You are on page 1of 10

Mechanism of Collapse of Space T russes with Steel Hollow Circular Bars with Flattened Ends

Alex Sander Clemente de Souza and Roberto Martins Gonalves


Department of Structural Engineering, University of So Paulo at So Carlos campus Av. Trabalhador So-Carlense, 400 CEP13.566590 So Carlos, SP Brazil e.mail: goncalve@sc.usp.br (Received 5th July 2005)

ABSTRACT: This paper presents the results of experimental analysis of space trusses using steel tubular bars with flattened ends. The connections are formed by overlapped bars connected by single bolt. Such system has been used due to the low cost and assembly facilities. The behavior and collapse modes were determined by experimental analysis on six space trusses with 1.5 m height and spans of 7.5 m 7.5 m and 7.5 m 15 m. Structures with steel nodes in the top corner and supports points were also tested. The structural collapse was caused by either connection collapse or yielding in the bar ends. Traditional theoretical analysis models (linear truss model) are not suitable for these structures.

INTRODUCTION
Over the last two decades there has been an increase in the use and development of space trusses for different applications like gymnasiums, exhibition pavilions and hangars. In other words, structures with large free areas and long spans. The space truss system has some advantages over the conventional ones1. A suitable design is necessary to explore the advantages of space trusses. It should integrate structure and architecture with economy of material, durability, production and assembly easiness. Guidelines for the design and classification of space trusses have been presented by MAKOWSKI1, IFFLAND2, AGERSKOV3 and MARSH4. In many types of space trusses hollow circular section bars are frequently used. The difficulty in making connections details is a disadvantage when hollow circular sections are used. The main problem of design and development in space trusses are the connection systems, because they
International Journal of Space Structures Vol. 20 No. 4 2005

have significant influence on the cost, structural behavior and collapse modes. The choice of a connection system depends on structural layout, types of sections and distribution of bars. Many patented or ad hoc connection systems are available and new ones continue to be invented but the code of practice does not include specific design rules for these connections. Bars are connected to a spherical node, hub or joint block in the patented connection systems. MERO system (fig. 1) was the first patented system for space structures1. Many other patented systems were also developed IMAI5, IWATA & KAMIYAMA6, LANDOLFO & MAZZOLANI7. Designers and researchers have used ad hoc systems GERRITS8, GANDOLFI9, De MARTINO10, COOD11, EL-SHEIKH12 and CUENCAS13 due to their lower cost than patented systems.
201

Mechanism of Collapse on Space Trusses with Steel Hollow Circular Bars with Flattened Ends

Figure 1. Mero connection system

There are systems without a special nodal piece like the ones studied here. In this case, the chord bars have flattened ends and can be continuous or not. Diagonal bars are flattened and bent at the ends, Fig. 2. The typical node, which is reported in this paper, is a connection system, similar to the one previously shown and largely used in Brazil, (Fig. 3). CUENCAS13 describes a gymnasium built in Spain where a similar typical node was used, Fig. 4.

Figure 4. CUENCA (2002)13 connection system

Connections like the ones shown in figures 2, 3 and 4 are easy to produce and assemble, and are low in cost, however aspects of structural behavior, collapse modes and theoretical analysis should be carefully evaluated. In Brazilian design offices, the space trusses are usually analyzed using linear elastic models (a linear truss model), without eccentricity, with flexible nodes and uniform sections along the bars length. However, in space trusses with a connection system previously described those hypotheses are not confirmed. In this case it is necessary to include the nodal stiffness, section variation in the bar ends, eccentricity in the connections, physical and geometrical nonlinearities.

TYPICAL NODES FEATURES


The typical node (Fig. 3) is the most popular connection system in Brazil. The main advantages are low cost, easy production and easy assembly. These connection systems have an important influence on the structural behavior of the compression bars and the whole space truss. Their flattened ends may lead to a reduction in the compression strength capacity and premature failure by local instability. A direct test of compression bars with flattened ends shows a reduction of up to 50% in compression strength in relation to the theoretical results calculated without flattened ends SOUZA14. Strut bars, with slenderness less than 60 are more critical. SOUZA14 presented an approximate procedure for the design of compression bars with flattened ends, but it is very complicated to use in design offices as the shape and length of the flattening do not follow a standard manufacturing process. The behavior of compression members is a function of node stiffness, bar slenderness and the other connected members in the same node. Therefore fullscale tests are necessary to obtain compression bars behavior and space trusses collapse modes.

Figure 2. Systems without a special nodal piece11, 12

Figure 3. Typical node


202

International Journal of Space Structures Vol. 20 No. 4 2005

Alex Sander Clemente de Souza and Roberto Martins Gonalves

Eccentricity is another problem in typical nodes. A local failure by yielding is possible because end flattened tubes have low flexural stiffness and eccentric loads induce bends in bars. A local failure by yielding in critical regions, such as support nodes or corner nodes, may lead to structural collapse. Connection failure is the most common cause of structural collapse which occurred in Brazil and are the object of investigation by the authors. Ten years ago the authors of the present paper began a wide theoretical and experimental study of space trusses. The objective was to determine the collapse modes for several connection systems, mainly the typical node, and develop criteria for analysis, design and construction. The authors conducted their research with tests in space truss prototypes of different connection systems15,16,17. The main experimental results for the typical nodes are presented in this paper.

The prototypes were divided into two groups (Table 1). In the top corner of ProtA-3 and ProtB-3 a steel node was used. Fig. 6 presents the details of the nodes and bars ends. Materials ASTM A325 bolts with 19 mm diameter were used. Connection plates and steel nodes were made of ASTM A36 steel (fy = 250 MPa). The tubular bars were made of ASTM A36 steel for series A and ASTM A570 steel (fy = 290 MPa) for series B.

EXPERIMENT AL PROGRAM
Six space truss prototypes with 1.5 m height and spans of 7.5 m 7.5 m and 7.5 m 15 m were tested see Fig. 5a and Fig. 5b.

Figure 5a. Space truss (A series: protA-1, protA-2, protA-3)

Figure 5b. Space truss (B series: protB-1, protB-2, protB-3)

Table 1 Prototypes tested Prototypes ProtA-1 ProtA-2 ProtA-3 ProtB-1 ProtB-2 ProtB-3 Dimensions 7.5 m 7.5 m Support Diagonal (mm) 76 2.0 88 2.65 88 2.65 60 2.0 88 2.65 88 2.65 Steel ASTM A36 Steel node in top corner and supports. ASTM A570 Steel node in top corner and supports. Observations

15 m 7.5 m

International Journal of Space Structures Vol. 20 No. 4 2005

203

Mechanism of Collapse on Space Trusses with Steel Hollow Circular Bars with Flattened Ends

a) Typical node corner (protA-1, protA-2, protB-1 e protB-2)

b) Typical node support (protA-1, protA-2, protB-1 and protB-2)

c) Steel node corner (protA-3 and protB-3)

d) Steel node support (protA-3 and protB-3)

Figure 6. Details of the nodes and bar ends

Assembly and Test Procedures The prototypes were assembled on a reactions slab with a rigid base of columns. Fig. 7 shows some steps of the assembly of the prototypes. Loads were applied to Series A prototypes to four bottom nodes and to ten bottom nodes of Series B prototypes. Special steel nodes were used in the load applications points.

The displacements were measured by LVDT in the bottom and top nodes. The same bars in the central regions and in the corners were chosen to measure strains. The measurements were taken in middle section bars and in two sections in the flattened ends. Fig. 8 shows the set up details for the test.

(a)

(b)

(c)

Figure 7. Assembly of prototypes

204

International Journal of Space Structures Vol. 20 No. 4 2005

Alex Sander Clemente de Souza and Roberto Martins Gonalves

fy fp = 0.5fy 0.1E

E = 20500 kN/cm2

Figure 8. Set up test

THEORETICAL ANALYSIS
Theoretical analyses by Finite Element using ANSYS software were performed. In the first analysis both nodal displacements and bar forces were determined by a linear truss model and the theoretical ultimate strength was determined as a function of the bars capacity. The second analysis with eccentricity, inertia variation in end bars and physical non-linearity was further included in the numerical analysis.
Tube 88 2.65
section 1 section 2 section 3

Figure 11. Stress vs. strain relationship

In order to include the section variation, the bars were divided as shown in Fig. 9. The end bends of diagonal bars were inserted in the analysis model to reproduce the eccentricities according to the nodal geometry Fig. 10. Experimental results show that the structural collapse in space trusses depends on both rotation and yielding in the bar ends. The flattened ends in the bars cause residual stress in the transversal sections; therefore the stress vs. strain relationship of Fig. 11 was adopted.

RESULTS AND DISCUSSION


circular section

140 mm

90 mm

90 mm

section 1

section 2

section 3

circular section

Figure 9. End bars for numerical analysis

eccentricity 50 mm e Diagonal eccentric node Node

The collapse mode has been generally associated with connections failure Figure 11. Only in prototype ProtA-3, due to the existence of steel nodes in the top corner and supports, the structural collapse was caused by the instability of the top chord bars Fig. 12. The experimental loads were lighter than the theoretical ones for ideal truss analysis Table 2. In the nonlinear analysis with eccentricity and sections variation the correlation between the experimental and theoretical loads was better Table 3. The difference between the experimental and theoretical loads was significant for the displacements even with eccentricity and nonlinear analysis in the theoretical model, (Table 4). Table 2 Experimental vs. theoretical loads elastic linear analysis Prototype Fexp (kN) 161.3 160.2 179 93 71 148.8 Fteo (kN) 259.7 245.9 259.7 129.7 162.8 162.8 Fexp/Fteo 0.62 0.65 0.69 0.72 0.43 0.91

138 mm

1500 mm

8 mm

Diagonal centered node

60 2.0 mm

Node

Figure 10. Modeling of eccentricities


International Journal of Space Structures Vol. 20 No. 4 2005

ProtA-1 ProtA-2 ProtA-3 ProtB-1 ProtB-2 ProtB-3

205

Mechanism of Collapse on Space Trusses with Steel Hollow Circular Bars with Flattened Ends

Table 4 Theoretical vs. experimental displacements nonlinear analysis Prototype ProtA-1 ProtA-2 ProtA-3 ProtB-1 ProtB-2 ProtB-3
a) ProtA-1

Dexp (kN) 7.5 6.56 2.6 4.62 4.56 7.9

Dtheo (kN) 3.75 3.75 3.0 5.5 3.55 7.6

Dexp/Dtheo 2.0 1.75 0.87 0.84 1.28 1.04

b) ProtA-3

The curve load vs. displacement shows the nonlinear behavior due to the yielding of the flattened ends and slips in the nodal region. These phenomena justify the differences between the experimental and theoretical results, as the slip was not considered in the theoretical analysis. The graphs of Fig. 13 and Fig. 14 show the load vs. displacement curves. According to the graphs the elastic linear analyses are not adequate for these space trusses. The eccentricity, inertia, end-bar variation and nonlinearities are very important for the behavior of space trusses, therefore they must be included in the theoretical analysis see Fig. 15 and Fig. 16.
200 180 160 140 Load (kN) 120 100 80 60 40 20 0 0 1 2 3 4 5 Displacement (cm) 6 7 8 9 ProtA-1 ProtA-2 ProtA-3 theoretical linear

c) ProtB-1

Figure 12. Collapse mode

Figure 13. Load vs. displacement experimental curves Series A


"  

 =@ 

Table 3 Experimental vs. theoretical loads nonlinear analysis Prototype ProtA-1 ProtA-2 ProtA-3 ProtB-1 ProtB-2 ProtB-3 Fexp (kN) 161.3 160.2 179 93 71 148.8 Ftheo (kN) 195 195 215 90 79 200 Fexp/Ftheo 0.83 0.82 0.83 1.03 0.9 0.74

 & $ "  


0 1 2 3 4 5
A J ?  2H J*  ANF 2H J* ANF

2H J* ! ANF JDA HAJE?= E A=H

,EIF =?A

Figure 14. Load vs. displacement experimental curves Series B

206

International Journal of Space Structures Vol. 20 No. 4 2005

Alex Sander Clemente de Souza and Roberto Martins Gonalves

200

1 00

Load (kN)

150

80

Load (kN)

60

100 ProtA-1 ProtA-2 Theoretical

40

50

20
0 0 1 2 3 4 5 6 7 8 9

ProtB-1 experimental ProtB-1 theoretical ProtB-2 experimental ProtB-2 theoretical


0 1 2 3 4 Displacement (cm) 5 6

Displacement (cm)
250 200
220 200 180 160 140 120 100 80 60 40 20

Load (kN)

100 50 0 0 1 2 3 4

Theoretical ProtA-3

Load (kN)

150

ProtB-3 experEmental ProtB-3 theoretEcal


0 1 2 3 4 5 6 7 8 9

Displacement (cm)

Displacement (cm)

Figure 15. Load vs. displacement experimental curves nonlinear analysis (Series A)

Figure 16. Load vs. displacement experimental curves nonlinear analysis (Series B)

100

80

Load (kN)

60

40

20

ProtB-1 1 - exp 2 - exp Theoretical


0 100 200 300 400 500

Points of strain measurements


100
100

Strain ()

80

80

Load (kN)

40

Load (kN)
ProtB-1 - experimental point 5 point 3
6000 5000 4000

60

60

40

20

ProtB-1 - theoretical point 5 point 3


3000 2000 1000 0

20

ProtB-1 - theoretical point 4 point 6


0 1000 2000 3000

ProtB-1 - experimental point 4 point 6


4000 5000 6000 7000

0 -7000

Strain ()

Strain ()

Figure 17. Strain on the support diagonal ProtB-1

International Journal of Space Structures Vol. 20 No. 4 2005

207

Mechanism of Collapse on Space Trusses with Steel Hollow Circular Bars with Flattened Ends

The behavior of the structures (except TrussB-3), mainly for the displacements in the first load steps, is well represented by the numerical model analysis. In ProtB-3 both structural arrangement and slip between bars in the nodal region justify the disagreement between the theoretical and experimental results. The theoretical analysis model does not include the slip between bars in the nodal region. Elastic linear strains were found in the middle sections of the bar, but in the end sections plastic strains with complicated pattern were found. Fig. 17 shows some experimental strain results. Fig. 17 shows that the results of theoretical nonlinear analysis with eccentricity and inertia variations in bars are very close to the experimental ones.

not be used in support nodes. The length of the flattened ends must always be as short as possible.

ACKNOWLEDGEMENTS
The authors would like to thank the Brazilian research institutions FAPESP and CNPq for their financial support for this research work.

REFERENCES
1. MAKOWSKI, Z. S., Review of development of various types of double-layer grids. In: MAKOWSKI, Z. S., ed. Analysis, design and construction of Double-layer grids. Applied Science., p. 155. 1981. IFFLAND, J., Preliminary planning of steel roof space trusses. Journal of the Structural Division, v. 108, n. 11, p. 25782589. Nov, 1982. AGERSKOV, H., Optimum geometry design of doublelayer space structure. Journal of Structural Engineering, v. 112, n. 6, p. 14541463. June, 1986. MARSH, C., Some observations on designing double layer grids. International Journal of Space Structures, v. 15, n. 3/4, p. 225231. 2000. IMAI, K. et al., The KT space truss system. In: INTERNATIONAL CONFERENCE ON SPACE STRUCTURES, 4, Guildford, UK, Sept. 1993. Proceedings. London, Thomas Telford. v. 2, p. 13741382. 1993. IWATA, M., KAMIYAMA, K., Development and projects of the NS space truss system. In: INTERNATIONAL CONFERENCE ON SPACE STRUCTURES, 4., Guildford, UK, Sept. 1993. Proceedings. London, Thomas Telford. v. 2, p. 14171426. 1993. LANDOLFO, R., Qualification analysis of a new structural system. In: INTERNATIONAL CONFERENCE ON SPACE STRUCTURES, 4., Guildford, UK, Sept. 1993. Proceedings. London, Thomas Telford. v. 1, p. 693702. 1993. GERRITS, J. M., Space structures in the Netherlands since 1975. In: INTERNATIONAL CONFERENCE ON SPACE STRUCTURES, 3., Guildford, UK, Sept. 1984. Proceedings. London/New York, Elsevier Applied Science. p. 2832. 1984. GANDOLFI, A., Hangar per il montaggio e la revisione di velivoli nello stabilimento aeritalia de Capodichino Nord. Costruzioni Metalliche, n. 6, p. 317324. 1989. DE MARTINO, A., Relazione generale: progettaziopne, lavorazione e montaggio. Costruzioni Metalliche, n. 1, p. 1454. 1992. COOD, E. T., Low technology space frames. INTERNATIONAL CONFERENCE ON SPACE STRUCTURES, 3., Guildford, UK, sept. 1984. Proceedings. London, Elsevier Applied, p. 955960. 1984. EL-SHEIKH, A. I., Experimental study of behavior of new space truss system. Journal of Structural Engineering, v. 122, n. 8, p. 845853. Aug, 1996. CUENCAS, L. S., The stainless steel structures of a sport stadium in quart. In: INTERNATIONAL CONFERENCE ON SPACE STRUCTURES, 5., Guildford, UK, Aug. 2002. Proceedings. London, Thomas Telford. v. 1, p. 547556. 2002. SOUZA, A. S. C., GONALVES, R. M., Influence of section variation by stamping of bar ends the behavior

2.

3.

CONCLUSIONS
Tests on space truss prototypes were performed to determine the collapse pattern. The results of experimental and theoretical analyses allow to conclude that: 1. The connections failure was the predominant collapse pattern and was caused by plastic strains in the end bars, node rotations and slip among bars in the nodal region. These facts caused an increase in the displacements and the premature structural collapse. 2. The nodes failure began in the top vertices, where there were the worst eccentricity conditions. 3. The structural collapse pattern was modified by the use of steel nodes (Fig. 5b) in the top vertices (ProtA-3 e ProtB-3). Top chord instability was the cause of collapse in ProtA-3 and excessive displacement was the convectional collapse in ProtB-3. 4. Collapse pattern differences between ProtA-3 and ProtB-3 occurred due to the differences in the ratio module length by span (3 for ProtA-3 and 6 for ProtB-3) and ratio height by span (5 for ProtA-3 and 10 for ProtB-3). 5. The relationship between the experimental and the theoretical loads was better in non-linear analysis with eccentricity and inertia variation in bar ends, but for displacements the experimental and theoretical agreement was not good. Differences of up to 100% between theoretical and experimental displacements were found. 6. Low cost is the main reason for the use of connection systems with flattened end bar. However the results presented here show critical points in the design: this connection system should

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

208

International Journal of Space Structures Vol. 20 No. 4 2005

Alex Sander Clemente de Souza and Roberto Martins Gonalves

15.

of space trusses. In: NORDIC STEEL CONSTRUCTION CONFERENCE, 9., Helsinki, Finland, 1820 June, 2001. SOUZA, A. S. C., GONALVES, R. M., MALITE, M., Behaviour of tubular space truss connections with stamped end bars. In: PARKE, G. A. R., DISNEY, P. (Ed.). Space structures 5 (Proc. 5th International Conference on Space Structures, Guildford, UK, 1921 August 2002). London, Thomas Telford, 2002. v. 1, p. 337345 (ISBN: 0-7277-3173-4).

16.

17.

MALITE, M., MAIOLA, C. H., GONALVES, R. M., SOUZA, A. S. C., Experimental analysis of the structural performance of space trusses commonly used in Brazil. International Journal of Space Structures, v. 16, n. 4, p. 253260, 2001. (ISSN: 0266-3511). SOUZA, A. C. S., GONALVES, R. M., MALITE, M., MAIOLA, C. H., Theoretical analysis of the structural performance of space trusses commonly used in Brazil. International Journal of Space Structures. v. 18, n. 3, p. 167179, 2003. (ISSN: 0266-3511).

International Journal of Space Structures Vol. 20 No. 4 2005

209

You might also like