manner, so that we have d’ = Td, d= Ta’. Thus we obtain the usual expression for the member stiffness matrices 1,2) (10.30) y= TK, Tl, The rotation matrix for a vertical member should be treated separately (since I and m are zero for these members). For these members there Js no rotation through the angle i (about the Z axis). Instead, the first rotation is through the angle x which may be either 90° or 270°. The second rotation matrix consists ofthe direction cosines of the 3, 9 and z axes with respect to the structure axes. Thus, 0 ow nsina cosa 0 (10.31) =neose sina 0, Tver Substitution of the rotation matrices (10.24) and (10.26) in the equation (10.30) gives the coefficients of the stiffness matrix K,. The clements of K,, matrix thus obtained are shown in equation (10.32) on page 531 ‘where es 4B Caa= Cat Cont = & C5Cy= (Cy GL F,= msin @+Incos a Fy= —mcos a+ In sin @ oa Be Fy= cosa Fy= sina S29530 Space Structures: Principles and Practice sre Structures: Prine and Practice, Fs= Usin a mncos & Fe= Leos a+ mnsin a PX= P+ m? = ad A= B=? = H=6 As shown earlier, the elements of the stiffness matrices K,y, Ky, and K,, are numerically same as the elements of K,, but with different signs. I the computer program given in Appendix I, the elements of K, alone are stored and withthe proper change of sign the elements of K, K,_ and K,, are formed. Then, by using these matrices the structure stiffness matrix is generated, 10.20 Stiffness matrices which include secondary effects In section 10.9 we have assumed that the deflections of the structure ‘are small compared to the overall dimensions and we derived the ‘equations of equilibrium for the undeformed structure. But in a refined analysis, especially when the deflections are large, the equilibrium in the deformed state of the structure has to be considered. In the deformed geometry, the deflections will not be proportional to the applied loads. The additional effects to be considered in the deformed state of equilibrium are: 1. The change in length of the member 2. The values of a, b and g, which are functions of the axial force, and are no longer equal to 4, 2 and 6 respectively (as given in equation 10.32), and 3. The difference in moment caused by the axial force, The values of a, f and 7 considering the axial forces also into account are derived in Appendix Il. It can be shown that these expressions will reduce to 4, 2 and 6, if the axial forces are zero. To take into account the effect of axial forces, an iterative Procedure is adopted. For the first iteration, the values of oy B and Analysis of Skeletal space ames selS ke whe al wl@xl— le els x Slo ST ES IS Slo & Seles Ly HFS Sle Sle slo s) SUS © SIE +h SSS SS SES SIS Fz SST ]s ws F a Sache ES Sas fis SSS leel Sle & elt (slg Sls § Sele els ele ely 2 Sele Se Se et ey SSP slo slS ole g CSHLENS METS gle « 2 lS Sp shg E Eels se Slo S FAs asskt ¥ SFEIoohk F SIN SS Fle F = a he Pag “? So 2 le x Sele gle gl Pelt Sie SclEch Gj STS SIS She eels x exile & Per 4 e|* ey EcE G g TS FF z te = Sk = gé 5 ¥ ag oFsa Space Structures: Priciles and Practice ‘yare taken as 4, 2 and 6 respectively. As a result of the linear analysis, the values of axial forces are obtained. With these values of the axial forces, the modified values of a, 6 and ¥ are found out using the expressions given in Appendix II. Using these values the analysis is repeated again to get new values of deformations and axial forces. This iteration is continued until two adjacent values of axial forces agree within specified limits, To take into account the effect of moment due to axial force zl the coefficients C, and C,, are modified as 27,—> oy Ein \ PB 7, it~ 2 respectively ‘The effect of change in length is not very important and hence is not usually considered. However its effect can also be incorporated caslly in the computer program shown in Appendix I 10.41 Stiffess matrices which take into account the offset members In practice, the purlins are placed above the main members and hence the axes of these members will be at a distance from the specified joint position. Also, by adoating a standard type of prefabricated connector or a gusset plate, the true end of the member will be offset in all the three directions. To include the effects of such cases, the stiffness matrix should be modified as shown below®. To include the above effects, the following two displacement transformations are necessary. Y= vx (10.33) Z=UY (10.34) T=uv (10.35) The first transformation from X to ¥ involves the rotation of axes from the frame reference axes to the member axes, or in other words, the transformation matrix V is the same as that given in Equation (10.28) The possible misalignment of member ends from specified joint positions are given in Fig. 10.13. The transformation matrix U to take into account these effects is 533 etal space ‘centroidal axis 10.13 Possible misalignment of member ends from specified joint positions Oe =a ate 0 Pew vu: Ge ~Pa 0 (10.36) Sooooe eooooHe oooHoo ° ° ‘The element stiffness matrix k for the offset member is of the form (10.37)Space Structuts: Principles and Practice where A is the area of cross-section, F is the Young's modulus, I, and I, are the moment of inertias about the principal axes and 1, & the ‘member product of inertia about the principal axes. Aya hede= lore (00.38) r (00.39) GI Ep. 2 tp (G1 ~ 2a + 21.) (10.40) By using the equations (10.30) and (10.33) to (10.37) the global stiffness matrix can be obtained. 10.42 Determination of the elastic critical loads of space structure When the extemal loads that produce a set of axial forces (tension and compression) in the members of a structure remain relatively small, any nodal displacement arbitrarily applied on the structure by some disturbing nodal force will vanish when the source of disturbance is, removed. If the loads are increased beyond a certain limit the structure will collapse suddenly under any small disturbing force. The loads causing such a collapse are called the critical or collapse loads. ‘Naturally, the designer of any new type of structure will be interested in knowing the value of the load factor, which would cause the structure to collapse. For all practical purposes the detailed information about the behaviour of the structure at or near the critical load is not important. This is fortunate since a rigorous analysis of a structure under these conditions must inevitably include a consideration of the effects of gross deformation and partial plasticity. Such an analysis is complicated even ‘when cartied out on a computer. In practice, a reasonable estimate of the collagse load of a rigid jointed space frame can be obtained from two quantities which are most easily calculated—the rigid plastic collapse load and elastic load. The calculation of the first of these is a tedious problem especially for space frames and is a linear programming problem. The second may conveniently be obtained by using the modified equilibrium equations described in section 10.10. ‘As discussed earlier the relationship between nodal forces and nodal displacements can be expressed as F=KD (10.41) When the loads reach the critical value, the displacements can assume analysis of Skeletal space frames any value without any change in the nodal forces. In other words, a small change of the displacements 6D can be produced without any change in the nodal forces, that is OF = K,6D=0 (10.42) ‘where Kyis the stiffness matrix of the structure at critical load. Since this, equation forms a set of homogeneous linear simultaneous equations, it leads to the stability criterion iK,|=0 (10.43) It should be kept in mind that the stiffness K, is obtained from the ‘matrix transformation of Equation 10.30, in which the assembled stiffness matrix K consists of the individual member matrices k with consideration of axial forces. 10.12.2 Method of analysis For calculating the buckling load of a structure, the linear analysis as explained in section 10.9 is first performed. In the first iteration the axial load in the members are assumed to be zero for the calculation of a B and y. At the end of the first cycle, the forces in the members are known and then the iteration is repeated till two adjacent values are reasonably close. Actually three cycles are enough and may give ‘good agreement. The coefficients of the equation of equilibrium are determined after determining the forces in the elements. These coefficients are intrinsic functions of the axial force and hence result in @ nonlinear eigen value problem. As explained in 10.12.1, the buckling load is that value of axial force which makes the determinant to vanish, ‘As this load cannot be determined directiy, the following procedure is adopted. First an arbitrary load is assumed and the corresponding determinant is found out. Then the loads are incremented by a factor A and the process is repeated until the determinant changes sign. The particular load at which the determinant changes sign is taken as the buckling load. It is a question of how large the increments should be in order to reach the buckling load at a reasonable speed. At first very large increments are chosen to traverse as wide a range as possible of the external loads. Once the order of buckling load is approximately determined, then a smaller increment is used covering @ narrower range of external loads. The required accuracy can be achieved by suitably adjusting the increments of 4. 40.12.3 Buckling in the inelastic rango ‘The stability of space frames in the inelastic range is highly complicated. Hence many approximate methods have been tried. In one such method,Space Structure: Pinciles and Practce {or the region beyond the onset of yielding, the value of Young's modulus E is replaced by the tangent modulus of the stress-strain curve and the iterative procedure as discussed in 10.12.2 is continued. The computed buckling loads based on the above method have been found to give fairly accurate values. ‘The computer program developed by the author uses this approximate ‘method and is explained in Appendix IL. This program can be used to analyze any skeletal space frame having any geometric configuration, and support condition for buckling in the elastic or inelastic range. In addition to the above methods of analysis, procedures have also been developed for the analysis of space frames for nonlinear geometry and nonlinear material properties. Methods for taking into account semi-rigid joints, nonprismatic members, discontinuities in members, etc, are also available!?>""8, The discussions on these methods are outside the scope of this book and an interested reader is requested to consult the books and papers given at the end of this chapter for further details of these analyses. The methods presented in this chapter provide adequate knowledge for analyzing any type of skeletal space frame encountered in practice. Nonlinear analysis, analysis to take into account semi-rigid Joints, etc., are required only for some special cases. 10.43 Software packages Space frame analysis techniques offer the advantage of generality of formulations. Even in the late 1980s it was estimated that there were about 500 user-oriented and several thousand research-oriented finite element program systems which could be used for the analysis of space frames. A number of software reviews and bibliographies have been published. As it is not possible here to review the capabilities and compare different commercially available finite element analysis packages only the names of some of the popular packages are given below : ABACUS, ADINA, ANSYS, ASKA, COSMOS, GT-STRUDL”, NISA, PAFEC, SAP”, SESAM-80, STRESS!5, MSC/NASTRAN, NIKE3D, PATRAN, STARDYNE, DYNA3D. The above list is not exhaustive. Many of the above packages have been developed for use in Mainframe computers. These programs are quite general in terms of loading, geometric configurations and support conditions. However, for large lattice systems the core storage was often exceeded. Hence substructure and/or condensation methods have als3 been developed to solve large space frames. It is also possible to grou the joints in some convenient way following the topological and geometrical relations of members and joints”, Recently a number of packages have been developed for use with IBM PC or compatible computers. Notable among them are SAP 90, STAAD Il, STAAD Pro, DAST, STRUCAD* 3D, GT-STRUDL, SODA, ROBOS and ‘alysis of Skeletal space frames ‘STRUDS. These packages were originally developed to work on MS-DOS. {version 6.0 or above) operating system and run on IBM-PC - Pentium or compatible machines having the following configurations : 4 MB RAM, 1 floppy disk drive, 4.3 GB hard disk drive, a mouse, a colour VGA monitor and a good quality printer (132 column, 300 cps). Most of these general ~ and special purpose packages can now be executed under MS Windows and/or Unix/X-Windows environment and include graphics pre and post processing capabilities. A large number of public. domain software are also available and are distributed at a nominal cost by agencies such as NISEE, COSMIC, USAWES, NTIS and LLNL. However, the advantages of low cost and availability of source code of public domain software are often offset by the disadvantages such as lack of maintenance and user support. source codes of simple second order and elastic plastic analyses of plane rigid frames (which can be extended to 3D frames) are also available in Ref. 68, 132, and 139-142. ‘With these programs, it is now possible to analyze any complicated geometry subjected to any pattern of loading (static, dynamic) and having any boundary condition or discontinuity. These programs are based on the direct stiffness method of solution and consist of the following steps: ‘Structural discretization Reading in the geometry and member properties of the structure Element stiffness computations Determination of joint loads Coordinate transformation Assembly of global stiffness matrix Application of boundary conditions Solution of simultaneous equations to find joint displacement Computation and display of element forces, stresses and displacements Some of the major features of the commercial software packages are listed below. a) Element Library: The element library is a very important component in a package and allows the user to idealize the system and arrive at the finite element model as realistically as possible. The element library may include structural elements such aS 3D rods (truss), beams, plane stress/strain and membranes in space, axi-symmetric solids, shear panels, plates, thin and thick shells, shells of revolution, 3D solids, discrete stiffeners for plate and shells and pipe elements. The ‘user should have options to use lower or higher order elementsSpace Structures: Principles and Practice 3) in all the above types. Other elements may include matrix elements, substructures and boundary elements. Other effects which ean be modeled include gaps, friction, viscous dampers, cables, plastic hinges and crack tip. Analysis Capabilities and Range of Applications: Linear static analysis = Non-linear static analysis-non-linear material, geometry, large deformations, stress stiffening, plasticity, creep hyper-elasticity and rubber like material behaviour temperature dependent inelastic properties, non-linear contact problems. ~- Dynamic Analysis - Eigen value - free vibrations, linear transient dynamics, spectrum analysis due to seismic loading, - Non-linear dynamic analysis - material and geometric non- linear effects, direct integration and different incremental approaches. - Harmonic response - steady state response of a linear structure subjected to harmonically time - varying loads, = Stability analysis - calculation of critical loads, bifurcation points, buckled shapes, large deflection analysis to determine the limit load for failure by bifurcation or snap through buckling. - Heat transfer - temperature distribution and heat flow within a body, linear and non-linear analysis for temperature dependent material properties and transient heat transfer analysis for time dependent temperature distribution. - Coupled field analysis - simultaneous solution of interaction of multiple field effect such as structural displacements, temperature and heat flows, solid and fluid flow interactions. ‘Types of loading: Option for analysis of the system due to several types of loading such as concentrated loads, line loads, axisymmetric loads, gravity loads, surface and volume loads, inital stresses, strains or velocities, thermal loading, centrifugal loading and contact loading, Boundary conditions and constraints: Sliding interfaces, prescribed displacements, support at contact points, elastic foundation, multi point constraints. ‘Material properties and Models: Material properties may be ‘temperature ~ dependent, isotropic, orthotropic or anisotropic and multi-layered composites. Non-linear material behaviour such as plasticity, creep, elastic-strain hardening, visco-elastic or plastic. ‘analysis uf Shletal space frames 4) Pre and Post-processing: The pre-processor takes minimum input from the user, creates the configuration of the space structure (finite element mesh) and other data required for analysis and displays the model for data check and correction, if any, to be made by the user in an interactive mode. They also provide automatic generation of finite element meshes for two and three-dimensional structures. Graphics oriented edit facility enables the user to edit and display the model and to input/generate additional data like loads, boundary conditions, ete. On-line assistance by way of help screens, windowing/zooming, model rotation, change of view direction, element shrinking, removing hhidden lines/surfaces, etc. are also available. (The StruCAD* 3D package allows the user to define and verify geometry and loads in a user-friendly 3D CAD style graphics environment with pop-up and pull-down menus). Some of these programs also have the following facilities: automatic member offset calculations, members with variable cross-sections, automatic surface ‘area load generation, automatic gravity and buoyancy load generation, automatic mass, added mass and entrapped mass generations. Automatic wind, wave and current load generations on structural and non-structural elements, material and surface area take off, automatic joint and member numbering, Tension only ‘members, compression only supports, Active/ inactive member specifications for _——_load-dependent structures. Rectangular/cylindrical coordinate system with mix and match capabilities, automatic moving load generation, seismic load generation as per the codes of practices, factored load ‘combinations. Segments of repetitive geometry may be used to generate complete structure model. Generate, Copy, Repeat and Mirror commands are available for quick and easy ‘geometry/generation. STAAD III also has facilities to read Auto CAD's DXF format drawings and generate corresponding STAAD Ill input. It also contains library of commonly used structures. CAD facilities include Zoom, Pan, Rotate, Section, Erase, etc The post-processor helps the user to judge the accuracy of results by means of various graphical outputs, to identify the critical values and regions for design and to decide upon the adequacy of modeling/design and the need for remodeling/redesigning the components/structure/solid. The displays of these packages also use colour codes to show the contours of stressed zones. They also allow display and plotting of deflection, shear force, bending moment, EigenSpace Strvtuts: Principles ans Practice vectors and code check diagrams of all/ selected joints, supports, members. If necessary, numerical values can also be seen in tabular form. User controlled scaling of deflecton and mode shapes are available. Powerful on-line query for analysis/design results are also available. They also provide for animation of deflection/mode shapes, external printer plots and. multiple windowing. Design optimization: Some of the packages now offer sophisticated family of computer programs for optimum structural design - the capabilities include minimization of material volume, mass and weight for fixed geometry by changing thickness and cross-sectional dimensions and optimization of structural shapes. Most of these packages have capacity to analyze frames having. up to 6,000 joints, 12,000 members, 36,000 degrees of freedom, 200 Basic and 200 combined load cases. They have builtin algorithms which utilize the hardware optimally for computation and storage. They also perform global static equilibrium check and joint equilibrium reports. Most of the programs also have steel/concrete] timber design capacities. STAAD IT package has built in Steel tables including U.S. British, European, Canadian, Japanese, Chinese, Indian, Australian and Korean steel tables. Code check/member selection can be done as per the codes of the above countries. These programs also do the weld design. 10.43.41 Specialized space frame packages ‘The companies which specialize in space frame construction, have developed their own special purpose space frame analysis and design programs. For example, KONDAR is a program system developed at the ‘Technical University of Darmstadt for the computer supported geometric construction and representation of spatial objects. This program has been used by the space structure company MERO for studying a variety of topological networks and geometric shapes that are encountered in the ‘computer based practical design of space structures. The details of this program are available in Ref. 92. ‘Another program called SECAD has been developed by Chu of “Xinhong Space frame Co. Ltd, China. This program runs on a Sun SPARC station in the windows environment with the graphic interface of ‘menus, buttons and icons making it convenient to use. Formex algebra (See chapter 11) concepts have been used in this program for the configuration data processing, Thus most types of space frames can be easily generated, graphically modified and automatically analyzed. The fabrication drawings of the bolted spherical jointed space frames can be plotted which can be sent directly to the workshop for fabrication. Thus the total design time is greatly reduced. This program has been ‘used in the design of more than 600 space frame projects between 1989 analysis of Skeletal space frames ‘and 1993. The details of the program and the various space structures designed by using it are given in Ref. 93. Sedlak of the University of New South Wales describe a conceptual Structural Design Aid (SDA) package, which may be used as a tool during the conceptual design stage to establish possible solutions interactively and make valid decisions about structural system choice and its impact on building design. Fig. 10.14 shows the typical conceptual structural design process and the components of SDA. This tool consists of a case study (LASE) database which contains details of 450 full projects and 1300 partially entered projects. Thus it provides a case-study database resource for rapid information access. It also provides a catalogue of possibilities related to shape, structure, use, building envelope and material as a design resource. An intuitive graphical tool-kit for the examination of structural adequacy of a proposed building structure during the design process has also been provided. It also provides guidance in choosing appropriate structural systems through statistical associations. The background and capabilities of the fully functional prototype of this tool is given in Ref. 117. It is also available in Internet!#. +10.13.2 Expert systems/knowledge based systems ‘To produce efficient design, the design engineer is faced with not only problems of engineering nature, but also of information handling, Tat is, much information about the previous designs of similar nature, the failures or deficiencies, etc., are unknown to him. However, with the use of expert systems such problems can be solved. Expert or knowledge-based systems are intelligent behavioural computer programs that solve problems just like an expert would. [smiewre}-{ pn] [coans}-{" pe] (Ct samo a} Develop mesa ‘SpateComey > Fig. 10.14 SDA and the typical conceptual structural design process”Space Structures: Principles and Practice Knowledge-based systems or interactive computer programs incorporate rules of thumb, heuristic knowledge, intuition and other expertise to provide knowledgeable advice on a variety of tasks. The heuristics are usually accumulated by a human expert over a number of years. Using heuristics, an expert system can make educated guesses, recognize promising approaches and avoid blind search; and consequently it can narrow down the search process in a solution space, Expert systems have three basic components : a knowledge base, an inference engine and an user-interface (see Fig. 10.15). To facilitate the development of knowledge based expert systems, expert system programming environments or shells have been developed. They contain specific representation methods and inference mechanisms. Few engincering problems can be solved by a purely heuristic approach. ‘Numerical algorithmic routines must usually be combined with heuristics. Thus an expert system shell suitable for engineering applications should ideally be able to handle scientific numerical computations within the systems, Several expert system shells have been developed to aid in acquiring knowledge, representing it in the knowledge base and verifying its correct usage by the system. ‘Some of the shells available on PCs are : EXPERT - EASE, M.1, Rule Master, INSIGHT 2+, 1st - class and VPEXPERT. INSIGHT 2+ shell is, divided into four integrated components - an inference engine, a text editor, a compiler and an extensive help system. It uses a production, rule language in which the rules are written. Rules are written in IF/THEN form with a wide range of AND/ORJELSE options. These features make the creation of complex rules possible. Knowledge in the expert systems is incorporated either as facts or rules. Facts may be in several form: simple facts, mumeric fact, string facts or attribute facts. The INSIGHT 2+ inference engine uses a backward-chaining algorithm, although a Kaowiedge tase Working Lynne) "tava le Keowee Leer] esi es) senition Tate necanion |} ——~--—+{ arg ¥ ‘ince a he crpmereeyyf Fig. 10.15 Components of an expert system ‘Anaya of Skeletal space frames Fig. 10.16 Elements of the expert system package developed at SERC'?° forward-chaining algorithm is also possible. INSIGHT 2+ is written in "TURBO PASCAL. ‘An expert system for discrete optimization of space trusses has been. developed by Niczyj and Paczrowski””. The Structural Engineering Research Centre (SERC) of the Council of Scientific and Industrial Research, Chennai, India has developed two knowledge based expert systems for the analysis and design of free standing microwave and transmission line towers. With these one can design steel towers supporting parabolic or shielded antenna and single or double circuit transmission line towers of various capacities. Blackboard architecture which organizes various modules has been used to develop these systems. {t forms a dynamic global shared database. In this expert system packages, the height of the ‘ower, its location and details about the antennae which the tower is to support are fed as input. This information is used by the processor to generate a suitable configuration. The system then calculates the loads due to wind acting fon each member of the tower. These are used by various modules which analyze the structure and if necessary modifies the design for these forces. The configuration processor module decides on the geometric configuration of the tower. The designer module, evaluates design based on criteria such as weight of the tower and tip deflection. One of the important features of these systems is the ability to improve their performance. Their knowledge is represented in the form of IF-THEN rules, for example: If the height of the tower is ‘H’ metres THEN the most economical base width lies in the range (H/6, H/4). Associated with each rule is a performance measure. This numberStruetues: Principles and Practice is an indication of the worthiness and is updated after every iteration (design). The elements of these expert systems are shown in Fig. 10.16. Ramaswamy describes an expert system called CCCXPERT, developed using the C language, developed for the design of double layer grids using circular and rectangular hollow sections. This system is used in conjunction with SAP 90 analysis package. The details of this expert system is available in Ref. 131. References 133 to 160 provide more information about the recent developments in the analysis of space frames. REFERENCES: 1, Almroth, 3.0. and Brogan, F.A., Automatic Choice of Procedures in Computerized Structural Analysis, Computers and Structures, Vol. 7, 1977, pp. 335-342, 2. Argyris, JH, and Kelsy, S., Energy Theorems and Structural Analysis, Buttervortts, London, 1960. 3. Bazant, Z.P, and El Nimeiri, M., Large-Deftection Spatial Buckling of Thin Walled’ Beams and Frames, Journal of the Engincering Mechanics Div., ASCE, Vol. 99, No. EM6, Dec. 1973, pp. 1259-1261. 4. Benedetti, D.. and Ionita, V, Nonlinear Analysis of Space Trusses, Mecanica, Journal of the Italian Association of Theoretical and Applied ‘Mecahnics, Vol. 9, No. 4, Dec. 1974, pp. 274-282. Bibliography on Latticed Structures, by the Subcommittee on Latticed Structures of the Task Committee on Special Structures of the Committee on Metals of the Structural Div, Journal of the Structural Div, ASCE, Vol. 98, No. ST7, July 1972, pp. 1545-1566. 6. Bray, KHM, Croxton, PCL, and Martin, LH., Matix Analysis of Structures, Edward Arnold, 1976. 7. Brooks, DF, and Brotton, D.M., Formulation of a General space frame Program for a Large Computer, The Structural Engineer, Vol. 44, No. Ii, Nov. 1966, pp. 381-396, 8. Bryan, E.R, The Stressed Skin Design of Steel Buildings, Crosby Lockwood Staples, London, 1973. 9. Chu, KH. and Rampetsreter, Large Deflection Buckling of space frames, Journal of the Structural Div., ASCE, Vol. 98, No. ST12, Dec. 1972, pp. 2701-2712. 10, Davies, RM. ed.), Space Structures, Blackwell Scientific Publications, Oxford, 1967. 1, Dean, DLL, and Avent R.R, State of the Art of Discrete Field Analysis of ‘Space Structures, Proceedings of the 24 International Conference on Space Structures, University of Surrey, Sept. 1975, pp. 7-16. 12, Dean, D.L, and Ganga Rao, H.V'S., Macro Approach to Discrete Field Analysis, Journal of the Engineering Mechanics Diu., ASCE, Vol. 96, No. EM4, August 1970, pp. 377-394. 13. Dickie, JF, and Dunn, D.J., Yield Pattern Considerations in Space Structures, The Structural Engineer, Vol. 53, No. 3, Mar. 1975, pp. 147-152. 14. Enami, A, and Sunami, 9. Prismatic’ Folded Plate Roof’ Structure consisting of Steel Pipe panels for a Factory Building (Erection, Loading ‘analysis of Skeletal space Frames 15, 16. 17. 18. 19. 20. 21 22, 23, 24, 25. 26. 27 28, 29, 20, 31 32, Test and Calculations), Proceedings of the IASS Congress on Large Span Shells, U.SSR., Moscow, 1968, pp. 175-189. Fenves, S.J», Logger, R.D., Mauch, S.P., and Reinschmidt, K.R, STRESS: A User's Manual, MLLT. Press, Cambridge, Mass. 1964. Flower, W.R., and Schmidt, LC, Analysis of Space Truss as Equivalent Plate, Journal of the Structural Div,, ASCE, Vol. 97, No. ST12, Dec. 1971, pp. 2777-2789, Forman, S.E., and Hutchinson, T: i ‘Structures, Intemational Journal of Solids and Structures, Vol. 6, 1870, pp. 909-932. Ganapathy Chettiar, C., and Subramanian, N., A Computer Program for the Buckling Aralysis of Space frames, Computer Society of India, Annual Convention, Pure, Jan. 1977. Ganapathy Chettiar, C., and Subramanian, N., Analysis of Steel Frame Folded Plate Roofs, Bulletin of the International Association for Shell and ‘Spatial Structures, Vol. XVII~3, No. 62, Dec. 1976, pp. 33-41, Gere, J-A., and Weaver, W., Analysis of Framed Structures, Van Nostrand Company Ltd, 1369, Gesund, H., Computer Analysis of Reinforced Concrete space frames, Conference ‘on Inelasticity and Nonlinearity in Structural Concrete, University of Waterloo, 1972, pp. 289-314, Harung, H.S., Lightfoot, E., and Duggan, D.M., The Strength of Scaffold ‘Towers Under Vertical Loading, The Structural Engineer, Vol. 53, No. 1, Jan, 1975, pp. 23-30. Heyman, J, The Limit Design of Transversely Loaded Square Grid, Journal of Applied Mechanics, ASME, Vol. 74, 1952, pp. 153-158. Hendry, A.W., and Jaeger, Z.G., The Analysis of Grid Frameworks and Related Structures, Chatto and Windus, 1958, Hodge, G-P, Plistic Analysis of Structures, McGraw Hill Book Co., 1959, P. 75-80. Hussey, MJ.L, Tarzi, A.l, and Theron, W.ED., Simply Supported Rectangular Double-layered Grids, Journal ‘of the Structural Div., ASCE, Vol. 97, No. ST3, March 1970. ICES—STRUDL I, Finite Element Program ofthe Integrated Civil Engineering System (ICES), M.LT., Cambridge. Jagannathan, D., Epstein, HL, and Christiano, P, Nonlinear Analysis of Reticulated space frames, Journal of the Structural Div, ASCE, Vol. 101, No. ST12, Dec. 1975, pp. 2641-2658. Keshava Rao, MN., and Raman, N.V., Computer Analysis of Rigid-jointed space frames, Indian Concrete Journal, Oct. 1968, pp. 394-404, Feb. 1967, pp. 71-78, Krishna, P., Cable Suspended Roofs, McGraw Hill Book Co., New York, 1978. Latticed Structures: State of the Art Report, Journal of the Structural Div., ASCE, Vol. 102, No. ST 11, Nov. 1976, pp. 2197-2230. Lightfoot, E., and Sawko, F, The Analysis of Grid Frameworks and Floor Systems by the Electronic Computer, The Structural Engineer, March 1960, pp. 78-87.Space Structures: Prinples and Practice 33. Livesley, R.K., The Application of an Electronic Digital Computer to Some Problems of Structural Analysis, The Structural Engineer, Jan. 1956, pp. 1-12, 34, Livesiey, R.K., Matrix Methods of Structural Anclysis, 2" ed., Pergamon Press, 1975, 35, Majid, K.1, Nonlinear Structures, Butterworths, Landon, 1973, 36. Makowski, Z.S., and Pippard, A.1.S., Experimental Analysis of Space ‘Structures with Particular Reference to Braced Domes, Proceedings of the Institution of Civil Engineers, Part II, Vol. 5, Dec. 1956. pp. 824-844. 37, Mallet, R.H., and Berke, L, Automated Method for the Large Deflection and Instability Analysis” for Three-Dimensional Truss and Frame Assemblies, AFFDL-TR-66-102, Wright-Patterson Air Force Base, Ohio, May 1968. 38. Mason, J., Contribuicao a Teoria das Estruturas em Superficie Anisotropa (in Portuguese), Rio De Janerio, 1963, 39. Mason, J, Nonlinear Buckling of Lattice Domes, Publications of the IASS, Vol. 32-11, 1972, pp. 107-119. 40, Mcteish, A., and Tatsa, E.Z., Overall Buckling of Reticulated Shells, JAS World Congress on Space Enclosures, Montreal, 1976. 41, Morino, S., Analysis of space frames, Ph.D. Dissertation, Lehigh University, 1970. University Microfilms, Ann Arbur, Michigan. 42. Naslund, K.C., Model Tests Predict space frame Behaviour, Architectural Record, Dec. 1961, p. 152. 43. Noor, A.K., Thermal Stress Analysis of Double-layer Grids, Journal of the Structural Div., ASCE, Vol. 104, No. ST2, Feb. 1978, pp. 251-263. 44, Olivares, et al, Experimental Analysis of a Grillage Girder Bridge, IABSE Publications, Vol. 21, 1961, p. 237. 45, Pagno, M., Theoretical and Experimental Research on Triangulated Steel Vaults, Hanging Roofs, Proceedings of IASS Colloquium, Paris, 1962, pp. 265-275, 46. Pederson, P,, Optimal Joint Positions for Space Trusses, Journal of the Seructural Div., ASCE, Vol. 99, No. ST12, Dec. 1973, pp. 2459-2476. 47. Ray, K.C,, Analysis of Grid Floors, The Indian Cencrete Journal, Vol. 34, No. 6, June 1960, pp. 227-232. 48. Renton, J.D., Stability of space frames by Computer Analysis, Journal of the Structural Dis., ASCE, Vol. 88, No. ST4, Aug. 1962. 49. Renton, J.D., A Finite Difference Analysis of the Flexural-Torsional Behaviour of Grillages, International Journal of Mechanical Sciences, Vol. 6, 1964, pp. 209-224. 50. Sadeghi, M.K., Berechung der Riumlichen Stabwerke unter Bericksichtigung der Wolbkrafttorsion nach der Deformations-methode, Bauingenieur, Vol. 52. Nov. 1977, pp. 413-420. 51. Salvadori M., and’ Levy, M., Structural Design in Architecture, Prentice Hall Inc., Englewood Cliffs, NJ., 1967. 52. Schmid, L.C, Altemative Design Methods for Parallel-Chord Space Trusses, The Structural Engineer, Vol. 50, No. 8, 1972, p. 245. 53, Schmidt, L.C., Morgon, PR., and Clarkson, J.A., Space Truss Design in Inelastic Range, Journal of the Structural Div. ASCE., Vol. 104, No. ST12, ec. 1978, pp. 1915-1919, ‘alysis of Skeletal space Frames 54, 55. 56. sn. 58, 59, 60. 6. 62, 63, 64. 65, 66. 67. 68. 69, ‘Shah, ML, and Gesund, H., The Analysis of Nonlinear Three-Dimensional Frames, Computers and Structures, Vol. 2, 1972, pp. 943-954. Soare, M.V., Contributions to the Analysis of Double-layer Mesh Grids, Proceedings of the 2" International Conference on Space Structures, University of Surrey, Sept. 1975, pp. 159-168. Stricklin, J.A., Heisler, WE, and Von Riesemann, W.A., Geometrically Nonlinear Structural Analysis by Direct Stiffness Method, Journal of the Structural Div,, ASCE, Vol. 97, No. STS, Sept,, 1971, pp. 2299-2314, Subramanian, 'N., Analysis of Space Structures with Computer as a Problem Solver, Newsletter of the Computer Society of India, Vol. 12, No. 3, Sept. 1976. pp. 7-8. Subramanian, N, Analysis and Behaviour of Steel Frame Folded Plate Roofs, Thesis Submitted for the award of the Degree of Doctor of Philosophy, Indian Institute of Technology, Madras, 1977. ‘Subramanian, N., and Ganapathy Chettiar, C., Experimental Investigations fon Steel Frame Folded Plate Roofs, Bulletin of the International Association for Shell and Spatial Structures, Vol. XIX-2, No. 67, ‘Aug. 1978, pp. 45-53. Subramanian, N. and Ganapathy Chettiar, C, Computer Analysis of space frames with Offset Members, Computers and Structures, Vol. 11, April 1980, pp. 287-303. ‘Supple, Wil, ed., Proceedings of the Second International Conference on ‘Space Structures, University of Surrey, England, Sept. 1975 Tezcan, S.S., Computer Analysis of Piane and space frames, Journal of the ‘Structural Div., ASCE, Vol. 92, No. ST2, April 1966, pp. 143-173, Tezcan, S.5., Large Deflection of Structures by Stiffness Method, Proceedings, 4° IKM International Colloquium on Applications of Mathematics in Engineering, Institute fir Mathematik, Weimar, East Germany, June 1967. Tezcon, SS., and Mahapatra, B.C, Tangent Stiffness Matrix for space frame Members, Journal of the Structural Div,, ASCE, Vol. 95, No. ST6, June 1969, pp. 1257-1270. Timoshenko, S.P, and Gere, J.M., Theory of Elastic Stability, MeGraw Hill Book Co., New York, 1961. Vinnakota, S., and Aoshima, Y., Spatial Behaviour of Rotationally and Directionally Restrained Beam Columns, Publications of the IASS, Vol. 34-11, 1974, pp. 169-194. Waiker, H.B., The Development of the Nodus Space Structural System, Lecture delivered during the Official Opening of the Space Structures Research Laboratory, University of Surrey, 12" April 1972, p. 8 Wang, CX Computer Methods in Advanced Structural Analysis, Intezt Educational Publishers, New York, 1973, Westlake, LR, A Handbook of Numerical Matrix Inversion and Solution of Linear Equations, John Wiley and sons Inc., New York, 1968. Wilson, EL, A General Structural Analysis Program (SAP), Report No. SESM 70-20, Department of Civil Engineering, University of California, Berkeley.n. n. B. m 15. 76. ”. 78. 19. a1 82, 83, 84. as. 6, 87. Space Structures: Pinciles and Practice Wright, D.T,, Membrane Forces and Buckling in Reticulated Shells, Journal of the Structural Dixz, ASCE, Vol. 91, No. ST1, Feb. 1965, pp. 173-201 Also see discussion June, August and October 1965 and April 1966, Wolf, J.P, Post Buckling Strength of Large Space Truss, Jounal of the Structural Dix., ASCE, Vol. 99, No. ST7, July 1973, pp. 1708-1712. Yamada, Y., ct al, Large Deformation and Critical Load Analysis of Framed Structures, Computational Methods in Nonlincar Mechanics, JT. Oden et al, eds., Texas Institute of Computational Mechanics, 1974, pp. 819-828. Gackstatter, G., Beitrag Zur Ermittlung Der Traglast und Zur Bemessung nach der Plastizitits theorie von Zweilagig-2weiliufigen Raumfachwerkrosten aus Stahl mit hilfe der losungen far das Elastisch- Plastische Kontinuum, Doktor-Ingenieur Dissertation, D 83, TU Berlin, Berlin, 1975, p. 123 Demidow, N., and Klimke, H., Optimierung einiger Parameter bei Vorgespannten Raumstabwerken, Der Bauingenieur, Vol. 55, 1980, pp. 137-140. Rother, H., Dickel, T, and Renner, D., Snap through Buckling of Reticulated Space Trusses, Journal of the Structural Div., ASCE, Vol. 107, No. STI, Jan. 1981, pp. 123-143. Coates, R.C, Coutie, M.G., and Kong, Fl 1972. Mahadevevappa, P,, Investigation of the Behaviour of Steel Braced Barrel ‘Vault Roofs, Ph.D Thesis, Indian Institute of Technology, Madras, 1981. Mahadevappa, P,, Subramanian, N., and Ramamurthy, LN., Tragverhalten stahlerner Fachwerktonnen, Der Stahlbau, No. 2, Feb. 1983, pp. 54-58. Subramanian, N., and Ganapathy, C., Prefabricated Steel Frame Folded Plate Roofs, Bulletin of the International Association for Shell and Spatial Structures, Vol. XXI, No. 73, Aug. 1980, pp. 47-53. Scheer, J, and Koep, H., Zur Optimierung von Raumfachwerken, Bauingenieur, Vol. 57, 1982, pp. 27-33. Ramchandra, Trikka, D.N. and Prem Krishna, Nonlinear Analysis of Stee! Space Structures, Journal of Structural Engg., ASCE, Vol. 116, No. ST4, April 90. Ramchandra, Krishna, P., and Trikka, D.N., Elastic Plastic Analysis of Steel Space Structures, Journal of Structural Engg., ASCE, Vol. 116, No. ST4, April 90. Dubina, D., Computation Models and Numerical Solution Procedures for Nonlinear Analysis of Single Layer Lattice Shells, International Journal of Space Structures, Vol. 7, No. 4, 1992, pp. 321-333. ‘Suzuki, T, Ogawa, T., and Ikarashi, K., Elasto Plastic Buckling Analysis ‘of Rigidly Jointed Single Layer Reticulated Domes, International Journal of Space Structures, Vol. 7, No. 4, 1992, pp. 363-368. Proc. Symposium on Preprocessing and Case-studies in Computer Aided Design of Space Structures, July 1990, Osaka, Japan. International Journal of Space Structures, Spl. Issue on Geometric Analysis of Space Structures, Vol. 6, No. 2, 1991 Structural Analysis, Nelson, Analysis of Steel space frames 88 Sumec, J, and Lichardus, 8, Discrete Structures-Statie Analysis-State- of-the-art report, Bulletin of IASS, Vol. 31, No. 104, Sept-Dec. 1980, pp. 153-159. 89, Ong, CE, Wakufield, D.S., and Barner, M.R, Interactive Graphic CAD for Tension Structures, Computers and Structures, Vol. 41, No. 6, 1991, pp. 1305-1312. 90. Anckwe, CM, Reduction Method of Analysis for Dense Flat Gris, Space Structures, Vol. 2, 1986 | 87, pp. 109-334 D1. Salalegheh, E., ond Vanderplaate, GN, An Efficient Approximation Method for Structural Synthesis with Reference to Space Structures, Space Structures, Vol. 2, 1986/87, pp. 165-175. 92. Alvarez, JS., Computer-based Modelling of Space Structures with the Program KONDAR, Proceedings of the 4° Intemational Conference on Space Structures, Univ. of Surrey, Sept. 1993, pp. 1250-1257, 93. Chu, F, SFCAD and its applications, Proceedings of the 4° International Conference on Space Structures Univ. of Suey, Sept. 1953, pp. 1258-1267 94. Sulo, F, A Reduction Method for the Solution of Large Sets of Equilibrium Equations, Proceedings ofthe 4% International Conference on Space Structures, Univ. of Surrey, Sept. 1993, pp. 1275-1284. 95. Majowiecki, M., and Travisan, R., Linked Interactive Graphic Software ‘Addressed to Siatic and Dynamic Analysis of 3D Structural Systems, Proceedings of the ¥ International Conference on Space Structures, Univ, of Surrey, Sept. 1993, pp. 1285-1294. 96. Subramanian, N., and Venugopal, M.S., A Micro Computer Program for Grids, First National Seminar on Computers & Civil Engineering, Chena, India, 26 Nov. 1983, 97. Subramanian, N, and Venugopal, M.S., Computer analysis and design of structures, Instruments & Electronics Developments, Vol. IV, No. 1A, ‘Aug. 1983, pp. 20-25. 98, Subramanian, N., and Venugopal, M.S., Computers and Civil Engineering Design, Indian Construction, Vol. 16, No, 30, Oct. 1983, pp. 39-42. 99, Subramanian, N., and Venugopal, M.S., A Micro Computer Program for Grids, Computerage, Vol. 2, No.5, Feb. 1984, pp. 51-54. 100. Rodrigues, S.J, Node Numbering Optimization in Structural Analysis, ‘Journal of tke Structural Division, ASCE, Feb. 1975, pp. 361-375. 101. Jiankeng, S., Stability of Braced Domes under Dynamic Loading, Proceedings ofthe 4" International Conference on Space Structures, Uni. of Surrey, Sept. 1993, pp. 323-330, 102. Sumec, J, Some Stability Aspects of Reticulated Shells, Proceedings of the 4% International Conference on Space Structures, Univ, of Surrey, Sept. 1993, pp. 339-348. 103, Karamanos, A.S., and Karamanos, S.A., Seismic Design of Double-Layer Space Grids and their Supports, Proceedings of the 4°" International Conference on Space Structure, Univ of Surrey, Sept. 1993, pp. 476-404, 104, Somerville, T, and Krishnamurthy, N., Plane Analysis of Steel space frames by Computer, International Symposium on Structural Stel Design and Construction, July 1985, pp. 548-584,Space Structures: Principles and Practice 105. Meek, J.L., and Loganathan, S., Large Displacement Analysis of space frame Structures, Computer Methods in Applied Mechanics and Engineering, Vol. 72, 1989, pp. 57-75. 106. Meck, JLL., and Tan, H.S,, Geometrically Nonlinear Analysis of space frames by an Incremental Iterative Technique, Computer Methods in Applied Mechanics and Engineering, Vol. 47, 1984, pp. 261-282. 107. Tan, T.H., and Schmidt, LC, Ulimate Load Behaviour of a Barrel Vault Space ‘Truss, Inferrational Journal of Space Structures, Vol. 2, 1986/87, pp. 1-10. 108, Kollar, I~, and Dulacska, F., Buckling of Shells for Engineers, John Wiley. & Sons, Chichester, 1984, 109. Collins, RJ, Bandwidth Reduction by Automatic Renumbering, International Journal for Numerical Methods in Engineering, 1973, Vol. 6, pp. 245-355, 110. Subramanicn, N., Computer Analysis of Plane and space frames using ‘a Micro Computer, First National Seminar on Computers & Civil Engineering, Chennai, India, 26 Nov. 1983, 111. Jonatowski, JJ, and Koutsoubis, D., A Microcomputer Aided Design System for Large Space Structures, Proc. 3° International Conference on Space Structures, Univ. of Surrey, 1984, pp. 301-305. 112, Mullord, P., A Review of Collapse Analysis of Space Structures, Proc. 3° Internationel Conference om Space Structures, Univ. of Surrey, 1984, pp. 647-649, 113. Fathelbab, F.A., Tangent Stiffness Matrix for space frame Members with both member and joint Imperfections, Proceedings of the 4" International Conference on Space Structures, Univ. of Surrey, Sept. 1993, pp. 1323-1333, 114. Wang, M., Chen, Cy and Shem, S., Geometric and Material Nonlinear Analysis of Latticed Shells of Negative Gaussian Curvature, Proceedings (of the 4 International Conference on Space Structures, Univ. of Surrey, ‘Sept. 1993, pp. 640-648. 115. Kawaguchi, K, Hangai, ¥, and Nabana, K., Numerical analysis for flding. of space structures, Proceedings of the 4'* International Conference on ‘Space Structures, Univ. of Surrey, Sept. 1993, pp. 813-823, 116, Avram, C., and Anastasescu, Space structures, Elsevier, Amsterdam, 1984, p. 377. 117, Sedlak, V, A Computer Aided Conceptual Structural Design Aid, Proceedings of the IASS International Symposium '97 on Shells and Spatial Structures, Nov. 1997, Singapore, pp. 745-754. 118. Tsuboi, ¥., Analysis, Design and Realization of space frames: A State-of- the-art Repert, Bulletin of International Association of Shell and Spatial ‘Structures, Wo. 84185, Vol. XXV-1/2. 119, Malla, RB, and Serrette, RL, Double-layer Grids: Review of Static and Thermal Analysis Methods, Journal of Structural Engineering, ASCE, ‘Vol. 122, Ne. 8, Aug. 1996, pp. 873-881. 120. Malla, R.B., and Serrette, R.L., Double-layer Grids: Review of Dynamic ‘Analysis Methods and Special Topics, Journal of Structural Engineering, ASCE, Vol. 122, No. 8, Aug. 1996, pp. 882-892. 121.Hanaor, A. and Ong, A.P, On Structural Redundancy in Space ‘Trusses, International Journal of Space Structures, Vol. 3, No, 4, 1988, pp. 237-241 Analysis of Sheetal space Frames 122, Madi, ULR., and El-Tayem, A., On the Effective Length of Compression Members in Double-layer Grids, International Joumal of Space Structures, Vol. 6, No. 1, 1991, pp. 33-39. 123,Murtha-Smith, E, and Bean, JE, Double-layer Grid space frame Euckling, International Journal of Space Structures, Vol. 4, No. 3, 1989, pp. 117-127. 124. Murtha-Smith, E,, Chaturvedi, A.. and Leary, S., Structural Integrity of Space Trusses, Spatial Lattice and Tension Structures, J.Abel, J.Leonard and C-Penalba, eds., ASCE, New York, 1994, pp. 409-418, 125.Holzer, S., Dynamic Consideration in Latticed Structures, Journal of ‘Structural Engineering, ASCE, Vol. 110, No. 10, Oct. 1984, pp. 2547-2551 126. shikawa, K., and Kato, S., Dynamic Buckling Behaviour of Single and Double-layer Latticed Domes Due to Vertical Earhquake Motions, Space Structures-4, G. Parke and CHoward, eds., 1993, Vol. 1, Thomas Telford, London, pp. 466-475. 127. Wiczyj, J, and Pacckowski, W., An Expert System for Discrete Optimization of Space Trusses, Space Structures 4, G. Parke and C. Howard, eds., 1993, Vol. 1, Thomas Telford, London, pp. 1756-1765. 128, Adel, H., Artificial Intelligence in Structural Engineering, Engineering “Analysis, Vol. 3, No. 3, 1986, pp. 154-160. 129. Sriram, 'D., Knowledge-based Approaches for Structural Design, Computational Mechanics Publication, Southampton, UK., 1987. 130. Murlidharan, TL, Aravind, H.B., Suryakumar, G.V. and Raman, N.V., Expert Tower Analysis and Design System, I: Architecture and Heuristics, Journal of Computing in Civil Engineering, Vol. 5, No. 2, 4pr91, pp. 175-192, Il, Search Strategies and Learning, pp. 193-210. 131, Ramaswamy, G.S., Expert System for space frames, International Course on Space Structures, Anna University, Nov. 97. 132, Harrison, H.B., Computer Methods in Structural Analysis, Prentice-Hall, Inc., Englewood Cliffs, New Jersy, 1973. 133. Chali, A., Neville, A.M., and Brown, T.G., Structural Analysis, A unified Classical and Matrix Approach, Fifth edition, Spon Press, London and New York, 2003, 844 pp. 134. McGuire, 'W., Gallagher, R.#., and Ziemian, B.D., Matrix Structural ‘Analysis, Second edition, John Wiley, New York, 2000, 460 pp. 135. Chen, WF, Goto, ¥., and Liew, JY.R, Stability Design of Semi-Rigid Frames, John Wiley & Sons, Inc., New York, 1996, 468 pp. 136.Faslla, C, Piluso, Vi, and Rizzano, G., Structural Steel Semi-Rigid Connections: Theory, ‘Design and Software, CRC Press, Boca Raton, Florida, 137. Chen, WE, and Tona, S., Advanced Analysis of Steel Frames, Theory, Software and Application, CRC Press, Boca Raton, Florida, 1993, 400 pp. 138. Chen, WF, and Sokal, 1, Plastic Design and Second Order Analysis of Steel Frames, Springer Verlag, 1985, 509 pp. 139. Wang, C.K, Structural Analysis\ on Microcomputers, Macmillan, New York, 1986. 140. Farrison,” H.B., Structural Analysis and Design, Part 1 and part2, Pergamon Press, Elmsfred, NY, 1990. 551Space Stuctures: Principe and Practice 14, Fleming, J.F., Computer Analysis of Structural Systems, McGraw Hill, New York, 1989, 574 pp. 142. Fleming, J.P, Analysis of Structural Systems, Prentice Hall, Inc., Englewood Cliff, N.J, 1997, 548 pp. 143, hetp://emulava.foc.unsw,edu.au:8080 144. Zhao, HL, Huang, WIM., and Zhao, C.Q., A method to calculate the critical loads of single layer shallow lattice domes with initial imperfections. Space Structures 4, Thomas Telford, London, , 1993, pp. 127-135. 145. Suzuki, T, Ogawa T., and Irakashi, K. Elastic buckling analysis of rigidly’ jointed single’ layer reticulated domes with random initial imperfection. International Journal of Space Structures. Vol. 7, No. 4, 1992, pp. 265-273. 146, Murakami, M., Numerical Analysis of Elastic Buckling of Single-Layer Latticed Domed under Gravity Load, Proceedings of International Congress of IASS-CSCE, Canada, 1992. 147, El-Sheikh, A., Numerical Analysis of Space Trusses with Flexible Member-end Joints, International Journal of Space Structures, Vol. 8, No. 3, 1993, pp. 189-197, 148, Saka, T., Taniguchi, ¥, and Tsutsui, T., Elastic buckling behavior of double-layer cylindrical lattice shell roofs composed of space trussed units with internal instability, Shells and Spatial Structures from Recent Past to the Next Millennium, Madrid, 1999, Vol. 2, B2.63-2.70. 149. Saka, T., Taniguchi, ¥., and Konishi, T, Member Design and Elastic Buckling Strength of the Mesh Double-Layer Braced Domes, Theor, Design and Realization of Shell and Spatial Structures, IASS 2001, Nagoya, Japan, TP100. 150. Horr, A. M., and Schmidt, L.C., Dynamic Response of a Damped Large ‘Space Structure: A New Practional-Spectral Approach, Int. Journal of ‘Space Structures, Vol. 10, No. 2, 1996, pp. 134-141, 151. Kavek, A., and lranmanesh, A., Comparative study of backpropagation ‘and improved counter propagation neural nets in structural analysis ‘and optimization, International Journal of Space Structures, Vol. 13, 1998, pp. 177-185. 152, Kaveh, A., and Servati, H., Design of double layer grids using back propagation neural networks, Computers and Structures, Vol. 79, 2001, pp. 1561-1568. 153. Takashima, H., Problems to Establish an Expert System for Predicting. Buckling Behaviors of Steel Single Layered Reticular Domes, Proceedings ‘Of IASS 2001 International Symposium on Theory, Design and Realization of Shell and Spatial Structures, Oct., 2001. 154, Takashima, H., Proposals to create ‘an Expert System to Predict Hlastic- Plastic Behaviors of Single Layered Lattice Domes, Lightweight Structures in Architecture, Engineering and Construction, IASS/IEAust/LSAA International Congress, Sydney, Vol. 1, Oct, 1998, pp. 473~480. 155, Bangash, MV. and Bangash, TL, Elements of spatial structures: ‘Analysis and Design, Thomas Telford, 2003, 669pp. ‘analysis of Skeletal space frames 156. Hori, A., and Sasagaw, Large deformation of inelastic large space frames, alysis, and Ul: Applications, Journal of Structural Engineering, ASCE, ‘Val. 126, No. 3, May 2000, pp. 580-588 and pp. 589-595, 157. Ramaswamy, G.S., Eekhout, M., and Suresh, G.R., Analysis, Design and Construction of Steel Space Frames, Thomas Telford, 2002, 240pp. 158. Liew, LY.R., Punniyakotty, N.M., and Shanmugam, NE., Limit State ‘Analysis and Design of Cable-tensioned structures, International Journal of Space Structures, Vol. 16, No. 2, Apr. 2001. 159, Rajasekaran, S., Optimization of large Scale Three Dimensional eticulated Structures using Cellular Genetics and Neural Networks, International Journal of Space Structures, Vol. 16, No. 4, Oct. 2001. 160.Kaveh, A., and Servati, H., Neural Networks for the Approximate Analysis and Design of Double Layer Grids, International Journal of ‘Space Structures, Vol. 17, No. 1, Jan. 2002. 553Formex Data Generation sss Formex Data Generation 11.4 Introduction In general, the solution to a computer-aided structural analysis problem may be considered to involve three stages: preprocessing, processing and post-processing. In the preprocessing stage, data required for the analysis of the structure is generated. The structure is analyzed in the processing stage and the results from the analysis are presented in a neat form in the final stage. In the previous chapter, we were concerned about the second stage, ie, processing stage. In this chapter we will discuss the first stage. Experience shows that data generation for structures, and in particular, large and/or complex structures are a costly, time consuming and error- prone process. This fact has opened a new field of research and a large number of approaches to solve these problems have been published in the last few years. To date, most approaches suffer from the lack of generality. It is because they were developed for certain types of structural elements or geometrical properties. Formex algebra, on the other hand, stands alone as an algebra which provides a powerful mathematical basis for a new approach to data generation for structures of various geometrical and topological properties. Hence we shall discuss formex algebra and its details in the next sections. 11.2 Formex algebra Formex algebra is a mathematical system that consists of a set of abstract objects, known as formices and a number of rules in accordance with ‘which these objects may be manipulated. It provides a convenient basis for solution of problems in data generation and computer graphics.Space Structures: Principles and Practice ‘The early ideas on which formex algebra is based were developed by Prof. H. Nooshin during the years 197219737. These ideas are general and have applications in many branches of science and technology. In particular, the concepts are suitable for use in computer-aided design and the design of space structures; itis this area of applications that will be the focus of attention in this chapter. The early ideas went through substantial transformations during the years that followed and a full account of the concepts of formex algebra was published in 1984 as a book entitled: “Formex Configuration Processing in Structural Engineering”. After the publication of this book, formex algebra started getting worldwide acceptance. Nooshin’ * presents a survey of the current work in formex configuration processing where reference is given to structural applications in relation to double-layer grids, barrel vaults, domes, offshore oil platform jackets, transmission towers, cable networks and finite element meshes. Examples of application in the fields of geotechnics, entomology and school education are also included. In this ‘chapter, the emphasis is on the manner in which the concepts of formex algebra are employed rather than the details of the formulations. No prior knowledge of formex algebra is necessary for following the material. The chapter does contain some formex formulations but these are included to give a feel for the appearance of formex formulations rather than their details being essential for ‘understanding of the material 14.3 Formex approach to data generation ‘The formex approach to data generation for structures is somehow different from other generation schemes in that it provides a framework with the basic facilities and the possibility of simple implementation of many others. Formex approach is performed in two distinctive steps, namely a topological step and a geometrical step. In the first step a formex which represents the interconnection pattern of a structural configuration is formulated. A set of retrocords and a retronorm are employed in the second step to give the geometrical characteristics of the structure under consideration. Data for the analysis of structures consists of information about Interconnection pattern Joint co-ordinates Support positions Load positions Cross-sectional areas, modulus of elasticity of the material, ‘magnitude of loads and constraint particulars. Information relating to items 1 to 4 above may be conveniently Formes B generated using the concepts of formex algebra. In the next sections, some of these basic concepts are presented. 4113.1 Formices Consider the configuration shown in Fig. 11.1 and let it be required to produce a numerical model representing this configuration. Let it be assumed that the configuration of Fig. 11.1 represents a structural system and that the numerical model is intended to be used as data for the purposes of structural analysis. The constituent pars of the configuration that are to be represented by the numerical model depend on the actual system that the configuration is supported to represent. For instance if the configuration of Fig. 11.1 is representing the plan view of a fiat arid and if the structure Is to be analyzed with every line segment in the configuration taken as a beam element then the numerical model should provide information regarding the disposition of the beam elements and the nodal points as shown in Fig. 11.2. Here, the line segments represent beam elements and the dots represent nodal points. Alterratively, the configuration of Fig. 11.1 may be intended to represent a finite element mesh consisting of quadrilateral elements with comer nodes. In this case the configuration to be represented by the numerical ‘model should be as shown in Fig, 11.3. In fact, the configuration of Fig. 111 may be interpreted in many different ways and the above interpretations should only be regarded as particular examples. Other possitilties include the interpretation of the configuration as a view of a stiffened plate with the line segments representing the ribs and the interpretation as a plan view of a cable net. "Now, let the case of the flat grid interpretation be pursued further. Im oréer to create a numerical model for the grid, one has to adopt a suitable reference system. Such a reference system may be provided by two families of lines, shown as dotted lines together witt a part of the grid in Fig. 11.4. A’reference system of this kind is referred to as a “normat’, with the lines themselves being called ‘normét lines: Each normet line is identified by an integer and two directions are specified as shown. An element such as Ri in Fig. 11.4 may now be represented Fig. 11.1 sorSpace Structures: Principles aed Practice relative to the normat by a construct such as (1,452, 1) ‘where 1, 1 indicate the ‘narmat co-ordinates’ of point A and 2,1 indicate ‘the normat co-ordinates of point B and where the semicolon implies an ‘interrelation between the points represented by the normat co-ordinates 1, 1 and 2,1, The element R2 may be similarly represented by (2,1; 3,1) and the combination of the elements Ri and R2 may be represented by + Second direction Forme Data Generation 550 {0,15 2, 11 12,153, Mp ‘where the square brackets enclose the normat co-ordinates that belong to a component of the configuration and the curly brackets enclose 2 sequence of the parts that represent components. The constructs used above for representation of the elements RI, R2 and their combination are examples of mathematical objects that are known as ‘formices: Now, using variables to represent formices, one may write, FL= (1,1; 21), F2= [21331] and Fa = (U4; 24, (2.35 3.11) where Fl, F2 and F3 are ‘formex variables’ representing elements Rl, R2 and their combination, respectively. 14.3.2 Formex composition ‘The above formex F3 may also be written as FA=FL#P2 Where F1 # F2 implies ‘composition’ of two formices and where the symbol # is read as ‘duplus. The operation of composition allows formices representing various collections of the component parts of a configuration to be combined together. For instance, the combination, of the elements R1, R2, R3 and R4 as shown in Fig. 11.4, may be represented by B= P34 {0315 4.1), 1415 5,1} 11.3.3 Formex functions ‘The combination of the elements Ri, R2, R3 and R4 may also be ( written as E=rin (1,41) | FL Where rin (1,4,1) is a ‘formex function’ which is referred to as a ‘rindle function. "The term ‘rindle’ is an old English word meaning ‘watercourse’ and this word is used to imply a serial replication in a particular direction. The function rin (1,4,1) implies four replications in. the first direction with one unit of translation at each step. The significance of various parts of the function is indicated in Fig. 11.5. Using a rindle function, one may represent all the eight elements of the grid along the line AB, in Fig. 11.4 by F=rin (1,81) FL