Sei sulla pagina 1di 45

12

Seismic Design of Steel Structures


12.1 Introduction 12.2 Historic Development and Performance of Steel Structures 12.3 Steel Making and Steel Material
Physical Properties of Structural Steel Mechanical Properties of Structural Steel

12.4 Structural Systems


Braced Frames Design Approach

12.5 Unbraced Frames

Ronald O. Hamburger
Simpson Gumpertz & Heger, Inc. San Francisco, CA

Special Moment-Resisting Frames Intermediate MomentResisting Frames Ordinary Moment Frames Special Truss Moment-Resisting Frames

Niaz A. Nazir
DeSimone Consulting Engineers San Francisco, CA

Dening Terms References Further Reading Appendix A

12.1 Introduction
In many ways structural steel is an ideal material for the design of earthquake-resistant structures. It is strong, light weight, ductile, and tough, capable of dissipating extensive energy through yielding when stressed into the inelastic range. Given the seismic design philosophy of present building codes, which is to rely on the inherent ability of structures to undergo inelastic deformation without failure, these are exactly the properties desired for seismic resistance. In fact, other construction materials rely on these basic properties of steel to assist them in attaining adequate seismic resistance. Modern concrete and masonry structures, for example, attain their ability to behave in a ductile manner through the presence and behavior of steel reinforcing. Timber structures derive their ability to withstand strong ground motion through the ductile behavior of steel connection hardware, including bolts, nails, and various steel straps and assemblies used to interconnect wood framing. Steel is a mixture of iron and carbon, with trace amounts of other elements, including principally manganese, phosphorus, sulfur, and silicon. Steel is differentiated from the earlier cast and wrought irons by the reduced amounts of carbon relative to these other alloys and the reduced amounts of other trace elements. These differences make steel both stronger and more ductile than cast and wrought irons, both of which tend to be quite brittle. Although iron alloys have been in use for centuries, steel is a relatively modern material. For practical purposes the advent of steel as a construction material can be traced to

2003 by CRC Press LLC

12-2

Earthquake Engineering Handbook

the mid-19th century, when Sir Henry Bessemer developed the iron-to-steel conversion process that allowed production of steel in large quantities. Initial uses of steel were in the railroad industry, where it was used extensively to produce rails, and for armaments, including rie and gun barrels. Andrew Carnegie imported the Bessemer process to the United States and constructed his rst steel mill in 1870, initially for rail and machinery production. By the 1890s, however, steel was being applied to building construction and, with the advent of the elevator and high-rise construction, rapidly became the building material of choice for the new generation of tall buildings. The same properties that make it a desirable material for high-rise construction (light weight, strength, ease of fabrication and erection) also make it a popular construction material for structures involving long, clear spans. Today it is used in a variety of construction applications ranging from bridges to industrial plants to buildings. Throughout the relatively brief history of their use, structural steel buildings have been among the best performing structural systems and, prior to January 1994, when previously unanticipated connection failures were discovered in some buildings following the Northridge earthquake (M 6.7), many engineers mistakenly regarded such structures as nearly earthquake-proof. A year later, the Kobe earthquake (M 6.9) caused collapse of 50 steel buildings, conrming the potential vulnerability of these structures. This experience notwithstanding, structural steel buildings, if properly designed, can provide outstanding earthquake performance. To assure good behavior of steel structures, it is necessary to: Congure the structural steel system so that inelastic behavior is well distributed throughout the structure, rather than concentrated in a few stories or elements Provide columns with sufcient strength to resist earthquake-induced overturning loads without buckling Provide adequate lateral bracing for exural members to prevent lateral-torsional buckling Proportion connections with sufcient strength that inelastic behavior occurs in the members themselves Select compact sections for those members intended to experience inelastic behavior, to avoid local buckling and the rapid loss of strength that accompanies such behavior In addition, as with all structural materials, it is very important to assure that the structures are actually constructed as designed, that quality is maintained in fabrication and eld welding operations, and that the structure is maintained over its life. This chapter discusses the historic performance of steel structures in earthquakes, the basic manufacturing processes and properties of structural steel, the basic structural systems used in steel structure design, and the current code requirements for the design of steel structures.

12.2 Historic Development and Performance of Steel Structures


Prior to the late 1800s, the most common building materials were either timber or masonry. Timber structures were limited in height by the strength of the material and seldom exceeded three or four stories. Masonry buildings could be constructed taller than this; however, it was necessary to make the loadbearing walls quite thick, 30 in. or more, in buildings of six stories or taller. Until the advent of the elevator, these were not limiting factors on construction, as it was impractical to inhabit structures taller than four or ve stories. With the elevator, however, it became practical to construct buildings that were ten or more stories in height. The elevator, together with structural steel being on the order of ten times stronger than masonry, enabled such construction to occur. The Home Insurance building, a 9-story structure erected in Chicago in 1885 and later expanded to 11 stories in 1891, is generally credited with being the rst skyscraper [Chicago Public Library, 1997]. Throughout the 1890s and early 1900s, major cities in industrialized nations around the world began to build tall steel structures. The use of structural steel as a building construction material also found rapid application in long-span industrial structures, where the high strength and light weight of the material found practical application in the construction

2003 by CRC Press LLC

Seismic Design of Steel Structures

12-3

FIGURE 12.1

Large steel spans (left) detail St. Pancras Station, London; (right) Hamburg Hauptbahnof.

FIGURE 12.2

Typical built-up member in early steel construction.

of trusses to span over large manufacturing operations. Throughout the early 20th century, these two applications, high-rise and industrial construction, were the primary applications of steel construction in the building industry. Later in the 20th century, as labor became more expensive in industrialized countries, steel began to nd application in low- and mid-rise construction as well, replacing the more labor-intensive concrete and masonry construction practices. As compared with other construction materials, steel construction requires a signicant industrial infrastructure and a skilled labor force. Therefore, even today, it is commonly used as a building material only in industrialized nations (Figure 12.1). Steel building construction is typically of two basic types: braced frames or unbraced frames (also commonly called moment-resisting frames), or a combination of these types. Prior to the 1920s, members of steel frames commonly were constructed as complex built-up members with gusset plates and built-up connections, as illustrated in Figure 12.2. The members and connections were riveted, and the entire steel frame was normally encased in masonry or concrete for re protection, and also to provide walls and partitions for the structure. Few, if any, of these steel structures were designed for seismic loading, since only wind load was considered prior to about 1930 (see Chapter 11). These buildings invariably included many stiff and strong unreinforced masonry walls and partitions. Structural engineers relied upon these walls and partitions to help resist lateral loads, but they performed no calculations of the stiffness and resistance provided by these walls [Roeder, 2000].
2003 by CRC Press LLC

12-4

Earthquake Engineering Handbook

FIGURE 12.3

Typical semirigid beam column connections with rolled structural shape.

Changes in steel frame construction began to evolve around 1920. Labor costs for the built-up elements were increasing at this time, and standard hot rolled shapes for beams and columns became the normal practice. The AISC Specication and Manual [AISC, 1928] was rst developed in this period. These rolled shapes commonly were connected with riveted angles and T-sections as illustrated in Figure 12.3, and members and the connections were still encased in concrete for re protection. These framing practices became standard and were designed with relatively simple calculations for the next 20 to 30 years. Unreinforced masonry curtain walls and partitions were still used, and the combined effect of the added strength and stiffness provided by these walls and the composite action due to the encasement supplied a major portion of the structural stiffness and resistance to lateral loads. These early structures tended to be highly redundant in that every beamcolumn connection was a semirigid moment-resistant connection, with a large but uncalculated stiffness, and signicant uncalculated resistance was also provided by nonstructural elements such as architectural masonry walls and the concrete encasement for re protection. This construction was commonly used until the mid-1950s or early 1960s. After about 1960, high strength bolts began to replace the rivets in the T-stub and double-ange-angle connection, but the connection details and geometry remained essentially the same as those used for riveted construction. Concrete encasement was also discontinued in favor of lighter re protection materials. However, buildings of the 1950s and 1960s still had a substantial uncalculated strength and stiffness due to cladding and partitions, and they were very redundant, since the moment-resisting connections were used at every beam-to-column joint. This construction continued into the early 1970s, at which time eld welding began to replace eld bolting, particularly for application to moment-resisting connections and, at about the same point in time, single plate shear connections began to replace double angles and tees for non-moment-resisting connections. The period 1970 to 1994 is characterized by a gradual trend of building frames with reduced redundancy, in which fewer but larger framing elements were used to provide lateral resistance for structures. Moment-resisting connections were typically fabricated using a standard welded ange, bolted web connection (Figure 12.4). Failure of a number of these connections in the 1994 Northridge earthquake resulted in revised practice, discussed in later sections. Given the relatively short history of steel building construction and the fact that this construction material was widely used only in industrialized nations, the earthquake performance history for steel structures is rather limited. It begins with the 1906 San Francisco earthquake (M 7.9) and re. At the time of the FIGURE 12.4 Typical welded ange, bolted web moment-resisting connecSan Francisco earthquake, construction in the city predomition commonly in use 1970 to 1994. nantly consisted of low-rise timber frame and masonry bearing
2003 by CRC Press LLC

Seismic Design of Steel Structures

12-5

FIGURE 12.5 View of downtown San Francisco following the 1906 earthquake and re, and showing the survival of a number of taller steel frame buildings.

FIGURE 12.6 Tall building (left) on re, 1906 earthquake; (right) in 1999, with newer cladding. (Photos: (left) anonymous; (right) C. Scawthorn)

wall buildings and taller steel frame buildings with unreinforced masonry inll walls. Engineers surveying the damage following the earthquake and re remarked that damage to the taller steel frame buildings was much lighter than for other structures [USGS, 1907] (Figure 12.5), and, in fact, a number of these buildings are still in service today (Figures 12.6 and 12.7). The superior performance of steel buildings observed in this earthquake initiated the perception, commonly held by some engineers until the mid-1990s, that steel frame buildings were practically invulnerable to earthquake-induced structural damage. Reis and Bonowitz [2000] postulate that this perception was more a result of a lack of exposure of steel frame buildings to strong motion, rather than a signicant body of data of good performance. However, it is probable that this perception evolved, at least in part, from the fact that the early steel frame structures, with extensive inll masonry, did perform remarkably better than contemporary structures of pure masonry or reinforced concrete construction. Hadley (n.d.), for example, reports that at the time of the great Kanto, Japan earthquake of 1923, steel
2003 by CRC Press LLC

12-6

Earthquake Engineering Handbook

FIGURE 12.7 Market Street at Third Street in San Francisco, April 1906. The tall steel frame building to the right is still in service today. (Source: anonymous)

building construction had been in use only for about 5 years in Japan. Four large steel buildings had been completed and two were almost complete. These buildings, with extensive masonry inll walls and partitions, suffered little to no damage of the frames, though extensive damage to the masonry inll is reported. Hundreds of collapses of masonry structures were reported. Similarly, reports of the 1925 Santa Barbara earthquake [CISC, n.d.] indicate that although 17 concrete and masonry buildings were destroyed or eventually demolished, 2 steel frame buildings with masonry inll, located closer to the epicenter, were barely damaged. The rst earthquake to affect a large number of modern steel buildings was probably the 1971 San Fernando (M 6.6), California earthquake. Steinbrugge et al. [1971] reported on a study of 30 completed steel buildings in the greater Los Angeles area, noting some damage to stairs, concrete walls, and nonstructural elements, but no structural damage to the completed steel frames. However, two noteworthy steel buildings, the two 52-story ARCO Towers then under construction in downtown Los Angeles, did experience damage to their structural frames. These buildings probably saw ground motion with a peak horizontal acceleration on the order of 0.15 g. Damage consisted of cracking of welded beam to column connections and of welded connections in transfer trusses. These cracks were ascribed to poor weld quality, and as the buildings were then under construction, were repaired as part of the ongoing construction work. Osteraas and Krawinkler [1989] note that the 1985 Mexico City earthquake (M 8.1) was probably the rst event in which a signicant number of steel buildings, including modern ones, were subjected to a severe test. They report data on 79 steel structures, including 41 moment-resisting frames, 17 braced frame structures and 21 structures with concrete shear walls. Of these, 12 buildings were reported as having moderate to severe damage, including 2 buildings in the Pio Suarez complex that were total collapses (Figure 12.8). Pio Suarez was a ve-building complex constructed over a subway station. A 21-story, braced frame structure collapsed onto an adjacent 14-story structure, also causing its collapse. Osteraas and Krawinkler ascribe this collapse to overstrength in the steel braces of the structure, which delivered greater overturning forces to the buildings built-up columns then they could withstand, inducing local, and then complete, buckling failure of the columns and building collapse. Study of this failure led to introduction of requirements in the building codes that columns in steel structures be designed considering the potential overstrength of the supported structure.
2003 by CRC Press LLC

Seismic Design of Steel Structures

12-7

FIGURE 12.8 Pino Suarez collapse, 1985 Mexico City earthquake. (Courtesy National Oceanographic and Atmospheric Agency) Shown as Color Figure 12.8.

Numerous other reports of damage to braced frame structures in various earthquakes may be found in the literature. Typical damage includes buckling of compression braces, fracture of braces at weak net sections and in the vicinity of local buckling of thin-walled brace sections, and failure of bolted end connections of braces. Figure 12.9, for example, shows a buckled brace in the three-story California Federal Savings Bank data center in Rosemead, California following the 1987 Whittier Narrows (M 5.9) earthquake. Such damage has typically been readily repairable. Extensive damage to welded moment-resisting connections was reported in buildings following the 1994 Northridge California (M 6.7) earthquake [Youssef et al., 1995]. The damage typically consisted of fractures, initiating at the root of complete penetration welded joints between beam bottom anges and columns. Once initiated, these fractures would extend in a variety of paths and, in some cases, extended nearly completely across columns (Figure 12.10). Although only one structure, a two-story building owned by the California State Automobile Association, was damaged so severely that it was deemed irreparable, the widespread occurrence of this unanticipated damage caused great concern in the design community. Responding to this concern, the Federal Emergency Management Agency (FEMA) sponsored a 6year, $12 million research program known as the SAC Steel Program, to determine the cause of the damage and recommend design and construction procedures to mitigate the problems identied. The FEMA/SAC program concluded that the damage was a result of large stress and strain concentrations induced by the typical connection conguration, the frequent presence of large defects and aws in the welded joints, and the common use of low toughness weld metals. An extensive series of design [SAC, 2000a], construction, and quality assurance [SAC, 2000b] recommendations were published by the FEMA/SAC project and are slowly being incorporated into standard design specications. The project also developed procedures for seismic performance evaluation of steel structures [SAC, 2000c], postearthquake damage assessment criteria [SAC, 2000d], and an extensive collection of background technical report documents. On January 17, 1995, exactly 1 year after the Northridge earthquake, the Hyogo Ken Nambu earthquake (M 6.9) struck the city of Kobe, Japan and surrounding areas. This earthquake also caused extensive brittle fracture damage to steel structures. Nakashima [2000] reports 988 damaged steel buildings, including 432 moment-resisting frames, 168 buildings with braced frames, and 388 with unidentied framing systems. Figure 12.12 presents a breakdown of the severity of damage, by building height. More than 50
2003 by CRC Press LLC

12-8

Earthquake Engineering Handbook

FIGURE 12.9

Buckled brace in California Federal Savings building, Rosemead, California.

FIGURE 12.10

Fracture through a column ange and web at a moment-resisting connection.

steel buildings collapsed in the Kobe earthquake, but none in excess of seven stories in height. Most of the collapsed buildings were older structures, employing tubular steel columns. Many of these buildings were very slender and experienced brittle fractures at column splices, resulting in overturning failures. As a result of the number of collapsed and severely damaged buildings, a study similar to that conducted in the United States following the Northridge earthquake was conducted by Japanese investigators. This study concluded that the failures were largely due to older construction practices and did not result in recommendations for major changes to design or construction practice.

12.3 Steel Making and Steel Material


The behavior of steel structures in earthquakes is dependent on key mechanical properties of the steel material, including its strength, ductility, and toughness. These properties, in turn, are dependent on
2003 by CRC Press LLC

Seismic Design of Steel Structures

12-9

FIGURE 12.11 Brittle fractures in massive braced frames of Ashiyahama Complex. (Photo: C. Scawthorn)

FIGURE 12.12 Distribution of steel building damage in the Kobe earthquake by severity and building height. (From Nakashima, M., 2000. Appendix C, State of the Art Report on Past Performance of Steel Buildings in Earthquakes, FEMA-355E, Federal Emergency Management Agency, Washington, D.C.)

the processes used to produce the material. Structural steel is a mixture of iron and carbon with varying amounts of other elements primarily manganese, phosphorus, sulfur, and silicon. These and other elements are either unavoidably present or intentionally added in various combinations to achieve specic characteristics and properties of the nished steel product. Table 12.1, excerpted from Frank et al. [2000], lists the primary elements found in structural steel and their effects on steel properties. Various steel-making furnaces have been developed over the years. The modern age of bulk production of steel was initiated with the Bessemer Converter. This was later replaced by the open hearth furnace and, more recently, the basic oxygen and electric arc furnaces. Steel making begins with a source of raw iron, either in the form of iron ore, reduced in a blast furnace, or scrap metal. The blast furnace reduces iron ore and other iron-bearing materials, coke, and limestone into pig iron. Coke is a carbon-rich material obtained by baking coal in an oxygen-free environment. The limestone acts as a cleaning agent by reacting with impurities in the ore. The iron ore and other iron-bearing materials, coke, and limestone proceed slowly down through the body of the furnace as they are exposed to a blast of hot air that burns the coke, releasing heat and gas, and reduces the iron ore to metallic iron. This metallic iron contains several chemical elements, including carbon, manganese, sulfur, phosphorus, and silicon in amounts higher than permitted for steel. Thus, it is drawn off periodically, to be rened further in a steel-making furnace.
2003 by CRC Press LLC

12-10

Earthquake Engineering Handbook

TABLE 12.1 Principle Elemental Components of Structural Steel


Carbon Principal hardening element in steel, increases strength and hardness, decreases ductility, toughness, and weldability Moderate tendency to segregate Increases strength and toughness Controls negative effects of sulfur Increases strength and hardness, decreases ductility and toughness Considered as an impurity, but sometimes added for atmospheric corrosion resistance Strong tendency to segregate Considered undesirable except for machineability. Decreases ductility, toughness, and weldability Adversely affects surface quality Strong tendency to segregate Used to deoxidize or kill molten steel Increases strength Used to deoxidize or kill molten steel Renes grain size, thus increasing strength and toughness Small additions increase strength Renes grain size, thus increasing strength and toughness Small amounts rene the grain size, thus increasing toughness Increases strength and toughness Increases strength Increases atmospheric corrosion resistance Primary contributor to atmospheric corrosion resistance Increases strength Increases strength and hardness May decrease ductility and toughness Small amounts increase hardenability, used only in aluminum-killed steels Most cost effective at low carbon levels

Manganese Phosphorus

Sulfur

Silicon Aluminum Vanadium and Columbium Titanium Nickel Chromium Copper Nitrogen Boron

The most common steel-making furnaces today are the basic oxygen furnace and electric arc furnace (EAF). In either furnace, the metallic iron from a blast furnace or scrap iron is charged into the furnace together with limestone and melted by gas jets, electric arcs, and oxygen lances. Fluxes are added to reduce sulfur and phosphorus contents to desired levels. As the melt progresses and a liquid pool can be contacted, the lanced oxygen burns dissolved oxidizable elements, such as carbon, manganese, silicon, and aluminum contained in the liquid; the energy from this reaction elevates the temperature of the liquid metal pool. In the nal stages of melting, the oxygen is used to decarburize the melt. Sacricial carbon is also commonly blown into the covering slag layer to react with excess oxygen. This reaction liberates additional energy. Working of the steel continues until the desired tap temperature and carbon level have been obtained. When this has occurred, the heat will be tapped into a refractory-lined ladle. Typical EAF heats range from 80 to 360 tons. Once the liquid steel has been processed to achieve the desired chemistry and temperature, it must be put into a solid form suitable for use by the rolling mill. The process of producing this solid product is known as casting. In traditional (historic) steel making, the liquid steel was poured from the ladle into a series of cast iron molds, cooled, and solidied into an ingot. Most modern steel production uses the continuous casting method. All structural shapes of domestic origin and the majority of foreign-produced shapes are continuously cast. In this process molten steel is continuously poured into a mold. The molds are made of copper, formed in the cross-sectional shape and size of the desired casting, and water-cooled. During steady-state casting, the steel streams into the open top of the mold and lls the mold cross section. Liquid steel that comes in direct contact with the water-cooled mold surface quenches to form a solid shell and joins to the existing shell already formed along the perimeter of the mold. As the shell forms, it is continually withdrawn from the bottom of the mold. During the short residence time within the mold, the thermal transfer is sufcient such that the shell grows to a thickness that is capable of maintaining its cross-sectional shape while containing a core of liquid steel. Outside of the mold, water and/or air sprays are employed to continue shell thickening. Mechanical restraint may also be used to help maintain the cast shape. Continuously cast shapes include billets, blooms, slabs, beam blanks, and near net shapes.
2003 by CRC Press LLC

Seismic Design of Steel Structures

12-11

Lamellar fracture

Plate subjected to thru-thickness tension strain

Welded joint

FIGURE 12.13 Schematic of lamellar tear-type fracture in heavy weldment.

Most structural shape and plate is produced from castings by the hot rolling process. The process consists of passing the hot cast material between a set of rolls revolving in opposite directions, and spaced such that the distance between the rolls is less than the thickness of the material entering the rolls. The rolls grip the piece, reducing its cross-sectional area and increasing its length. This process forms the steel into the desired cross-sectional shape, while improving the mechanical properties by modifying the original cast structure. Hot rolling causes grain renement and elongation of those grains, deformable inclusions, and inhomogeneities in the rolling direction. The preferential alignment of structure along the rolling direction results in a shape with anisotropic properties. This is particularly true for ductility and fracture toughness. Hot rolling also tends to elongate segregated elements, such as sulfur, into at discontinuities. When rolled steel is subjected to through-thickness tensile stresses, these discontinuities can produce planes of weakness that can later result in a form of failure known as lamellar tearing. Lamellar tearing is characterized by a step-like fracture surface, generally running parallel to the rolling direction (Figure 12.13). Lamellar tearing has occasionally occurred in thick materials under highly constrained conditions, such as in certain welded joint details. The tearing usually occurs during fabrication as a result of thru-thickness shrinkage strains and tensile stresses that occur as welded joints cool. Nondestructive testing procedures, including ultrasonic testing (UT) and radiographic testing (RT), can be used to detect zones of lamellar tearing. Modern steel produced by the EAF and continuous casting processes has reduced levels of sulfur relative to historic steels, and such steel is thought to be less susceptible to lamellar tearing.

12.3.1 Physical Properties of Structural Steel


On the microstructural level, all metals are composed of grains. Grains are a three-dimensional matrix of atoms arranged in a regular and repeating crystal structure. The characteristics and properties of steels are a function of the microstructure and grain distribution. Microstructure, in turn, is determined by the chemistry, deformation, and thermal history of the steel. The microstructure of most common structural steels consists of a primary matrix of ferrite grains with a small dispersion of pearlite. The characteristics of the ferritic grain structure of steel dictate the properties and behavior at normal service temperatures. A ne grain size promotes increased strength, toughness, and weldability. The past thermal history of steel has signicant inuence upon properties of steel products. The principal thermal history effects are due to phase transformations and grain growth events. Steel is an unusual material in that, as its temperature decreases from the liquid state to ambient, it not only

2003 by CRC Press LLC

12-12

Earthquake Engineering Handbook

undergoes a liquid to solid-state change (freezing), but also two separate and distinct solid-state phase transformations. Very simplistically, solid steel grains are composed of a three-dimensional crystal matrix of regularly arranged iron atoms. The atomic diameter of carbon is roughly half the size of iron. Thus, carbon atoms easily t into the interstices (spaces) between the iron atoms. The packing arrangement, and hence the interstitial hole size and distribution, is different in the different solid-state phases. Structural steel grades, upon solidication, form a solid phase known as delta-iron (-iron). This phase exists only at highly elevated temperatures. Upon further cooling, the atomic arrangement of iron atoms in -iron transforms to a different packing conguration known as austenite or gamma-iron (-iron). Atomic packing density of iron in the austenitic state is such that up to 2% of carbon can be dissolved into the iron matrix. Further cooling of austenite will induce the iron matrix to transform to a higher packing density known as ferrite or alpha-iron (-iron). The volume of interstices in the new matrix is reduced, resulting in a maximum carbon solubility of 0.02%. The transformation from austenite to ferrite occurs over a temperature range that is dependent on chemical composition. Under equilibrium conditions, as the temperature is decreased through the transformation range, the excess carbon that is rejected by the formation of ferrite diffuses through the solid steel, concentrates, and forms iron carbide. Iron carbide forms in islands of alternating ferrite and iron carbide, known as pearlite. The total percentage of pearlite developed within steel depends on the carbon content. The pearlite lath spacing is a function of temperature and time of formation. The size of the pearlite islands and the spacing between laths strongly inuence the hardness, ductility, and strength of the steel. Examples of structural steels having ferrite-pearlite microstructure are ASTM A36, A572, and A992, all of which are deemed to have suitable toughness, ductility, and weldability for use in seismic force-resisting systems. The solid-state diffusion (transport) of carbon atoms through the solid steel matrix is dependent on both time and temperature. If the temperature of the steel is lowered rapidly (quenched) through the transformation range such that sufcient time for carbon diffusion is not provided, metastable lowtemperature transformation products bainite or martensite will form. These phases are characterized as being harder, stronger, and less ductile than ferrite pearlite steels and not desirable components of seismic force-resisting systems. Martensite and bainite can form in structural steels if rapidly heated by welding, ame or arc cutting unless proper preheating is performed to control the cooling rate and avoid quenching. If after quenching, the temperature of the steel is again raised, sufcient thermal energy will be restored to the system for solid-state carbon diffusion to reinitiate. Ductility and toughness of the steel will be improved, but at the expense of the strength and hardness that bainite and martensite offer. By carefully controlling the temperature and time of reheating, the amount of decomposition can be controlled and thus a balance between increased strength and hardness can be obtained, with acceptable toughness and ductility. This process is known as tempering. ASTM A913 is an example of a structural steel material that is processed through a quenching and tempering process and has excellent properties of toughness and ductility. It is also suitable for use in seismic force-resisting systems.

12.3.2 Mechanical Properties of Structural Steel


The primary properties of structural steel that are important to seismic performance are Yield strength Tensile strength Ductility Fracture toughness

Each of these depends on the metallurgy and thermomechanical processing history, as discussed in previous sections, as well as the load application rate, temperature, and conditions of restraint at the time of load application.

2003 by CRC Press LLC

Seismic Design of Steel Structures

12-13

YIELD POINT Fy STRESS f (Ksi)

TENSILE STRENGTH

y 0 0.004

= 0.005

s 0.012 STRAIN (in./in.) 0.016 0.020 0.024

0.008

FIGURE 12.14

Typical stress-strain curve for structural steel.

12.3.2.1 Tensile Properties Most mechanical properties important for design are determined from a standard tension test. In this test, a machined specimen, with standard cross section, is loaded in a universal testing machine while loadelongation data are recorded. These data are reduced into the form of a stress-strain curve (Figure 12.14). The initial straight line segment of the stress-strain curve represents the specimens elastic behavior where stress is proportional to strain and related by the Youngs Modulus, which has a value of 29,000 ksi for steel. As strain increases, stress and strain become nonlinear and the specimen experiences permanent plastic deformation. Many mild carbon steels exhibit a peak stress immediately after the stress-strain curve deviates from linearity, known as the yield point. Immediately after achieving the yield point, the stress dips with increasing strain, then remains at a constant value, known as the yield strength, for considerable amounts of additional strain. Thereafter, the steel strain hardens with increasing stress, until a peak, or ultimate, tensile strength is reached. With increasing strain beyond the tensile strength, the material exhibits necking and, ultimately, fracture. Standard ASTM material specications include controls on the yield strength or yield point of the material as well as the tensile strength and elongation of the material at fracture. Although yield point is of no engineering signicance, ASTM specications permit mills to report either yield point or yield strength. Therefore, it is possible that some material conforming to the ASTM material specications will have slightly lower yield stress than the nominal value contained in the specication. More typically, due to variations in the production process, yield and tensile properties will substantially exceed the minimum specied values, sometimes by as much as 40% or more. In general, tensile properties of steel vary with temperature. Tensile data for various steels show that their yield strength and ultimate strength increase by approximately 60 ksi when the temperature decreases from 70 to 320F [Barsom, 1991]. Similarly, when steel is elevated to about 900F it loses about half of its room temperature strength and modulus of elasticity. However, for the range of temperatures of interest for most structures (60F < T < 120F), stress-strain properties of steel may be considered to be essentially constant. Tensile properties also vary with rate of loading. Tensile data for various steels subjected to monotonic dynamic loading show that the yield strength increases by about 4 to 5 ksi for each order of magnitude increase in rate of loading. The difference in yield strength under static loading as opposed to full impact loading (time to fracture <0.001 sec) is about 25 ksi. Tensile tests performed by mills to demonstrate compliance with ASTM material specications are performed in accordance with ASTM standard A370. This standard permits tensile tests to be performed at rates between 10 and 100 ksi/min. Typical response

2003 by CRC Press LLC

12-14

Earthquake Engineering Handbook

FIGURE 12.15

Three principal stress components and planes of maximum shear stress.

of structures to seismic loading has rise times to peak loading on the order of one to a few seconds, which is approximately one order of magnitude larger than the test rate. Thus, the effective yield strength of structural steel when responding to earthquake motion may be 4 to 5 ksi larger than that reported by the mills. With the possible exception of some braces in braced frames, structural elements are seldom loaded in a uniaxial state of stress. The stress eld for a typical material element can be described by three principal stresses that are aligned normal to each other on three orthogonal planes (Figure 12.15). Shearing stresses inclined with regard to these planes can be calculated from the principal stress components. Assuming that 1 is the largest of the principal stress components and 3 the smallest, the maximum shear stress component, max , is given by the equation: max =

(1 3 )
2

(12.1)

For a uniaxial tension test specimen, the applied tension produces tensile stress 1, while 2 and 3 are both null. Therefore, in such a specimen, the maximum shear stress, max has a value equal to half the applied tensile stress 1. Since plastic deformation and yielding occur when max reaches a critical value, in elements that are subjected to multiaxial loading, which is the more typical situation, the critical value of max may be achieved at maximum principal tensile stress values that are either lower or higher than attained in the uniaxial tensile test. For example, if one principal stress plane is in compression while a second is in tension, 3 in Equation 12.1 will have a negative value, and a critical value of max will be attained at a lower value of applied tension 1 than in the uniaxial case, lowering the effective yield strength. More commonly, if the material is constrained, as often occurs in connections, the material can develop a state of triaxial tensile stress. Under such conditions, max will approach 0 regardless of the applied tensile stress and yielding may never occur. Instead, the material will behave in an elastic manner until ultimate tensile stress is reached, at which point it will fracture. This is one reason why structures designed for seismic resistance should be proportioned such that yielding occurs in members, such as beams, columns, braces, etc., rather than at the connections of these members. 12.3.2.2 Ductility Material ductility is an important index of the ability of a material to withstand inelastic deformation without fracture and is a key property important to seismic performance of structures. Material ductility, M , is usually represented by the equation: M = u y (12.2)

2003 by CRC Press LLC

Seismic Design of Steel Structures

12-15

Mo

Undeformed shape

Deformed shape
FIGURE 12.16 Deformation of beam under applied external moment.

where u and y , respectively, are the material strains at fracture and onset of yield. At room temperature, typical structural steels will have static ductility on the order of 20. Under cyclic applied loads, the availability material ductility decreases as the material is subject to fatigue. Generally, the more cycles of strain the material must undergo, the lower available material ductility. Typical structural steels may be regarded as having available ductility for seismic purposes on the order of about 10. Ductility can also be expressed in terms of the inelastic deformation capacity of individual members. In the case of an axially loaded tension member, the member ductility, , would be given by the equation: = u y (12.3)

where u is the total brace elongation at failure and y is the total brace elongation at yield. Since axially loaded members have uniform stress and strain distribution across their cross sections and along their length, the ultimate elongation of such a brace would be given by the term: u = u L where u is ultimate material strain, previously discussed, and L is the member length. Similarly, the member yield deformation is given by: y = yL (12.5) (12.4)

where y is the material yield strain. By substitution of the relationships of Equations 12.4 and 12.5 into Equation 12.3, it is seen that for axially loaded tensile members, member ductility and material ductility are the same. The same cannot be said for member ductility of exural members. Consider the beam shown in Figure 12.16. This beam has uniform cross section, is pinned at both ends, is subjected to an applied moment at one end, Mo, and as a result of this, experiences an angular deformation at that end, . We will consider the behavior of this beam as the moment and rotation applied at the end are progressively increased in magnitude until failure occurs. Under initial, low-level loading, the beam will behave elastically. The moment M and curvature at any station, a, along the beam will be given by the equations: M (a ) = M o and (a) = M (a ) EI Mo a L (12.6)

(12.7)

2003 by CRC Press LLC

12-16

Earthquake Engineering Handbook

y+dy

Strain

y Fully Plastic Stress Distribution

y+dy Strain Hardening Stress Distribution

Partially Plastic Elastic Stress Stress Distribution Distribution

FIGURE 12.17

Strain and stress distribution in symmetric cross section at various levels of plasticity.

where the curvature, , is given by the equation: = T + C d (12.8)

and T and C are, respectively, the extreme tensile and compressive strains on the cross section at station a and d is the cross section depth. The rotation at the end of the beam is obtained by integrating the curvature along the beam and is given by the equation: = ML 3EI (12.9)

Once the strain at the outermost beam bers reach the yield strain, however, these relationships no longer apply. Figure 12.17 shows the strain distribution across the beam cross section, which in this case is assumed to be symmetric, and the corresponding stress distribution in the beam as the curvature progresses from elastic to partially plastic to fully plastic behaviors. As the applied rotation at the end of the member increases beyond the elastic range, plasticity spreads from the outer beam bers toward the center of the section until ful plasticity is achieved. Then, as further rotation is applied, the strains at outer bers begin to enter the strain hardening range and the stresses slowly increase beyond the yield level until ultimate strain levels are achieved and the section fractures. For most beam sections, strain hardening will actually begin to occur at outer beam bers before full plasticity is achieved across the cross section. A plot of applied moment at the beam end against applied curvature would typically take the form of Figure 12.18. As can be seen, as the cross section is deformed into increased levels of plasticity, the momentcurvature relationship is quite nonlinear. Eventually, as the section becomes fully plastic, the behavior again becomes approximately linear with a small postyield modulus, typically on the order of 2 to 3%. The rounded form of this curve, as the section plasties, makes denition of a specic yield point difcult. Some engineers dene the point of yield as that point at which the extreme beam ber rst yields. However, more typically, an inferred yield point is taken as the intersection of the extended elastic behavior and strain hardening behavior lines, with the yield curvature and yield rotation taken at this point. As indicated by Equations 12.6 and 12.7, when the beam is elastic, the distribution of curvature along the length of the beam is linear. As beam behavior becomes inelastic, the distribution of curvature along the beam also becomes nonlinear. Due to the effects of strain hardening, plasticity tends to distribute out along a nite length of the beam, lp , termed the plastic hinge zone (Figure 12.19), where a nearly
2003 by CRC Press LLC

Seismic Design of Steel Structures

12-17

Moment Mo

Inferred yield point

actual behavior

Curvature
FIGURE 12.18 Inelastic moment-curvature behavior of beam.

Mp

Plastic hinge zone

lp L
FIGURE 12.19 Idealized moment-curvature distribution for inelastic beam.

constant moment Mp occurs. Mp is commonly termed the plastic moment, and has a lower bound value, neglecting strain hardening effects, of ZFy , where Z is the plastic section modulus. For structural members, the plastic hinge length, lp , is typically about three quarters of the beam depth, d, though this varies depending on the length of the beam and the moment gradient. Neglecting the small moment gradient within the plastic hinge zone, the end rotation of the beam and the curvature of the beam within the plastic hinge zone can be related by the expression: = Mp 3L

(L l )
p

+ p

lp l L p 2 L

(12.10)

As can be seen, for exural members, the relationship between plastic curvature, p , and end rotation, , is not directly proportional and an available material ductility of 10 does not imply an available beam ductility of the same amount. This becomes even more complex when the global ductility of an entire structural system is considered. Global system ductility can be dened by the expression: = u y (12.11)

2003 by CRC Press LLC

12-18

Earthquake Engineering Handbook

7000

Total Lateral Base Shear Force - kips

6000
Vy

5000

4000

3000 Du=39.5 2000 Dy = 8.5 Dmax = 31.8

1000
Dy

0 0 5 10 15 20 25 30 35 40

Displacement (in.)

FIGURE 12.20

Pushover curve for an arbitrary structure.

where u and y are, respectively, the lateral deection of the structure at failure and onset of yielding. As with individual beams, determination of the point of yielding in a structure is somewhat difcult, as is selection of a point of ultimate strength. Figure 12.20 presents a pushover curve for an arbitrary steel structure, consisting of a plot of the applied lateral force on the structure against the resulting lateral displacement of the structure. A number of behaviors are discernible on this plot, including the Yielding of individual beams, initiating at a displacement of about 5 in. Failure of individual connections, initiating at a lateral deection of about 30 in. Eventually, at a lateral displacement of about 40 in., total loss of stability and collapse Typically the yield point for such a structure is taken, not at the point that the rst beam yields, but rather at an approximate point taken as the intersection of the lines representing elastic and stable postelastic behaviors, in this case at a lateral deection of about 8.5 in. Using this somewhat arbitrary denition of yield, for this structure, individual failures begin to occur at a deection of 31.8 in., or a global ductility of about 3.75. Global failure occurs at a deection of 39.5 in., or a global ductility of about 4.6. Typical seismic design codes anticipate structures will have global ductility capacities on the order of 3 to 4. As discussed above, it is important to recognize that global ductility is not synonymous with either material or member ductility. Most structures are proportioned to exhibit yielding behavior in relatively limited zones, for example, over the length of a brace, or in the end plastic hinge region of a exural member. If structural yielding is not well distributed throughout the structure, approaching a state of uniform strain, in order to achieve global ductilities on the order of 3 to 4, it may be necessary to achieve material ductilities that are several times larger, approaching or even exceeding the capacity of the material. This is why it is important to congure steel structures to spread inelastic behavior throughout the structure. 12.3.2.3 Toughness Structural steels can fracture in either a ductile or a brittle manner. Brittle fracture is undesirable as it occurs at stresses and strains that are signicantly less than the nominal ultimate values, and therefore has the effect of sharply reducing the availability of ductility. The fracture mode is governed by the metallurgical properties of the material, temperature at fracture, the rate at which loads are applied and the magnitude of the restraints that prevent plastic deformation. The effects of these parameters on the mode of fracture are reected in the fracture-toughness behavior of the material. In general, fracture toughness increases with increasing temperature, decreasing load rate, and decreasing restraint.
2003 by CRC Press LLC

Seismic Design of Steel Structures

12-19

The Charpy V-notch impact test has been the most widely used for characterizing fracture-toughness behavior of steels. This test consists of striking a standard notched bar specimen with a standard hammer, imparting energy to the specimen. These specimens may be tested at different temperatures, and the impact fracture toughness at each test temperature may be determined from the energy absorbed during fracture, the percent shear (brous) fracture on the fracture surface, or the change in the width of the specimen (lateral expansion). At low temperatures, structural steels exhibit a low value of absorbed energy (about 5 ft-lb), and zero brous fracture and lateral expansion. The values of these fracture-toughness parameters increase as the test temperature increases until the specimens exhibit 100% brous fracture and reach a constant value of absorbed energy and of lateral expansion. This transition from brittle to ductile fracture behavior occurs at different temperatures for different steels, for individual heats of steel within a given specication, and even for different locations within a rolled structural section. Fracture toughness is an important parameter for elements that have notches, abrupt geometric discontinuities, or other defects and aws. If the fracture toughness of a material is known, fracture mechanics concepts can be used to determine, based on the geometry of the discontinuity or notch and the fracture toughness of the material, the critical applied stress at which brittle fracture will initiate. A complete discussion of fracture mechanics concepts is beyond the scope of this chapter, but is presented elsewhere [e.g., Barsom and Rolfe, 1999]. In general, members with defects at their outside face are more susceptible to fracture than members without such defects or with defects within the interior of the member. Members under conditions of high restraint are more susceptible to fracture than members not under such restraint. Members subjected to rapidly applied loads are also more susceptible to fracture. When fracture toughness is an important parameter, the design engineer must establish and specify the necessary level of fracture toughness for the material to be used in the particular structure or in a critical component within the structure. Fracture toughness has been found to be important to the performance of exposed structures in cold regions. It has also been found to be an important parameter for welded joints in seismic service, as such joints typically have high restraint and may have large defects and aws which can initiate brittle fracture. Recent design specications identify minimum Charpy toughness values for deposited weld metal in the seismic-force-resisting systems of steel frames. 12.3.2.4 Effect of Cold Working Prior load history can also have signicant impact on a structures strength and ductility. Figure 12.21 shows loading and unloading behavior for a carbon steel tensile specimen. Loading path ABCDE is a schematic illustration of the stress-strain behavior for a specimen loaded monotonically to failure. Loading and unloading the specimen in the inelastic region will alter the stress-strain behavior under future cycles of loading. For example, if the specimen is loaded on path ABCC', the residual stress-strain curve for the element will be given by C'CDE. The resulting specimen will have substantially reduced ductility available. This effect will be even more pronounced if the specimen is loaded along path ABDD''. Upon reloading, the residual stress-strain capacity would be given by the path D'DE, a behavior that exhibits both elevated yield strength and substantially reduced ductility. A specimen that is strained into the strain-hardening region, then unloaded and allowed to age in the unloaded condition for several days at room temperature, or for shorter times at moderately higher temperatures, may follow the reloading curve shown in Figure 12.22. This phenomenon is known as strain aging, and has the effect of further increasing yield strength, increasing tensile strength, and decreasing ductility. Fabrication and production of steel shape and plate frequently involve deforming the material into the strain-hardening range. One common source of such straining is the roller straightening process used by mills to ensure that the straightness of structural shapes meets industry standard tolerances. This can result in local reduced ductility and elevated strength in structural shapes before any fabrication is performed. Bending operations, to form curved or bent elements, produce further strain hardening and cold working of the steel. For these reasons, detailing of steel structure should avoid relying on inelastic behavior in bent or curved structural elements.

2003 by CRC Press LLC

12-20

Earthquake Engineering Handbook

ELASTIC RANGE INCREASE IN YIELD POINT FROM STRAIN HARDENING B STRESS C

INELASTIC RANGE STRAIN-HARDENING RANGE

D E

A C' STRAIN D' DUCTILITY AFTER STRAIN HARDENING RESIDUAL STRAIN DUCTILITY AFTER DEFORMATION WITHIN PLASTIC RANGE DUCTILITY OF VIRGIN MATERIAL

FIGURE 12.21 Strain-hardening behavior of steel under repeat inelastic load applications.

INCREASE IN YIELD POINT FROM STRAIN AGING

INCREASE IN TENSILE STRENGTH FROM STRAIN AGING

STRESS

INCREASE IN YIELD POINT FROM STRAIN HARDENING

STRAIN

DUCTILITY AFTER STRAIN HARDENING AND STRAIN AGING

DUCTILITY OF VIRGIN MATERIAL

FIGURE 12.22 Strain-aging behavior of structural steel.

2003 by CRC Press LLC

Seismic Design of Steel Structures

12-21

12.4 Structural Systems


The lateral-force-resisting systems for structural steel buildings and structures can generally be categorized into one of three broad classes: 1. Braced frames 2. Unbraced or moment-resisting frames 3. Dual systems, composed of a combination of the two systems, designed to act in unison Design and detailing requirements for these various structural systems are governed by the building code and building standards law requirements specic to each country. In the United States, limitations on the applicability of the various structural systems to different structures, minimum loading criteria and limitations on building stiffness are specied by the model building code. A companion chapter in this publication provides discussion of these issues (see Chapter 11). Criteria for proportioning and detailing of steel members and their connections for purposes of seismic resistance are governed by the AISC Seismic Provisions [AISC, 1997] and supplements [AISC, 2000]. The sections below describe the various structural systems commonly employed for seismic resistance of steel structures, the performance issues associated with these structural systems, and brief discussion of the U.S. building code requirements for design of these systems. In U.S. building codes, structural systems are typically categorized based on type, for example, braced frame, moment-frame, etc., and also based on the extent to which the structure is congured and detailed for superior seismic performance. Structures with favorable conguration and superior detailing are classed as special, while structures with limited control on conguration and detailing are classed as ordinary. Structures with intermediate levels of control on conguration and detailing are classed as intermediate. Structures classed as special may be proportioned for reduced lateral design forces, and therefore increased ductility demand, as compared, respectively, to intermediate and ordinary structures. Special structures may be used in any location, regardless of seismicity, or for any occupancy structure, while the use of intermediate and ordinary structures is restricted in application based on structure size, occupancy, and, for some structures, may be used only in regions of low seismicity. In addition to these three categories, U.S. building codes also dene a class of steel structure that is designed for seismic forces, but not provided with any seismic detailing or conguration controls. Such structures are permitted to be used only in zones of low seismicity and must be designed for relatively large design force levels. Readers are referred to the companion chapter on building codes for more information on these topics (see Chapter 11).

12.4.1 Braced Frames


In braced frames, lateral stability of the structure is provided primarily through the presence of diagonal braces within the vertical plane of beamcolumn framing. Braced frames may generally be categorized as concentric braced frames (CBFs), eccentrically braced frames (EBFs), or incomplete braced frames. In a CBF, the frame is congured such that at the connection of the braces to the frame, the work-points for the braces and other elements are coincident, or nearly so. Such CBFs behave as vertical cantilevered trusses and resist lateral forces through the development of axial forces in the various members (Figure 12.23). Other braced frame congurations rely on a combination of vertical truss action plus bending in the beams and/or columns to provide lateral resistance. Figure 12.24 presents schematic elevations of a number of common braced frame congurations, though many variations of these congurations may be found. In general, the single diagonal, X-braced, chevron-braced, V-braced, and zipper-braced frames may be categorized as CBFs.

2003 by CRC Press LLC

12-22

Earthquake Engineering Handbook

V
s co V/

V V(h 4)/D
4

4
V(h4 )/D

s co V/

V(h 4+h3)/D

h3

V(h4+h3)/D

h4

s co V/

V(h4+h3+h2)/D

VH/D
FIGURE 12.23

VH/D

Vertical truss action in braced frame.

X-Braced

Single Diagonal

Chevron Braced

h1

V(H)/D

s co V/

h2

V(h 4+h3+h2)/D

V-Braced

H
Zipper Braced Knee-Braced

K-Braced

Eccentric Braced

Incomplete Braced

FIGURE 12.24

Common braced frame congurations.

12.4.1.1 X-Braced Frames X-braced frames are among the most economical and popular congurations of braced frame structure, particularly in industrial applications, where the architectural impacts of large diagonals across a bay are often not a concern. Many X-braced frames are designed assuming that the braces are capable of resisting tensile loads only. When this approach is taken, the designer typically assumes that the compressive braces buckle under negligible load and can therefore be neglected in the analysis, resulting in a statically determinate structure, and a simplication of the design. Also, this assumption permits very slender elements to be used for the braces, including rods, single angles, at bars, and cables, making for an economical structure.
2003 by CRC Press LLC

Seismic Design of Steel Structures

12-23

FIGURE 12.25

Fractured rod brace in support structure for an elevated water tank.

Although tension-only braced frames are popular with designers, they have not performed particularly well in past earthquakes and modern building codes discourage their use, except for smaller structures or in zones of low seismicity. There are several problems associated with tension-only braces. Many such braces are installed with some slack. When they are stressed into tension by the structures response to ground shaking, they experience an impact-type loading as slack is taken up. This impact loading can cause a brittle fracture of the brace. Similar effects can occur as tension braces that have been buckled by induced compression are pulled into tension, as the structure cycles from one direction of response to the opposite. This effect can be made more severe by the fact that buckling of the braces in compression often results in localized plastic hinging near end connections and at the mid-span of the brace. This can cause very high local strains and can lead to low-cycle fatigue-type fractures under a few cycles of motion. Figure 12.25 is a picture of fractured rod bracing on a water tank, one of many examples that are available in the literature. As a result of these problems, some building codes require that tensiononly frames be designed to resist design shaking in an elastic manner. While this does result in a stronger frame than would otherwise be required, it does not directly address the problems associated with this structural system and is probably an ineffective solution. If X-braced frames are designed for both tension and compressive behaviors, the braces must often be made quite heavy in order to meet minimum slenderness requirements contained in the codes. A common design question relates to the effective length that should be used when proportioning the braces for compression. Research conducted by Goel [1986] and others suggests that in X-braced frames the tension brace effectively braces the compression brace both for in-plane and out-of-plane loading so that the length of X braces may be taken as half the full diagonal length. In addition, if end connections provide signicant rotational restraint, as is commonly the case for in-plane buckling behavior, effective length factors less than one may typically be applied. Hollow structural sections are commonly used as brace elements in X-braced frames because they are an economical section for obtaining large radius of gyration and meeting minimum slenderness requirements for compressive design. However, when these sections buckle they are subject to local buckling and a type of behavior known as oil-canning. This has frequently lead to brittle brace fracture. Goel [1992] recommended minimum widththickness ratios for the anges of tubular sections to avoid this behavior, or the lling of tubular members with a solid material, such as concrete, to brace the anges

2003 by CRC Press LLC

12-24

Earthquake Engineering Handbook

ac Br

Column

2t

Plastic hinge line

Beam

FIGURE 12.26

Illustration of gusset plate detail for special CBF.

against local buckling. Codes specify minimum widththickness ratios for elements of braces of various cross sections for the same reasons. X-braced frames may qualify either as special CBFs, ordinary CBFs, or nondetailed frames. To qualify as special CBFs, brace elements must be selected with adequate widththickness ratios to minimize the potential for local buckling, braces must conform to minimum slenderness criteria, connections must be designed with adequate strength to ensure that inelastic behavior of the frame will be governed by brace buckling and/or yielding, rather than connection failure, and end connections of braces must be proportioned to accommodate buckling of the brace, without damage to the gusset plate. Figure 12.26 illustrates such a detail. In this detail, a plastic hinge line, perpendicular to the axis of the brace, is drawn across the gusset plate. The gusset plate must be free to rotate, in bending about this line, without restraint from the attachment of the plate to the beams or columns. The brace is held back a distance equal to twice the plate thickness from this free rotation line. Connections detailed in this manner will be able to accommodate brace buckling through benign plastic hinging of the plate along this rotation line. Ordinary X-braced frames may meet somewhat relaxed criteria for brace slenderness and widththickness ratios relative to special X-braced frames, but the braces themselves must be designed somewhat stronger than for special frames. Brace connections must still be designed strong enough to force inelastic behavior to occur in the brace as opposed to the connection; however, there are no specic detailing or conguration requirements for the gusset plate. Nondetailed X-braced frames may be designed without regard to special limitations on slenderness or widththickness ratios, except as contained in the standard design specication. Special CBFs were introduced as a structural system in the mid-1990s, due to specic concern about the poor performance of ordinary braced frames, as observed both in laboratories and in real buildings following earthquakes. Preferential design criteria, in the form of reduced design forces, were provided for special frames as inducement for engineers to specify them, but no penalties were applied to ordinary frames. It is likely that future codes will severely restrict the use of ordinary braced frame systems or further penalize the design requirements as a means of discouraging their use. 12.4.1.2 Single Diagonal Frames Single diagonal frames are similar to X-braced frames except that rather than pairs of opposing braces in each braced bay, only a single diagonal brace is provided. Since steel members are subject to signicant
2003 by CRC Press LLC

Seismic Design of Steel Structures

12-25

Undeformed shape

Floor beam deflects downward

Buckled compression brace

FIGURE 12.27

Deformed shape of chevron-braced frame.

postbuckling strength degradation, this system tends to have poor inelastic cyclic response behavior. When this system is used, this can be mitigated somewhat by providing pairs of opposing frames with the braces in each of the frames oriented in opposing directions. This assures that regardless of the direction of ground motion, there will be some braces acting in tension at each level. Single diagonal frames can be designed as either special, ordinary, or nondetailed systems. The design requirements for this system are similar to that for X-braced systems. In addition, for special and ordinary systems, there is a requirement that not more than 70% of the lateral force in any direction be resisted by tension braces. This ensures a reasonable distribution of tension and compression braces in each potential direction of response. 12.4.1.3 Chevron, V-Braced, and Zipper Frames Chevron and V-braced frames are very popular for commercial building construction because they are an economical alternative and also because the V or inverted-V pattern of the bracing provides opportunity for relatively large and unobstructed window and door openings. When these frames are loaded laterally, the lateral loading will tend to induce tensile forces in one brace in each bay and compression forces in opposing braces. If the beams in braced bays are designed such that they rely on vertical support from the braces to resist dead and live loads, this structural system should be classied as a bearing walls system for the purposes of establishing the design seismic forces. Otherwise, the system may be classied as a building frame system under the provisions of the building code. Regardless of whether the beams are designed to support tributary gravity loads independent of the braces, this frame conguration has a signicant performance issue related to post-elastic behavior. The nonlinear behavior of most chevron and V-braced frames is controlled by buckling of the braces that are loaded in compression. Once the compression side braces buckle, they tend to degrade in strength very quickly, while the tension brace will have reserve strength. As a result of this, as the compression braces degrade in compressive strength and the tensile braces remain effective in resisting additional load, each pair of braces will place an unbalanced vertical component of load on the beams. This load can be very large and has often resulted in undesirable vertical plastic deformation of the beams and supported oor systems (Figure 12.27). Starting in the mid-1980s, this undesirable performance characteristic came to be recognized and the design specications were revised to require that chevron and V braces be designed stronger than their X-braced counterparts. This was done in part as a penalty, to dissuade engineers from using this system and also as a means of minimizing the amount of damage that occurred. However, the penalty was not sufciently large to accomplish either objective. Chevron and V-braced frames may be designed as special, ordinary, or nondetailed systems. When designed as special CBFs, the beam at the apex of the chevron or V must be designed with sufcient strength to resist, in addition to gravity loads, the unbalanced vertical component of forces from the braces, assuming that the postbuckling strength of the compression brace is equal to 30% of the buckling load and that the tension brace develops its full yield strength. This results in a considerable design penalty.
2003 by CRC Press LLC

12-26

Earthquake Engineering Handbook

Undeformed shape

Buckled compression brace

Deformed column
FIGURE 12.28 Deformed shape of K-braced frame.

The zipper-braced frame is an alternative approach to dealing with the problem of postbuckling behavior of multistory chevron and V-braced frames. In this frame system, a vertical post, or zipper column, is installed between the beams supporting the apex of the V or inverted-V braces, but not extending to the ground (Figure 12.24). When the compression braces start to buckle, this zipper column redistributes the unbalanced load to other stories of the structure, providing for improved inelastic response and reduced oor beam deformation. 12.4.1.4 K-Braced Frames K-braced frames are similar in conguration to chevron or V-braced frames except that the V formed by the braces is oriented horizontally and has its apex at a column rather than at a beam. This is an extremely undesirable conguration for braced frames intended for seismic resistance as, once one of the braces buckles in compression, the unbalanced force from the tensile brace will result in large lateral deformation on the column, which can initiate column buckling and structural collapse (Figure 12.28). For this reason K-braced frames are not permitted to be designed either as special CBFs or ordinary CBFs. They are permitted in nondetailed systems in zones of low seismic risk only. 12.4.1.5 Eccentric Braced Frames Eccentric braced frames (EBFs) are a relatively recent innovation, developed in the early 1980s, primarily on the basis of research conducted at the University of California at Berkeley [Popov et al., 1989]. A variety of EBF congurations are possible (Figure 12.29). In this framing system, the diagonal braces are intentionally congured such that their work points are nonconcentric either with beam column joints in the case of single-diagonal systems, or with other braces in multidiagonal systems. The resulting eccentricity induces exural and shear stress in the beams. These systems are intentionally designed so that nonlinear behavior is developed through plasticity in the beams. This protects the braces from buckling and results in a substantial amount of energy dissipation, a desirable property for earthquake resistance. Either shear plasticity or exural plasticity may occur, depending on the link length. The segment of the beam in which the plasticity occurs is called the link and the beam itself is called the link beam. Beams with long links will be controlled by exural plasticity and the development of plastic hinges at either end of the link. Beams with short links will develop plasticity through shear yielding distributed along the length of the link. Short shear-dominated links are preferred to longer exural-dominated links [Malley and Popov, 1984]. This is because the distributed plasticity along the length of the shear link provides for increased energy dissipation with reduced concentrated deformation and damage, as compared with exural links.

2003 by CRC Press LLC

Seismic Design of Steel Structures

12-27

Link

FIGURE 12.29

Typical EBF congurations.

EBFs tend to be more exible than CBFs, resulting in somewhat increased damage to nonstructural elements at low levels of ground motion. However, their behavior under intense ground shaking is far superior to that of CBFs. Consequently, they are permitted to be designed for substantially reduced lateral forces, relative to CBFs. Critical consideration in the design of EBFs include providing adequate bracing of the link beam, so that it can develop plasticity without experiencing exural-torsional buckling, and provision of sufcient stiffener plates within the web of the link beam to avoid lateral buckling of the web. Design specications include prescriptive requirements for these aspects of the design. In addition, the columns of the EBF frame must be designed with sufcient strength to resist the axial forces resulting from development of a full mechanism in the frame. 12.4.1.6 Incomplete Braced Frames Incomplete braced frames have diagonal bracing in some stories of the frame, but not in others. This is a common form of construction for industrial applications where vehicle or maintenance access is facilitated at the lower story. Essentially, this system behaves more like a moment-resisting frame than a braced frame. Deformation in these structures occurs primarily as a result of bending of the columns in the unbraced stories and nonlinear behavior typically occurs through development of plastic hinges in the columns and the formation of a story mechanism (Figure 12.30). This is an undesirable behavior mode as it can result in the development of very large interstory drifts within the unbraced story, and potentially, development of P-delta (P-) instability and collapse. This system is permitted only as a nondetailed seismic system, in zones of low seismic risk. 12.4.1.7 Knee-Braced Frames Knee-braced frames are another hybrid system with behavior characteristics more like that of a momentresisting frame than a braced frame. Essentially, in a knee-braced frame, rigidity of the beamcolumn joint is achieved by placing a short brace between the end region of the beam and the column. If properly designed, these structures can have good performance characteristics similar to those of a momentresisting frame. Important design considerations include ensuring that the brace and frame connections are able to develop the strength of the connected members, ensuring that nonlinear behavior of the frame is controlled through a preferential mode, such as exural hinging of the beams, and ensuring that the intersection points of the knee braces with the beams and columns are adequately braced to prevent
2003 by CRC Press LLC

12-28

Earthquake Engineering Handbook

Plastic hinge

FIGURE 12.30

Deformed shape of incomplete braced frame.

initiation of torsional buckling. The codes do not include any provisions directly governing the design or detailing of these systems; however, they can be designed as a nondetailed system. If the above design considerations are not accounted for in the design, performance of such frames can be poor. 12.4.1.8 Unbonded Braced Frames Unbonded braced frames, also known as buckling-restrained braced frames, are a new technology developed in Japan (see www.unbondedbrace.com). These frames are congured similar to X-braced frames or single-braced frames; however, the braces are specially detailed to allow both tensile and compressive yielding to occur without onset of buckling or fracture. The brace consists of a at bar structural member, inside a larger tubular member. The tubular member is provided only as a buckling restraint for the at bar, which carries the actual bracing force. A ller material or devices are placed between the at bar and the tube walls so that the bar cannot buckle. The resulting structural system has a number of seismic performance advantages. It is a relatively stiff system, so that the building will experience relatively little deformation and damage under low levels of ground motion. Under more intense levels of ground motion, the braces yield, both in compression and tension, producing extensive hysteretic energy dissipation and effectively reducing the level of structural response. Important considerations in the design of these systems include assuring that the connections are adequate to develop the yield strength of the brace and assuring that yielding is uniformly distributed along the length of the brace so that large inelastic strains and ductility demands, potentially leading to fracture, are not developed. Ensuring that the at bar brace does not bind against the side walls of the tube is important to achieving this uniform stress and strain distribution.

12.4.2 Design Approach


Braced-frame members are designed to resist the forces specied by the building code based on the type of structural system selected and the location of the building site relative to various faults and seismic source zones, as determined from seismic risk or zonation maps. Under the requirements of the AISC Seismic Provisions the brace members of an ordinary braced frame, except chevron congurations, are designed for the force corresponding to the application of the specied base shear force per the applicable building code. In the case of chevron or V braces, the design force is increased by 150%. However, this requirement of 150% increase in the design force is not applicable if the chevron is designed as a special concentric brace frame (SCBF). All bracing connections are required to be capable of resisting the maximum expected force that could be delivered to them by the bracing conguration. The design intent is that the strength of all of the brace frame components (beams, columns, connections) be larger than the expected maximum capacity of the brace member. By ensuring this, the failure of a braced-frame system is intended to be controlled by yielding and buckling of the braces only, not the other elements of the frame. As soon as braces yield in tension or buckle in compression, they start to plastify under increasing lateral loads. As
2003 by CRC Press LLC

Seismic Design of Steel Structures

12-29

full plastication occurs, the stiffness and load-carrying capacity of the brace are limited and, therefore, the load that may be attracted to the brace frame as a whole is limited. As a result only the brace member will be damaged and will require repairs after an earthquake, whereas all other components of the braced frame will be undamaged and require no repair. Note that as previously discussed, the design requirements for ordinary V- and chevron-braced frames are not adequate to accomplish this objective, as the beams at the apex of the V or chevron are vulnerable to damage. The AISC Seismic Provisions require that brace connections be designed for the lesser of the following forces: The strength of the brace in axial tension, given by As Fy , where As is the area of the brace and Fy is the yield strength of the steel. An overstrength factor (0) times the design force in the brace including gravity loads. The value of 0 is specied by the building code, depending on the structural system, but typically has a value ranging between 2 and 3. The maximum force that can be transferred to the brace by the system, considering other limiting factors, such as the capacity of diaphragms to transfer shear forces to the braced frames. Columns in braced frames must also be designed with adequate strength to resist the axial loads resulting from the above three load conditions. U.S. building codes permit structural steel elements to be designed to either of two specications: An allowable stress design (ASD) format specication [AISC, 1989] A load and resistance factor (LRFD) specication [AISC, 2001] Although both are permitted, the LRFD specication is far more suitable to design for seismic loading, since it directly considers the yield state of various members and their connections.

12.5 Unbraced Frames


Unbraced, or moment-resisting, frames rely on the exural rigidity of their beams, columns, and beamcolumn connections for lateral-force resistance and stability. Figure 12.31 illustrates the typical lateral deformation pattern experienced by moment-resisting frames in the elastic regime of behavior. As the frame deforms laterally, there is a tendency for the angle between the beams and columns to change. The rigidity of the beamcolumn connection resists this change through development of bending moments and shears in the beams and columns. Connections in moment-resisting frames are designated either fully restrained or partially restrained, depending on the stiffness and strength of the connection, relative to that of the beams and columns. A fully restrained connection must be capable of holding the angle between the beam and column essentially constant, until the weaker of the beam and or column yields in exure. All other connections are classied as partially restrained because they permit some change in the angle of intersection between the beam and column at load levels less than that which initiates yielding of the frame elements. Partially restrained connections may either be partial strength connections or partial rigidity connections. Partial strength connections may be fully rigid but exhibit yielding of the connection elements and increased exibility at load levels lower than those that would cause yielding of the beams or columns. Partial rigidity connections are exible and permit some angular rotation to occur at the beam column connection, even at low levels of loading. Both fully restrained and partially restrained frames may be classied as special moment-resisting frames, intermediate moment-resisting frames, ordinary moment-resisting frames, or nondetailed moment-resisting frames. Properly congured and detailed moment-resisting frames are capable of dissipating extensive amounts of energy and, therefore, can provide superior seismic performance. However, even frames with appropriate conguration and detailing tend to be quite exible, which can lead to extensive damage to nonstructural building components.
2003 by CRC Press LLC

12-30

Earthquake Engineering Handbook

Undeformed shape

Deformed shape M M

V/2
FIGURE 12.31

V/2

Lateral deection of moment-resisting frame.

The conguration of a moment-resisting frame is a critical factor in its performance capability. Because moment-resisting frames tend to be relatively exible, they can develop large interstory drifts when subjected to strong ground shaking. If interstory drifts become too large, the frame can develop P-delta instability and collapse. The amount of drift induced in a building by an earthquake is a function of the intensity of the ground shaking, the spectral content of the ground shaking, the dynamic properties of the structure, primarily, its natural period of vibration and natural mode shapes, and its nonlinear deformation characteristics. If a structure responds to ground shaking within its elastic range of behavior, the structures fundamental mode shape is particularly important. Figure 12.32 shows the natural mode shape for two twostory structures. In the structure on the left, the stiffness of the structure is uniformly distributed vertically, relative to the mass of the structure. Therefore the structure experiences approximately the same amount of lateral deection, termed interstory drift, in each story. The structure on the right is much stiffer in the upper story than it is in the lower story. As a result, most of the lateral deection in this structure is accommodated through interstory drift in the rst story. Such structures are commonly termed softstory structures. If the two structures in the gure have the same natural period of vibration, the total lateral displacement demand (i.e., at the top of the structure) induced on the two structures by a given ground motion will be the same. However, in the structure with uniform stiffness, each story of the structure will experience approximately the same amount of interstory drift, approximately equal to one half the total drift demand on the structure. On the other hand, the weak story structure will accommodate almost all the drift in a single story. This can lead to an accumulation of damage in this story, premature degradation and failure of the framing in that story, and also the development of P-delta instability and collapse. For this reason, it is particularly important to design moment-resisting frame structures without soft-story congurations. Even structures that have uniform distributions of elastic stiffness can develop inelastic soft-story conditions if an inappropriate pattern of plastic hinges form in the structure as it undergoes nonlinear lateral deformation. When a moment-resisting frame structure becomes completely plastic it is said to form a mechanism, that is, a behavioral mode in which the structure exhibits neutral stability and can deform laterally without application of load. Frame structures can form a variety of different types of lateral mechanisms. The most common types are illustrated in Figure 12.33. The frame on the left side of the gure develops inelastic behavior through the formation of plastic hinge zones at the ends of the beams, adjacent to the columns. Once a plastic hinge forms at both ends of each beam and also at the base of each column, the frame can undergo free lateral translation through a rigid body rotation of the columns. This is termed a beam-hinge mechanism and is a preferred mode
2003 by CRC Press LLC

Seismic Design of Steel Structures

12-31

Uniform Stiffness Distribution

Soft Story Stiffness Distribution

FIGURE 12.32

Mode shape for structures with uniform stiffness and soft-story stiffness distributions.

Beam Hinge Mechanism Panel Zone Hinge Mechanism

Column Hinge Mechanism

FIGURE 12.33

Various lateral frame mechanisms.

of behavior because it distributes inelastic deection uniformly throughout all stories. The frame in the middle of the gure develops inelastic behavior through the formation of shear yielding in the panel zone of the connections between the beams and columns. These so-called panel zone hinges allow the beams to rotate freely, relative to the column, and also permit free lateral displacement of the frame to occur through rigid body rotation of the columns. This behavioral mode is not as desirable as the beam hinging mode because yielding of the column panel zone, though quite ductile and capable of dissipating extensive energy, can result in local kinking of the column anges and can initiate failure in certain types of beamcolumn connections. The frame on the far right in Figure 12.33 develops a mechanism through the formation of plastic hinges at the tops and bottoms of the columns in a single story. This permits free lateral displacement of the structure through rigid body rotation of the columns in that particular story, without participation of the other columns. This condition is known as a single-story mechanism. Like the weak story conguration, it is undesirable because it concentrates deformation and damage in a relatively small region of the structure and can result in large interstory drifts in a single story and possible P-delta instability. Five basic moment-resisting frame systems are available in the building codes. These are termed: Special moment-resisting frames Intermediate moment-resisting frames Ordinary moment-resisting frames Special-truss moment frames Nondetailed moment frames

2003 by CRC Press LLC

12-32

Earthquake Engineering Handbook

As with nondetailed braced frames, the nondetailed moment frame may be used only in regions of very low seismic risk and must be designed for relatively large design forces. There are no specic requirements with regard to the detailing and conguration of the frame, other than the standard limitations on slenderness and section compactness contained in the design specications for resisting loads other than seismic. Frames that employ partially restrained connections are subject to the same requirements as frames employing fully restrained connections, except the reduction in stiffness and strength of the frame introduced by these connections must be accounted for in the structural analysis used to evaluate the design forces and lateral deections. Often, this requires the use of a nonlinear analysis methodology. The remaining sections discuss the design considerations for special momentresisting frames, intermediate moment-resisting frames, ordinary moment-resisting frames, and specialtruss moment-resisting frames.

12.5.1 Special Moment-Resisting Frames


Special moment-resisting frames (SMFs) are congured and detailed to provide a high level of system ductility. Members of special moment-resisting frames must be either hot-rolled structural shapes or welded built-up shapes and must conform to material specications capable of providing adequate ductility. In the United States, A-36, A572, Grade 50, and ASTM A913 Grade 50 or Grade 65 are approved for use in special moment-resisting frames. Structural shapes in moment-resisting frames must conform to strict limitations on widththickness ratios, called compact section criteria, so that plastic hinging can form without premature fracture of the section. In addition to these requirements, the frame must be congured such that there are no soft or weak stories, as described above. Under the AISC Seismic Provisions, weak stories are avoided through the requirement that at each beamcolumn connection, the following relationship be satised:

M M

* pc * pb

> 1.0

(12.12)

In this equation, the top term is the sum of the plastic moment capacities of the columns framing into the joint, reduced for the effect of axial load on the columns, while the bottom term is the sum of the plastic moment capacities of beams framing into the joint. The plastic moment capacity of the beams framing into the joint are calculated considering potential overstrength of the beam, that is, the potential that the actual yield strength of the steel will be somewhat larger than the nominal strength for the material specication and also strain-hardening effects. This is accomplished through the following formula: M pb = 1.1 Ry Fy Z (12.13)

where Ry is a material specication-related coefcient that represents the ratio between the median strength of steel produced to a particular specication and the nominal strength, Fy is the nominal yield strength, and Z is the plastic section modulus for the beam section. For most structural steels Ry is assigned a value of 1.3. Frames that satisfy this requirement are known as strong-column frames because the columns are stronger in exure than the beams and the frame will develop either a beam-hinge or column panel-zone type hinge mechanism. Early moment-resisting frames were highly redundant with nearly every beamcolumn connection provided with moment resistance so that the entire structural frame would participate in lateral force resistance. As bolted connections were replaced by welded connections in the 1970s, the labor cost associated with making moment-resisting connections became signicant; to economize, engineers began to congure frames with fewer participating beams and columns, so as to minimize the number of welded connections required. As the number of beams and columns in the frames was reduced and the frames
2003 by CRC Press LLC

Seismic Design of Steel Structures

12-33

became less redundant, it became necessary to use relatively large members for the frames providing lateral resistance. This trend continued until 1994 and the Northridge earthquake, after which a number of moment-resisting frames were found to have experienced brittle fractures of welded moment-resisting connections. The lack of redundancy in modern steel frame construction both surprised and shocked a number of engineers and code writers [Holmes and Somers, 1996]. In addition, under the FEMA/SAC program research that was initiated in response to these failures, it was determined that momentconnections in deep members are more likely to experience brittle fracture than those in shallower members. Consequently, in 1997, the building codes were amended to require that minimum levels of redundancy be provided in SMFs. This is accomplished through the calculation of a redundancy coefcient, x , per the following equation: x = 2 20 rmax x Ax (12.14)

where Ax is the oor area, in square feet at level x of the structure, and rmaxx the maximum of the sum of the shears in any two adjacent columns in the plane of a moment frame divided by the story shear. Special moment-resisting frames are not permitted to have congurations that result in values of x exceeding 1.25. During the period 1970 to 1994, most moment-resisting frames were constructed using a standard connection in which beam shears were transferred to the column through a shear plate that was welded to the column and high-strength bolted to the beam web, while exural stresses were transmitted through a complete joint penetration butt weld of the beam anges to the column. This fully restrained connection was considered to be capable of providing acceptable performance based on limited cyclic testing conducted at the University of California at Berkeley [Popov and Stephen, 1972]. Only a few such tests were performed, and many of these tests were on small structural sections. However, in 1985, the building codes adopted provisions requiring that this connection be used unless suitable test data were provided to substantiate the performance capability of alternative details. During research on the behavior of eccentric braced frames, Ksai and Popov [1986] and Engelhardt and Popov [1989] noted the vulnerability of this connection when subjected to cyclic inelastic deformation. However, these warnings were largely overlooked by the profession until the occurrence of the 1994 Northridge earthquake. The FEMA/SAC program conducted extensive research into the causes of the brittle fractures of these connections. Roeder [2000] determined that the behavior of these connections was both brittle and erratic. Based on extensive review of cyclic tests performed on these connections, Roeder suggested a mean plastic rotation capacity, p, in radians, for this connection as given by the equation: p = 0.051 0013 db where db is the beam depth, in inches, and a standard deviation given by the equation: p = 0.0044 + 0.0002 db (12.16) (12.15)

For 36-in.-deep beam sections, these equations provide a mean plastic rotation capacity of 0.004 radians and a coefcient of variation of 280%. For shallower sections, the behavior is somewhat better. For example, 24-in.-deep sections have a mean plastic rotation capacity of 0.02 radians and a coefcient of variation of 50%. The FEMA/SAC project concluded that this poor behavior was due to a variety of factors previously discussed above. Based on the FEM/SAC recommendations the building code adopted requirements that connection details used in SMFs be qualied through approved programs of cyclic inelastic testing, to be capable of withstanding at least 0.03 radians of cyclic plastic rotation without developing brittle fracture or experiencing signicant strength degradation. The FEMA/SAC project developed a series of details
2003 by CRC Press LLC

12-34

Earthquake Engineering Handbook

Reduced Beam Section (RBS)

Free Flange (FF)

Welded Flange, Welded Web (WUF-W)

Bolted End Plate (BEP)

Bolted Flange Plate (BFP)

Bolted T-Sub (BTS)

FIGURE 12.34

Prequalied special moment-resisting frame connections.

L/2 Plastic hinge L L

s
Note: If 2Mpr /L is less than the gravity shear in the free body (in this case P/2 + wL/2), then the plastic hinge location will shift and L must be adjusted accordingly.

P Mpr=1.1RyZbFy Mpr Vp w
L

VA A Mpr

taking the sum of moments about A = 0 Vp ={Mpr + Mpr + P L/2 + wL2/2}/L

FIGURE 12.35

Determination of beam shear at formation of plastic mechanism.

that were recommended as prequalied for this service, based on testing and analysis of these connections performed as part of the project. These connection details are illustrated in Figure 12.34. Included in the FEMA-350 guidelines are specic design criteria for each of these connection types. The design criteria include specic detailing requirements, weld metal and base metal physical property requirements, quality assurance requirements, and limitations on the size of member for which the connection is qualied. The design procedure for each of these connections is strength based and considers the formation of a beam-hinge type mechanism in the frame. Specically, the connection is designed so that a plastic hinge will form in the beam at a predetermined distance, s, from the face of the column. Based on this, the shear in the beam at the formation of a beam-hinge mechanism is determined, as indicated in Figure 12.35. Then a series of free bodies are taken of the connection and a portion of the column, and using these free bodies, as shown in Figure 12.36, the critical strength demands on various elements of the connection are determined. Each of these various elements, including the column panel zone, as well as the various welds, plates, and bolts, must be provided with adequate strength to ensure that plastic hinging in the beam can occur. Appendix 12A illustrates application of the design procedure for a typical reduced beam section-type connection.
2003 by CRC Press LLC

Seismic Design of Steel Structures

12-35

Plastic hinge

Plastic hinge

Mf

Mpr Vp
x

Mc
dc x+dc/2

Mpr Vp

Mf=Mpr +Vpx
Critical Section at Column Face

Mf=Mpr +Vp(x+dc/2)
Critical Section at Column Centerline

FIGURE 12.36

Typical free body diagrams used to determine required design strength of connection elements.

Mct Vct Cfct Tfct Tfb Vpz Vb

Vcb

Tfcb

Cfb Cfcb

Mb

Mcb

FIGURE 12.37

Forces on column panel zones.

Since the range of applicability of the prequalied connections is limited to congurations and section sizes tested and evaluated under the FEMA/SAC program, the prequalied connections are not applicable to all possible frame designs. For those cases where no prequalied connection is applicable, the AISC Seismic Provisions require a program of qualication testing and evaluation to substantiate the performance capability of the design. Appendix S of the AISC Seismic Provisions species the requirements of the qualication program. Column panel zones at beamcolumn connections are a key consideration in SMF design. The transfer of exural stresses between beams and columns at a moment-resisting connection is accomplished through shearing of the panel zone. Figure 12.37 illustrates this behavior. The gure shows a momentresisting beamcolumn connection. The panel zone is that portion of the column web between the top and bottom beam anges. The stress resultants applied by the beam to the panel zone are a moment Mb and shear Vb . The top and bottom portions of the column, respectively, apply moments and shears Mct , Vct, Mcb , Vcb to the panel zone. Each of these moments can be idealized as being applied as a couple of
2003 by CRC Press LLC

12-36

Earthquake Engineering Handbook

forces, applied at the centroid of the beam and or column anges. These are shown in the gure as forces Tfb , Cfb for the beam, and with similar notation for the column. By cutting a section through the panel zone, just below the beam top ange, it is seen that the panel zone has a net shear demand Vpz, given by the equation: Vpz = Tfb Vct (12.17)

If the panel zone shear exceeds the shear strength of the material, the panel zone will yield, creating the effect of a pinned connection between the beam and column and limiting the forces on these elements. In the 1970s, design specications required that panel zones in SMFs be capable of developing the strength of the beam. Panel zone strength was calculated as the quantity 0.55Fydctw , where Fy is the nominal yield strength of the column steel, dc is the column depth, and tw is the column web thickness. Often this required that the panel zones be reinforced, typically with the placement of a at plate, commonly termed a doubler plate, adjacent to the web, welded to the column and any stiffener plates present. In the 1980s, concurrent with research performed on eccentric braced frames, it was postulated that panel zone shear yielding was a desirable behavior, capable of extensive energy dissipation and that this, in fact, might be preferable to frame behaviors that rely on the formation of plastic hinges in the beams. The 1985 Uniform Building Code [ICBO, 1985] was modied to permit the use of weaker panel zones by permitting panel zones that could develop only a fraction, 85% of the beam plastic exural strength. In addition, a revised formulation was used for calculation of panel zone shear strength that accounted for the fact that complete yielding of the panel zone also required exural yielding of the column anges. This new formulation was: 3b t 2 Vy = 0.6 Fydc t w 1 + cf cf dbdc t w

(12.18)

where bcf and tcf are the width and thickness of the column ange, db is the depth of the beam, and other terms are as previously dened. These combined actions resulted in panel zones that were substantially weaker than those previously used. In the FEMA/SAC research, it was determined that excessively weak panel zones were detrimental to connection ductility. In particular, the exural yielding of the column anges, relied on in Equation 12.18, resulted in severe distortion of the column anges at the beamange connection, and this could initiate weld fracture at this location. The research also found that maximum frame ductility could be achieved when the yielding at beamcolumn connections was shared between the beam and the column panel zone. Therefore the design procedures for the new prequalied connections attempt to balance the strength of the panel zone with the exural strength of the beam. In addition to the prequalied connections developed by the FEMA/SAC program, several proprietary connection technologies are available in the marketplace. The developers of these proprietary technologies have developed qualication data for these connections, using procedures generally following those contained in Appendix S of the AISC Seismic Provisions, and now license the use of these technologies for individual projects. One such technology, known as the slotted web connection, is similar to the standard bolted web, welded ange connection commonly used prior to the Northridge earthquake, except that long slots are cut into the beam web adjacent to the top and bottom beam anges. These slots signicantly reduce the stress concentrations inherent in the standard pre-Northridge connection. A second technology, known as the side plate connection, uses a series of heavy shear plates to transfer exural stresses from the beam to the column. A pair of shear plates is placed on each side of the column and beam, and additional plates are used to transfer exural stresses from the beam anges to these shear plates, then back to the column anges. Information on these and other proprietary technologies may be obtained directly from the licensors.

2003 by CRC Press LLC

Seismic Design of Steel Structures

12-37

12.5.2 Intermediate Moment-Resisting Frames


Intermediate moment-resisting frames (IMFs) are similar to SMFs except that the detailing and conguration requirements are somewhat relaxed and, in particular, a broader collection of beamcolumn connection details may be used. In particular, strong columnweak beam and panel zone strength requirements specied for SMFs are not required for IMFs. Also, connections in IMFs need only be qualied to be capable of resisting 0.01 radians of plastic rotation demand, as specied in Appendix S of the AISC Seismic Provisions. A wide range of partially restrained and fully restrained connection details are available to meet these requirements. The building codes require that IMFs be designed stronger than SMFs, reducing the amount of ductility demand on IMF frames as compared to SMFs. In addition, the building codes generally limit the use of IMF structures to buildings in zones of moderate seismicity and to certain classes of low-rise construction in zones of higher seismicity. These frames are commonly used in light residential construction due to their relative economy.

12.5.3 Ordinary Moment Frames


Ordinary moment frames (OMFs) are similar to IMFs except that a single prescriptive beamcolumn connection detail, similar to that commonly in use for SMFs prior to the Northridge earthquake, is specied by the AISC Seismic Provisions for this system. Just as with the original SMF connection, commonly used in the 1970s and 1980s, the prescriptive connection employs a shear plate, welded to the column and bolted to the beam web for shear transfer and complete joint penetration groove welds to join the beam anges to the column. Several improvements have been specied for this connection, however. First, weld ller metal used in joining beam anges to columns must provide minimum specied toughness. Second, weld runoff tabs and backing bars must be removed from the completed weld; the weld must be backgouged and rewelded to remove any surface defects or discontinuities. Finally, weld access holes, cut in the webs of beams to permit welding continuously across the beam ange, must conform to a specied shape criterion to minimize the potential for stress and strain concentrations around this discontinuity. OMFs must be designed with larger strength levels than IMFs and are subject to additional limitation with regard to the size of structure and zones of seismicity in which they can be utilized.

12.5.4 Special Truss Moment-Resisting Frames


Long-span and industrial structures commonly employ fabricated trusses, rather than rolled shapes, as the horizontal members of frames. For many years, buildings and other structures were commonly designed as SMF structures, but using trusses as the horizontal framing elements. Unfortunately, the exural strength of a truss used as a girder in a frame will almost always exceed the exural strength of frame columns, and therefore moment-resisting frames employing trusses as the girder elements almost always have weak-story congurations and will form single-story mechanisms, an undesirable condition. The special truss moment-resisting frame (STMF) was developed based on research performed at the University of Michigan [Goel and Itani, 1994] specically to address this problem. In the STMF, the truss is specically designed to have inelastic behavior controlled by shear yielding of the truss, in specially designed panel zones, near the truss mid-span. Figure 12.38 illustrates this behavior. In order to accommodate this inelastic behavior, prescriptive requirements are specied for the design of the special shear zone of the truss as well as other members in the system, to ensure that yielding will indeed occur in the special zone. Yielding must be limited to the middle one third of the truss span and X-pattern diagonals must be used in this zone. Truss chords must be compact and laterally braced. Though relatively few of these structures have been constructed, it is anticipated that they will perform in a manner comparable with other SMF structures.

2003 by CRC Press LLC

12-38

Earthquake Engineering Handbook

Undeformed shape

Deformed shape

Shear Yield Zone

Plastic hinge

FIGURE 12.38

Inelastic deformation of special truss moment frame.

Dening Terms
Brace An axially loaded diagonal member of a frame. Braced-frame A structure that relies on braces for lateral resistance. Butt weld A type of welded joint in which two pieces of metal that abut are joined by lling the
groove between the two pieces with molten metal. See also groove weld.

Complete joint penetration A method of welding in which weld metal is deposited for the full depth
or thickness of the structural elements being joined.

Concentric braced frame A braced frame structure in which the line of action of beams, columns,
and braces at a connection are coincident.

Connection The assemblage of welds, bolts, and other joining techniques used to fasten one member
to another.

Ductility A property of a material, structural element, or structure, by which it can withstand inelastic
straining without failure. Elastic A type of structural behavior in which deformation is proportional to applied force and in which upon unloading, a structure will return to its original undisturbed condition. Fracture The formation of an unstable rupture plane or crack within a piece of material. Fracture mechanics The engineering mathematics used to predict the propensity of materials and structures to fracture. Fracture toughness A mechanical property of a material that relates to its ability to absorb and dissipate energy without developing fracture. Groove weld A type of welded joint in which a groove between two pieces of metal is lled with molten weld metal, also sometimes called a butt weld. Hysteresis The property of a steel structure or element by which it dissipates strain energy through plastic straining. Mechanism A behavioral state for a structure under which it can experience free motion. Mode shape A characteristic deformed shape, in which a structure can move in free vibration. Modulus of elasticity A material-dependent proportionality constant that expresses the relationship between stress and strain during elastic loading of a structure; for steel it has the value 29,000,000 lb/in.2. Moment-resisting frame A structure that relies on the exural rigidity and strength of beams and columns for lateral resistance.
2003 by CRC Press LLC

Seismic Design of Steel Structures

12-39

P-delta (P-) The effect of vertical forces on a laterally displaced structure, typically a condition of
secondary stress.

Period The amount of time, in seconds, it takes a structure in free vibration to experience one
complete cycle of motion. Plastic A type of structural behavior in which increasing deformation can occur under constant load, which upon unloading is not recoverable. Plastic hinge A zone in a exural member where inelastic exural rotation occurs. Principal stress The direct stress on a material, perpendicular to a plane across which there is no shear stress. Rolled shape A structural member with a cross section formed by a rolling process. Strain A dimensionless measure of deformation within a material or member, reective of the percent elongation or contraction of the material or member under applied loading. Strain hardening A type of structural behavior that occurs at loading exceeding the plastic range, in which increased loading is required to produce increased deformation. Stress A measure of the distribution of force across a structural section, in units of pounds per square unit length, such as pounds per square inch (lb/in.2). Tensile strength The peak stress in a standard axially loaded steel tensile specimen, prior to onset of fracture; also called ultimate stress. Truss A structure comprised of multiple elements including diagonal members that resists load primarily through axial behavior of its members. Ultimate stress See tensile strength. Weld A means of joining two pieces of metal together by heating them sufciently to allow melting of both pieces, and resolidication, perhaps with supplemental molten metal, into a single piece. Yield strength The stress in a standard axially loaded steel tensile specimen at which plastic deformation can occur at uniform stress, typically the stress at an elongation of 0.005 in./in.

References
AISC. 1928. Steel Construction, 1st ed., American Institute of Steel Construction, New York. AISC. 1989. Specication for Structural Steel Buildings, Allowable Stress Design and Plastic Design, American Institute of Steel Construction, Chicago. AISC. 1997. Seismic Provisions for Structural Steel Buildings, American Institute of Steel Construction, Chicago. AISC. 2000. Seismic Provisions, Supplement No. 2, American Institute of Steel Construction, Chicago. AISC. 2001. Load and Resistance Factor Design Specication for Structural Steel Buildings, American Institute of Steel Construction, Chicago. Barsom, J.M. 1991. Properties of Bridge Steels, in Highways Structure Design Handbook, Vol. 1, Steel Bridge Alliance, Chicago, chap. 3. Barsom, J.M. and Rolfe, S.T. 1999. Fracture and Fatigue Control in Structures: Applications of Fracture Mechanics, 3rd ed., American Society for Testing and Materials, Conshohocken, PA. CISC. n.d. Why Take a Chance? California Institute of Steel Construction, Southern California Division. Chicago Public Library. 1997. City of Chicago, a Chronological History of Chicago 1673Present, Chicago Public Library, Chicago. Engelhardt, M.D. and Popov, E.P. 1989. On Design of Eccentrically Braced Frames, Earthquake Spectra, 5, 495511. Frank, K.H., Barsom, J.M., and Hamburger, R.O. 2000. State of the Art Report on Base Metals and Fracture, FEMA-355E, Federal Emergency Management Agency, Washington, D.C. Goel, S.C. 1992. Cyclic Post Buckling Behavior of Steel Bracing Members: Stability and Ductility of Steel Structures Under Cyclic Loading, CRC Press, Boca Raton, FL. Goel, S. and El-Tayem, A.A. 1986. Effective Length Factor for the Design of X-Bracing Systems, AISC Eng. J., 23, 4145.
2003 by CRC Press LLC

12-40

Earthquake Engineering Handbook

Goel, S.C. and Itani, A. 1994. Seismic Resistant Special Truss Moment Frames, J. Struct. Eng., 120. Goel, S.C. and Lee, S. 1992. A Fracture Criterion for Concrete-Filled Tubular Bracing Members under Cyclic Loading, Proceedings of the 1992 ASCE Structures Congress, American Society of Civil Engineers, Reston, VA. Hadley, J.M. n.d. How Buildings Withstood the Japanese Earthquake and Fire, Portland Cement Association, Portland, OR. Holmes, W.T. and Somers, P. 1996. Steel Buildings, in Northridge Earthquake of January 17, 1994, Reconnaissance Report, Vol. 2, chap. 2, Earthquake Spectra, Suppl. C to Vol. 11. ICBO. 1997. Uniform Building Code, International Conference of Building Ofcials, Whittier, CA. Ksai, K.C. and Popov, E.P. 1986. General Behavior of WF Steel Shear Link Beams, J. Struct. Eng., 112, 362382. Malley, J.O. and Popov, E.P. 1984. Shear Links in Eccentrically Braced Frames, J. Struct. Eng., 110, 22752295. Nakashima, M. 2000. Overview of Damage Observed to Steel Building Structures in the 1995 Kobe Earthquake, Appendix C, in State of the Art Report on Past Performance of Steel Buildings in Earthquakes, FEMA-355E, Federal Emergency Management Agency, Washington, D.C. Osteraas, J. and Krawinkler, H. 1989. The Mexico Earthquake of September 19, 1985, Behavior of Steel Buildings, Earthquake Spectra, 5, 5188. Popov, E.P. and Stephen, R.M. 1972. Cyclic Loading of Full-Size Steel Connections, AISI Bulletin, 21, American Iron and Steel Institute, Washington, D.C. Popov, E.P., Engelhardt, M.D, and Ricles, J.M., 1989. Eccentrically Braced Frames: U.S. Practice, Eng. J., 26, 6680, American Institute of Steel Construction, Chicago. Reis, E. and Bonowitz, D. 2000. State of the Art Report on Past Performance of Steel Buildings in Earthquakes, FEMA-355E, Federal Emergency Management Agency, Washington, D.C. Roeder, C. 2000. State of the Art Report on Steel Connections, FEMA-355D, Federal Emergency Management Agency, Washington, D.C. SAC. 1996. Background Reports on Metallurgy, Fracture Mechanics, Welding Moment Connections and Frame Systems Behavior, FEMA 288, Federal Emergency Management Agency, Washington, D.C. SAC. 2000a. Recommended Seismic Design Criteria for Welded Steel Moment-Frame Structures, SAC Joint Venture, FEMA-350, Federal Emergency Management Agency, Washington, D.C. SAC. 2000b. Recommended Specications and Quality Assurance Guidelines for Welded Steel Moment-Frame Construction for Seismic Applications, SAC Joint Venture, FEMA-353, Federal Emergency Management Agency, Washington, D.C. SAC. 2000c. Recommended Seismic Evaluation and Upgrade Criteria for Welded Steel Moment-Frame Buildings, SAC Joint Venture, FEMA-351, Federal Emergency Management Agency, Washington, D.C. SAC. 2000d. Recommended Postearthquake Inspection, Evaluation and Repair Criteria for Welded Steel Moment-Frame Buildings, SAC Joint Venture, FEMA-352, Federal Emergency Management Agency, Washington, D.C. Steinburgge, K.V., Schader, E.E., Bridgestone, H.C., and Weers, C.A. 1971. San Fernando Earthquake, February 9, 1971, Pacic Fire Rating Bureau, San Francisco, CA. USGS (U.S. Geological Survey). 1907. The San Francisco Earthquake and Fire of April 18, 1906 and their Effects on Structures and Structural Materials, reports by G.K. Gilbert, R.L. Humphrey, J. S. Sewell, and F. Soule, Bull. 324, Government Printing Ofce, Washington, D.C. Youssef, N., Bonowitz, D., and Gross, J.L. 1995. A Survey of Steel Moment Frame Buildings Affected by the 1994 Northridge Earthquake, NISTIR 5625, National Institute of Standards and Technology, Washington, D.C.

Further Reading
Selected state-of-the-art reports [Reis and Bonowitz, 2000; Roeder, 2000; Frank et al., 2000] and the SAC series of reports [SAC 1996, 2000a, 2000b] are recommended for additional detail on seismic aspects of steel design and performance. AISC [1997] and Barsom and Rolfe [1999] are also recommended.
2003 by CRC Press LLC

Seismic Design of Steel Structures

12-41

Appendix A: Design Procedure for a Typical Reduced Beam Section-Type Connection


Design a reduced beam section (RBS) connection at the second-oor level for the multistory (end bay) special moment-resisting frame shown in Figure A-1. The design is based upon FEMA 350 guidelines.
Third Floor

Moment Connection under Consideration

W14x311

Second Floor

W21x73

W14x311

13 ft.

13 ft.

22 ft. 8 in.

FIGURE A-1 Special moment-resisting frame.

Frame dimensions
Story h (above) = Story h (below) = Bay width L = 156 in. 156 in. 272 in. htop = 67.38 in. hbot = 67.38 in. Column exterior

Section properties
Beam W 21 73 db = 21.24 in. bf = 8.295 in. tf = 0.74 in. tw = 0.455 in. Zxb = 172 in.3 Sb = 151 in.3 Fyb = 50 ksi Column W 14 311 dc = 17.12 in. bcf = 16.23 in. tcf = 2.26 in. tcw = 1.41 in. Zxc = 603 in.3 Sc = 506 in.3 Fyc = 50 ksi

2003 by CRC Press LLC

12-42

Earthquake Engineering Handbook

Beam loads and assumed axial stresses in columns


w = 0.17 k/in. fa = 15 ksi

STEP 1

Trial values for RBS dimensions a, b, and c


a = (0.5 to 0.75) bf = 4.1 in. to 6.2 in. b = (0.65 to 0.85) db = 13.8 in. to 18.1 in. c = 0.2 bf < 0.25 bf = 1.7 in. to 2.1 in. a-avrg = 5.2 in. b-avrg = 15.9 in. c-avrg = 1.9 in. Try a Try b Try c R = 5 in. = 16 in. = 2 in. = 17.00 in.

STEP 2

Plastic and section modulus at the minimum section of the RBS


ZRBS = Zxb 2.c.tf.(db tf ) = 111 in.3 SRBS = Sb 2.c.tf.(db tf )2/db = 92 in.3

a Column

c REDUCED SECTION BEAM (RSB) c tf c

Continuity plate

Panel zone

Beam

tw

db

Continuity plate

Beam reduced section

STEP 3

Expected yield stress of the beam and column


Ryb = 1.1 Ryc = 1.1 Fyb = Ryb.Fyb = 55 ksi Fyc = Ryc.Fyc = 55 ksi

STEP 4

Probable plastic moment and at the center of the RBS


Mpr = 1.15*ZRBS*Fye =7021 in.-kip

2003 by CRC Press LLC

Seismic Design of Steel Structures

12-43

STEP 5

Shear force at the centers of the RBS at each end of the beam
L = L dc 2(a + b/2) = 229 in. Vp = 2Mpr /L' + wL'/2 = 81 kips Vp = 2Mpr /L' wL'/2 = 42 kips
dc a b b a dc

Uniform load (w)

Plastic hinge location

Plastic hinge location

L' = L-dc-2(a+b/2)

L Uniform load (w)

Mpr

Mpr

L' = L-dc-2(a+b/2) Vp = 2Mpr/L' + wL'/2 V'p = 2Mpr/L' - wL'/2

STEP 6

Maximum moment and shear force expected at the face of the column
Mf = Mpr + Vp(a + b/2) + w(a + b/2)2/2 = 8088 in.-kip Vf = Vp + w.(a + b/2) = 83 kip

Mf

Mpr

Vp

Vf a+b/2

2003 by CRC Press LLC

12-44

Earthquake Engineering Handbook

STEP 7

Check beam exural capacity at the face of the column


Mpe = Zxb.Fye = 9460 in.-kip Mf/Mpe = 0.85 Beam capacity OK dimensions a, b, c

STEP 8

Check beam shear capacity


Vn = Aw.Fy = (db.tf ).Fyb = 786 kips Vf/Vn = 0.11 Beam capacity OK dimensiones a, b, c

STEP 8

Strong column-weak beam check


h = (htop + hbot +db) = Vc = [Mpr + Vp*(dc/2 + a + b/2)]/h = Vc = [2Mpr + (Vp+Vp)*(dc/2 + a + b/2)]/h = Mctop = Vc * htop = Mcbot = Vc * hbot = Mctop + Mcbot = S [Zc*(Fyc fa)]/(Mctop + Mcbot)] = 156 in. 56 kips 3787 in.-kip 3787 in.-kip 7574 in.-kip 5.57 average story height exterior column interior column

OK

Vc

ht Mct Mpr Mcb V'p Vp hb Vc


a+b/2 dc a+b/2

Mpr db

STEP 9

Continuity plate requirement


minimum required thickness of column ange when no continuity plates are required 0.4*sqrt(1.8*bf.tf.FybRyb/(FycRyc)) = 1.3 in. bf/6 1.4 in. continuity plates not required Recommended thickness of continuity plate (one-sided exterior connection) = thickness of beam ange Use continuity plates 3/4 thick

2003 by CRC Press LLC

Seismic Design of Steel Structures

12-45

STEP 10 Panel zone strength requirement


Cpr = Cy = 1/(Cpr*Zbrs/Sbrs) = h = t req = Cy.Mf.(h db)/h/(0.9*0.6Fye.dc.(db tf )) = t wc = t doubler plates = (treq tcw)/2 No doubler plate required 1.15 0.72 156 in. 0.49 in. 1.41 in. 0.462 in. peak connection strength coefcient average column height required thickness web thickness

2003 by CRC Press LLC

Potrebbero piacerti anche