Sunteți pe pagina 1din 7

Engineering Structures 22 (2000) 116122 www.elsevier.

com/locate/engstruct

Structural stability: from theory to practice


W.F. Chen
*

School of Civil Engineering, Purdue University, West Lafayette, IN 47907, USA Received 27 July 1997; accepted 9 January 1998

Abstract Over the past 40 years drastic improvements in our knowledge regarding the behavior, strength and design of steel building frames has been achieved. This paper provides several specic examples in which new knowledge has been implemented and better design methods have been advanced in engineering practice. The directions of possible immediate implementation of some recent developments in advanced analysis for practical frame design are outlined. 1999 Elsevier Science Ltd. All rights reserved.
Keywords: Advanced analysis; Buildings; Design; Effective length factor; Plasticity; Stability; Steel; Structural engineering

1. Introduction What measuring stick should be used to assess the accomplishment of structural stability research for steel building frames in the past 40 years? Should it be the volume of papers presented, or the number of journal articles published, or the number of PhD theses produced or the number of new courses in the universities offered? I believe that the bottom line for the structural engineering profession should be the amount of research which nds its way into practice. The profession has developed in the past decades in important ways as a direct result of these intensive research activities worldwide. The SSRC-related success stories that can be attributed to these developments fall into a number of broad categories. For the Symposium honoring Professor T.V. Galambos, it is the most appropriate occasion to summarize these categories in this paper and provide several specic examples within each category where new knowledge has been implemented and, in some measure, a better understanding of the behavior of structural members and systems has been developed and better design methods have been advanced. Directions of possible immediate implementations of some recent developments for engineering practice are outlined here.

2. Behavior and design of structural members Perhaps in no other area has there been such drastic improvement in our knowledge regarding the behavior, strength and design of structural members including columns, beams and beam-columns using the mainframe computing and nite element methods in the 1960s and 1970s. There has been a steady ow of results from SSRC research into the development of improved codes and standards governing the design of structural members in building codes. Major changes were made, for example, in the design of biaxially loaded columns. The studies ranged from full-scale tests to complex niteelement analysis of beam-columns under various load combinations in plane and in space. The information produced has been implemented in AISC building codes and Euro-codes and has become standard practice in the 1980s. These and other related developments were summarized in a two-volume book by Chen and Atsuta in 197677 [1] as well as the 1988 SSRC Guides edited by Galambos [2]. Another area where signicant advances have been made is in the design of large fabricated cylindrical members as used in deep-water offshore structures. Several specic areas where the research results were instrumental in bringing about major changes in API codes and other practices for engineering in offshore structures include the effect of hydrostatic pressure on column strength, the beam-column strength and behavior considering dent damage effects, and the assessment of the

* Tel: 1-765-494-2254; Fax: 1-765-496-1105

0141-0296/00/$ - see front matter 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 1 4 1 - 0 2 9 6 ( 9 8 ) 0 0 1 0 0 - X

W.F. Chen / Engineering Structures 22 (2000) 116122

117

strength of internally grout-repaired damaged members. This information and related developments were summarized in the books for example by Chen and Han [3] and Chen and Toma [4] among others.

3. Structural design with K factors In current engineering practice, the interaction between the structural system and its members is represented by the effective length factor (Fig. 1). This classical approach to structural design is described clearly in the 1981 SSRC Technical Memorandum No. 5 [5], which provides the basis for the development of modern steel design methods including the popular load resistance factor design (LRFD) and allowable stress design (ASD) methods [6]. The effective length method generally provides a good method for the design of framed structures. However, despite its popular use in the past and present as a basis for design, the approach has major limitations. The rst of these is that it does not give an accurate indication of the factor against failure, because it does not consider the interaction of strength and stability between the member and structural system in a direct manner. It is a well recognized fact that the actual failure mode of the structural system often does not have any resemblance whatsoever to the elastic buckling mode of the structural system that is the basis for the determination of the effective length factor K. The second and perhaps the most serious limitation is probably the rationale of the current two-stage process in design: elastic analysis is used to determine the forces acting on each member of a structural system, whereas inelastic analysis is used to determine the strength of each member treated as an isolated component. There is no verication of the compatibility between the isolated member and the member as part of a frame. The individ-

ual member strength equations as specied in specications are not concerned with system compatibility. As a result, there is no explicit guarantee that all members will sustain their design loads under the geometric conguration imposed by the framework. The other limitations of the effective length method include the difculty of computing a K factor, which is not user-friendly for a computer-based design, and the inability of the method to predict the actual strength of a framed member, among many others. To this end, there is an increasing awareness of the need for practical analysis/design methods that can account for the compatibility between the member and system. With the rapid increase in the power of desktop computers and user-friendly software in recent years, the development of an alternative method to a direct design of structural system without the use of K factors becomes more attractive and realistic. The real challenge is making this type of new approach to design work and competitive in engineering practice. An extensive research on this topic, now known as advanced analysis to design, has been made at several universities around the world for many years, and signicant advancements have been made, although much more remains to be done. 4. Advanced analysis to design Extensive research has been devoted to the development and validation of several advanced analysis methods. A promising technique of developing highorder beam elements (making only one or two necessary to describe the behavior along a members length) is in progress at Cornell University [7]. An intensely rigorous method using workstations and super-computers to solve thousands of degrees of freedom has been in development around the world for many years [8,9]. Simple calibration techniques and practical approaches have been researched here at Purdue [10]. Intermediate solutions include plastic-zone, quasi-plastic hinge, elasticplastic hinge methods and various modications thereof. All in some way account for residual stresses, geometric imperfections, non-linearities and moment redistribution throughout a structure. Briey they are outlined below: 1. Plastic zone [8,9]: a. Discretized nite elements along the length and through the cross-section. b. Captures the incremental load-versus-deection response considering the second-order geometric distortion. c. A constant residual stress pattern is assumed. d. The spread of plasticity is traced. 2. Quasi-plastic zone [7]: a. A compromise between plastic zone and elastic plastic hinge methods.

Fig. 1. Interaction between a structural system and its component members.

118

W.F. Chen / Engineering Structures 22 (2000) 116122

b. c. d. e.

The spread of plasticity is considered by exibility coefcients. A simplied residual stress pattern is used. The fully plastic cross-section is calibrated to the plastic zone solution. There is no potential to upgrade this from its current two-dimensional restriction.

3. Elasticplastic hinge [11] a. Zero length plastic hinges. b. No spread of yielding through the cross-section, or along the length. c. No consideration of residual stresses. d. Second-order geometric effects can be considered. 4. Rened plastic hinge method [12]: a. A step up from the elastic plastic model for two dimensions. b. Distributed plasticity-smooth stiffness degradation of a hinge. c. Inelasticity is considered indirectly by forces rather than strains. Tangent (Et) modulus is used to describe the effect of residual stresses. d. Stiffness degradation function is used for gradual yielding. e. Connection exibility can be modelled using rotational springs. 5. Practical rened plastic hinge method [13]: a. The rened model (4. above) is made practical by calibration to the LRFD empirical code equations. b. A separate modication of tangent modulus (Et) is imposed to consider geometric imperfections. c. The CRC tangent modulus model is used allowing residual stresses to be considered separately. For the last method to work effectively on popular commercial programs in use in design ofces today further changes are needed. The Purdue method [14] was developed to perform designs, using simple modications to elastic parameters familiar to LRFD users, comparable to those achieved by traditional code proceduresbut a unique program was required. The advanced capabilities included two modications for material non-linearity and one for geometric imperfections. The degradation of stiffness due to gradual yielding of a cross-section subject to exural moments was dened as shown in Fig. 3. Residual stresses were accounted for by reducing the tangent modulus Et as plotted in Fig. 2. The last non-linear effect, geometric imperfections, was considered by applying a further reduction to Et. Details of this development will be summarized in the following section.

Fig. 2. Moment-curvature relationship for a perfect plastic and work hardening hinge.

Fig. 3. Member tangent stiffness degradation derived from the CRC column curve.

Two difculties were noted with regard this method, which, being changed, would result in a practical method of using current software to achieve the same types of analysis. 1. The choice of using stability functions to account for the P- effect does not translate well into the nite element world. Stability functions use small deformations theory implicitly to capture an effect that many nite element analyses account for explicitly. The advantage of this technique is only one element is needed per member. This is becoming less and less of a driving issue with the speed of computer analysis today, and the price of having more elements along the length (while capturing behavior taking place outof-alignment with the centre-line of the members) is small. This allows the nite element method (as it exists in most all structural analysis packages) to be used as they stand.

W.F. Chen / Engineering Structures 22 (2000) 116122

119

2. The degradation in stiffness modication factors was applied by the Purdue team to coefcients in the stiffness matrix of each element. It was required that the fundamental solution algorithms be re-coded in order to consider this non-linear behavior simply. This difculty will be overcome with this method by taking advantage of the iterative capabilities inherent (or soon to be implemented) in common analysis packages [15]. Non-linearities due to material behavior fast becoming available in commercial codes, and using this to dene a failure criterion would present a convenient and practicable advanced analysis method.

The reduction factors are selected as 0.85 for axial strength and 0.9 for exural strength just as the LRFD specication does. 5.3. Residual stresses The CRC tangent modulus is employed to account for the gradual yielding effect due to residual stresses along the length of members under axial loads. In this approach, the elastic modulus E instead of the moment of inertia I is reduced to account for the reduction of the elastic portion of the cross-section, because the reduction of elastic modulus is easier to implement than that of the moment of inertia for different sections. The reduction rate in stiffness for both strong and weak axis is taken to be the same and this reduction is reected by the CRC Et as (Fig. 2). Et 1.0E Et 4 for P 0.5Py (3a) (3b)

5. Structural design without K factor In the following, I shall briey summarize a practical solution to the problem by simply modifying an elastic program with a modied tangent modulus based on the familiar CRC column strength equation together with a rened plastic hinge concept. These modications consider the following key behavioral effects of a steel member: second-order, gradual yielding associated with residual stresses and exure and geometric imperfections. To meet the current LRFD requirements, these modications have been calibrated against the LRFD specication. 5.1. Second-order effects To capture second-order effects, the simplied stability functions reported by Chen and Lui [16] is adopted. The incremental force-displacement relationship of a member may be written as, in the usual notations

P P E 1 for P > 0.5Py Py Py

5.4. Distributed plasticity When idealized plastic hinges are formed at the member ends, the elastic stiffness at the ends will be reduced abruptedly to zero (Fig. 3). To represent a gradual transition from the elastic stiffness at the onset of yielding to the stiffness associated with a full plastic hinge at the ends, the parameter representing a gradual stiffness reduction associated with exure is introduced with 0 1.0 according to the parabolic expression (Fig. 4)

4(1 ) for > 0.5

(4)

A A M S1 S2 0 EI B B S2 S1 0 M I 0 0 A/I e P

(1)

where S1 and S2 are stability functions, for in-plane bending of a prismatic beam-column. 5.2. Cross-section plastic strength The LRFD cross-section plastic strength curves are adopted for both strong and weak-axis bending P 8 M 1.0 cPy 9 bMp P M 1.0 2cPy bMp for P 0.2 cPy (2a)

for

P 0.2 cPy

(2b)

Fig. 4. Parabolic plastic hinge stiffness degradation function with o 0.5 based on LRFD sectional strength equation.

120

W.F. Chen / Engineering Structures 22 (2000) 116122

where is the force-state parameter obtained from the limit state surface corresponding to the member ends (Fig. 5)

P 8 M Py 9 Mp P M 2Py Mp

for

P 2 M Py 9 Mp

(5a)

for

P 2 M Py 9 Mp

(5b)

similar to the LRFD plastic sectional strength expressions. This is known as the rened plastic hinge concept. It reects the distributed plasticity effects associated with bending actions at the member ends. When these rened plastic hinges are present at both ends of a member, the incremental elastic forcedisplacement relationship as given in Eq. (1) can now be modied to include both the inelasticity within the member by using Et instead of E and the distributed plasticity at the ends by using the rened plastic hinge concept using the parameter as

Fig. 6. CRC and reduced tangent modulus for members with geometrical imperfection.

5.5. Geometric imperfections The degradation of member stiffness due to geometric imperfections may be simulated by a further reduction of member stiffness. This may be achieved simply by a further reduction of the tangent modulus Et as (Fig. 6) E t 0.85 Et (7)

P EtI L

A M B M

(6)
S2 2 (1 B) S1

A S1

ABS2
S2 2 B S1 (1 A) S1 0

ABS2
0

0 A I


A B e

where A and B are stiffness reduction factor as given in Eq. (4) at end A and end B, respectively. Details of this development are given elsewhere [14].

Herein, the reduction factor 0.85 is used to reduce further the CRC Et as given in Eq. (3a) and (3b) to include the effect of geometric imperfections (Fig. 6). Thus, if the modulus E in the Euler buckling formula is replaced by E t , the column strength curve as specied by the LRFD specication will be obtained within a maximum error of no more than 5% (Fig. 7). The further reduced modulus E t is applicable for both braced and unbraced members and frames.

Fig. 5. Smooth stiffness degradation for a work-hardening plastic hinge based on LRFD sectional strength curve.

Fig. 7. Comparison of column strength curves with further reduced tangent modulus method.

W.F. Chen / Engineering Structures 22 (2000) 116122

121

6. Behavior and design of structural system When I rst began to work in structural stability over 35 years ago, evaluation of the rst-order response of a structural system was a signicant problem. This includes linear elastic analysis and simple plastic analysis, and the progress made to the present state-of-the-art, which deals routinely with second-order inelastic analysis of complicated structural systems having hundreds of thousands of degrees of freedom, is miraculous. The modelling of all types of structural systems of high-rise buildings can now be handled quickly and efciently on relatively inexpensive computers. The primary limitation is a sufcient understanding of the response of some secondary structural elements such as concrete oor slabs, composite joints and walls that make up that system to develop simple but realistic models that can be incorporated into the analysis programs. The advent of personal computers, particularly in the computing and graphics performance of engineering workstations, has made more sophisticated methods of analysis feasible in design practice. While the use of rst-order analysis for elastic or plastic design is still the norm of engineering practice, a new generation of codes has emerged that recommends the second-order theory as the preferred method of analysis. The basic theory for second-order inelastic analysis is well established and documented in open literature. The real challenge is making this type of analysis work in engineering practice. The advanced analysis approach to design as illustrated in Fig. 8 can predict more accurately the possible failure modes of a structure, exhibit a more uniform level of safety, and provide a better long-term serviceability and maintainability.

rigid or exible joints in a steel frame (Chen and Sohal, [18]). The power of this tool is as follows: with the ability to predict the actual moment distribution at load levels that require members to sustain their plastic moment capacity, breaks can be strategically located throughout a structure. These structural fuses can be designed to fail themselves without the risk of the building as a whole falling down, while leaving the majority of the connections in satisfactory condition (AISC, [19]). This would not only limit the amount of post-quake repair necessary, but would also indicate where the failed connections were and thus greatly reduce the expense of exploratory procedures. Some of the new aspects that can be further considered in design practice when performance based (via advanced analysis) becomes standard practice are: Semi-rigid connections and thus partly restrained members have been researched extensively but this knowledge has not been able to be easily implemented into code-based design practice. Unless computer methods are adopted, small practical use can be garnered from the new knowledge of connection rotation characteristics and their effect on the global behavior of frames. Furthermore, three-dimensional behavior is a natural extension. Even only partly considering out-of-plane behavior (e.g. lateral torsional buckling) becomes possible.

8. Summary The practical design method presented here introduces the potential (and emphasizes the necessity) of advanced analysis procedures. Long accustomed to isolated member-by-member capacity checks, one analysis now considers all components and their interdependence. The global analysis provides information on the failure mode and thus allows an assessment of damage sustained at collapse loads. If the damage can be predicted, it can be controlled by design procedures calibrated to maintain adequate performance criteria (as opposed to the Codes traditional safety levels). Only when design engineers are assured of the validity and convinced on the practicality of performance-based analysis and design will this and other advanced capabilities be implemented in dayto-day ofces and the results of many years of research be granted their true fulllment: a place in Practice.

7. Seismic design with structural fuse The analytical capability of tracing the performance of a structure into the non-linear range required by seismic loads is also available [17]. Advanced analysis combines the theory of stability with the theory of plasticity and traces the gradual plastication of members with

References
[1] Chen WF, Atsuta T. Theory of beam-columns, Vol. 1, In-plane behavior and design, 1976, Vol. 2, Space behavior and design, 1977. New York: McGrawHill. [2] Galambos TV, editor. Guide to stability design criteria for metal structures, 4th edn. WileyInterscience, 1988.

Fig. 8.

Analysis and design methods.

122

W.F. Chen / Engineering Structures 22 (2000) 116122

[3] Chen WF, Han DJ. Tubular members in offshore structures. London: Pitman, 1985. [4] Chen WF, Toma S. Analysis and software of cylindrical members. Boca Raton, FL: CRC Press, 1996. [5] SSRC. General principles for the stability design of metal structures. Technical Memorandum No. 5, Civil Engineering, ASCE, February 1981, pp. 5354. [6] Chen WF, Lui EM. Structural stability: theory and implementation. New York: Elsevier, 1987. [7] Deierlein GG. Steel-framed structures. Progress in structural engineering and materials, Vol. 1, No. 1. London: CRC Ltd., September 1997. [8] Clarke MJ, Bridge RQ, Hancock GJ, Trahair NJ. Advanced analysis of steel building frames. Journal of Constructional Steel Research 1992;23(13):130. [9] McGuire W. Computer-aided analysis. In: Dowling PJ, Harding JE, Bjorhovde R, editors. Constructional steel designand international guide. Elsevier Applied Science, 1992:915932. [10] Chen WF, Toma S. Advanced analysis of steel frames. Boca Raton, FL: CRC Press, 1994. [11] White DW, Chen WF. Plastic-hinge based methods for advanced analysis and design of steel frames. Bethlehem, PA: Structural Stability Research Council, Lehigh University, 1993.

[12] Liew JYR, White DW, Chen WF. Second-order rened plastic hinge analysis for frame design: Parts 1 and 2. Journal of Structural Engineering, ASCE 1993;119(11):3196237. [13] Kim SE, Chen WF. Practical advanced analysis for braced steel frame design and practical advanced analysis of unbraced steel frame design. Journal of Structural Engineering, ASCE 1996;122(11):125974. [14] Chen WF, Kim SE. LRFD steel design using advanced analysis. Boca Raton, FL: CRC Press, 1997. [15] Chen WF, Han DJ. Plasticity for structural engineers. New York: SpringerVerlag, 1988. [16] Chen WF, Lui EM. Stability design of steel frames. Boca Raton, FL: CRC Press, 1992. [17] White DW, Chen WF, editors. Proceedings of the USJapan Seminar on Innovations in Stability Concepts and Methods for Seismic Design in Structural Steel [special issue]. Engineering Structures 1997;20(46). [18] Chen WF, Sohal I. Plastic design and second-order analysis of steel frames. New York: Springer-Verlag, 1995. [19] AISC. Load of resistance factor design specication for structural steel buildings. Chicago, IL: American Institute of Steel Construction, 1997.

S-ar putea să vă placă și