classification of damage to steel buildings observed in the...

ELSEVIER eIh S0141-0296(97)00019-9

Engineering Structures, Vol. 20, Nos 4-6, pp. 271-281, 1998 © 1997 ELsevier Science Ltd

All rights reserved. Printed in Great Britain 0141-0296/98 $19.00 + 0.00

Classification of damage to steel buildings observed in the 1995 Hyogoken-Nanbu earthquake M. Nakashima

Disaster Prevention Research Institute, Kyoto University, Gokasho, Uji Kyoto, Japan

K. Inoue and M. Tada

Department of Architectural Engineering, Osaka University, Yamadaoka, Suita, Osaka, Japan

This paper presents an overview of the damage to steel buildings observed in the 1995 Hyogoken-Nanbu earthquakes. It summarizes the Japanese seismic design concept and the level of earthquake forces exerted on steel buildings built in severely shaken regions. Statistical data on the damage are presented with respect to the building height and type and the location of damage and typical damage is highlighted. The damage to welded beam-to-column connections is explained from the perspectives of materials, welding, connection details and plastic rotation demand. Finally, the damage was classified with respect to the previous awareness and the difficulty in providing solutions. © 1997 Elsevier Science Ltd.

Keywords: Hyogoken-Nanbu earthquake, steel building, damage, distribution, beam-to-column connection

1. Introduction

Although neither the most powerful nor deadly earthquake of this century, the magnitude 7.2 Hyogoken-Nanbu (colloquially called Kobe) earthquake wreaked havoc throughout Kobe and the surrounding regions and destroyed or damaged a large percentage of the building infrastructure. Over 600(I people were confirmed dead, 26 000 people reported injured and more than 108 000 buildings and homes damaged beyond repair, which rendered more than 300 000 people homeless immediately after the earthquake.

Modem steel buildings sustained serious damage for the first time in the Japanese history of experienced large earthquakes. The 1978 Miyagikenoki earthquake struck urban Sendai City and its surrouoding areas, in which many modem steel buildings had been built, but the reported damage to those buildings remained minimal. Why then, did the Hyogoken-Nanbu earthquake torture modem steel buildings so badly? Ground motions, particularly those experienced in areas where hunda'eds of buildings were clustered, were reported much larger than those observed in previous earthquakes. The Kobe area is one of the earliest urban developments and, as a :result, contained a much larger inventory of steel buildings. Whatever reasons may be

given, the fact remains that modem steel buildings experienced significant damage and it refuted the popular myth that steel buildings are immune to strong earthquakes.

This paper presents an overview of damage to steel buildings observed from the Hyogoken-Nanbu earthquake, presents plausible causes of the damage being discussed in the Japanese engineering community and introduce some of the progress to date for improving the seismic safety of modem steel buildings.

2. Damage to old steel bui ldings

The Architectural Institute of Japan conducted a preliminary reconnaissance from 24 to 26 January 1995 and identified 4530 engineered buildings damaged, including 1067 collapsed or fatally damaged beyond repair 1,2. The Kobe area is one of the earliest urban developments and contained a large stock of relatively old steel buildings more than 35 years old, built before the major economic growth of the post-war era. Figure 1 show examples of damage to such old steel buildings. As shown in Figure lb, these buildings were constructed with bundled light-gauged sections for columns and trusses for beams, consisting of light- gauged sections and round bars. Unfortunately, these old buildings were significantly lacking in earthquake resist-

271

272 Damage to steel buildings in the 1995 Hyogoken-Nanbu earthquake: M. Nakashima et al.

Figure 1 Damage to old steel buildings: a a collapse

ance because of the then premature seismic design and construction technologies and also because of material deterio- ration as shown in Figure lc. According to a preliminary estimate, over 70% of the damaged steel buildings located in Kobe City were of this construction type.

3. Ground motion recorded

Very large ground motions were recorded in the Hyogoken-Nanbu earthquake. However, the recorded data were scarce in the Shindo 7 region, i.e. the most severely shaken region, where the intensity of motions ranked approximately IX to XII on the modified Mercalli intensity scale. Figure2a shows the elastic pseudo-acceleration response spectra of the motions recorded at the Japan Meteorological Agency (JMA) at Kobe together with those of the Sylmar and Newhall records observed in the 1994 Northridge earthquake. [Note that their power appears to be similar.] However, the location of the JMA record, which was one of the largest ground motions recorded, was not in the midst of the most damaged area. This record was obtained near the edge of the Shindo 7 region and the building damage in this neighborhood was not extensive. Several attempts have been made to estimate the ground motion that must have occurred in the Shindo 7 region during the main shock. Figure 2b shows the one predicted in Hayashi and Kawase 3 (designated as Point-B-20) and it is for the center of downtown Kobe. Its elastic pseudo-acceleration response is significantly larger than those of the JMA Kobe record. The difference in damage potential between the two ground motions will be discussed later.

4. Design earthquake forces in Japan

The present Japanese seismic design code, adopted in 1981, provides two levels of design earthquake forces. The first level is for small to medium earthquakes with maximum ground accelerations ranging between 80 and 100 cm/s2; to ensure their serviceability, structural systems are required to remain elastic during such earthquakes. The second level is for large earthquakes with maximum ground acceler-

Figure 1 Damage to old steel buildings: b construction with light-gauged sections; c corroded sections

Damage to steel buildings in the 1995 Hyogoken-Nanbu earthquake: M. Nakashima et el. 2 7 3

Acceleration (g) JMA Kobe O~s)

............. JMA Kobe (EW~

. . . . . . . . S:Im~r (NS) . . . . . . . . . . . . . . . . . . . . . cwhaUft~).-.

|

0 1 2 3 Period (see)

(a) JIVLA Kobe Record

Acceleration (~)

0

0 1 2 3 Period (sec)

(b) Predicted Ground Motion (Ref.3 Point-B-20)

Figure2 Pseudo elastic acceleration response spectra of recorded and predicted ground motions: a JMA Kobe record; b predicted ground motion (~Point-B-20)

ations ranging approximately from 300 to 400 cm/s 2. For such large earthquakes some degree of damage (meaning plastic deformation in their members and connections) to structures is permitted. B~tsed upon these maximum ground accelerations, the Japanese seismic design code stipulates 0.2 and 1.0 g (g as the acceleration of gravity) of the mini- mum design base shear coefficient for the first level (serviceability) and second level (ultimate), respectively. Figure 2 clearly shows that response accelerations are significantly larger than 1.0 g for shorter periods, indicating that the motions were fitr beyond what have been considered in the present Japanese seismic design. In consideration of such large recorded ground motions, as well as the Japanese seismic design concept, it appears that not a few steel buildings stood in severely shaken regions might have to sustain significant structural damage even if they were built in conformance with the present design and construction practice.

5. O v e r v i e w o f d a m a g e to m o d e r n steel bu i ld ings

5.1. Share of steel buildings in Japan How popular is steel in Japanese building construction? Figure 3a shows the total floor area constructed each year with respect to other structural materials. Wood has been ranked first for years and it is used almost exclusively for houses. Steel has been ranked second, followed by reinforced concrete (RC) and steel-encased reinforced concrete (designated SRC in Japan). As such, steel is a very

Floor Area Constructed ( x 10,000m 2)

,0.o I ............... Z

F 5.ooo . . . . . . . . . . .,:-:

1960 1965 1970 1975 1980 1985 1989 Year of Construction

(a) Distribution of Floor Areas with Respect to Su'uctural Material

• >16F • 10-15F [] 6-9F [] 3-5F [] 2F • IF Floor Areas Constructed ( x 100,000 m 2) IA%

100% 337 380 414 445455 490 573 661738 752 733 689 6~. 1.5% - :

iii iiii

40% 50.9%

20% " 16.6%

0 % . . . . '----'- . . . . . '- - ' - - - : - - - - ; - - - - ' - - ' - - - - : - - - : - - ' -

Y e a r o f Construclion

~ ) DisSbution of Floor Areas ¢ t h Respect to Number of Stories

Figure 3 Market share of Japanese steel construction (floor areas constructed per year): a distribution of floor areas with respect to structural material; b distribution of floor areas with respect to number of stories

popular structural material in Japan. Figure 3b shows the total floor area of steel buildings constructed each year with respect to the number of stories, suggesting that the vast majority of steel buildings are shorter than five stories. In fact, most of steel buildings in Japan are low-rise, used for offices, shops, shops and residence combined, and plant structures.

5.2. Structural types, members and connections used in damaged buildings From the mid February to mid March, 1995, the Steel Com- mittee of the Kinki Branch of the Architectural Institute of Japan conducted a detailed survey into the damage to mod- ern steel buildings and located 988 steel buildings damaged 4. Among those buildings, 90 were rated as collapsed, 332 as severely damaged, 266 as moderately damaged and 300 with minor damage. [The ratios of these numbers are approximately 1:3:3:3.] Figure4 shows the number of buildings with respect to the damage level, indicating that most of the collapsed buildings are two to five stories tall and no building with seven stories or more collapsed.

The following summarizes structural types, cross-sections used for members and connection types employed in the buildings surveyed. Although the statistics given below were for the damaged buildings only, these statistics are believed to reflect those of Japanese modem steel buildings.


0 1 2 3 4 5 6 7 8 9

Number of Buildings 50 100 150 200 250 300 350 400

Table 2 Types of connections used in damaged steel buildings a columns

Type of connection Total

Weld 186 Bolt 19 unknown 514

b braces


Weld 43 Bolt 135 unknown 283

c beam-to-column connections


Figure 4

Number of Stories • Collapse I Severe [] Moderate [] Minor

Damage level with respect to number of stories 4

The damaged building types were classified as R-R (the unbraced frame in two horizontal directions), R-B (the unbraced frame in one horizontal direction and braced frame in the other direction), and B-B (the braced frame in two horizontal directions). The numbers of damaged buildings were 432 (R-R), 134 (R-B) and 34 (B-B), with 388 left unknown. These statistics indicated that the majority of damaged buildings were unbraced. Table 1 shows the cross-sectional types used for columns, for beams and for braces. For columns, wide-flange (H) sections were used most extensively, followed by square-tube sections. It is notable that, for the past 15 years, square- tube (commonly cold-formed) sections have been used more frequently. Beams consisted almost exclusively of wide-flange sections, either rolled or built-up and rods, angles, fiat bars, round-tubes, wide-flanges, square-tubes and channels were used for braces.

Field welding 40 Shop welding 271 Through diaphragm 144 Exterior diaphragm 6 Interior diaphragm 8 Stiffener plate 161 unknown 516

d column bases


Standard 270 Concrete Encased 70 Embedded 86 unknown 569

Table 1 Cross-sections used in damaged steel buildings a columns

Cross-section Total

[ ] (Cold-formed) 235 © (212) H 8 B-shaped 409 built-up 70 unknown 55

228

b beams


weld 12 bolt 397 unknown 457

c braces

Cross-section Total

rod 77 angle 44 flat plate 44 © 42 H 8 [ ] 6 channel 4 unknown 227

Table 2 shows the type of connection details in the damaged buildings; column-to-column splices are mostly accomplished by welding; beam-to-beam splices are almost always accomplished using high tension bolts; and braces are connected mostly by bolting, except for small rod and fiat bar braces, which are generally welded. Figure 5 shows three typical types of beam-to-column connections, namely the through-diaphragm connection, the interior diaphragm connection and the exterior diaphragm connection. Among these connections, the through-diaphragm connection is by far most popular as indicated in Table 2c. In the through- diaphragm connection, a long square-tube is cut into three pieces: one used for the column of the lower story, one for the connection's panel zone (a short piece, often called a

(a) Through- Co) Interior (c) Exterior Diaphragm D i a p h r a g m Diaphragm

Figure 5 Types of beam-to-column connections used in damaged steel buildings: a through-diaphragm; b interior diaphragm; c exterior diaphragm

Damage to steel buildings in the 1995 Hyogoken-Nanbu earthquake: M. Nakashima e t al .

(a) Base Plate Connection

(b) Concrete Encased Column Base

i Z:- U (c) Embedded

Column Base

Figure 6 Type of column base connections used in damaged steel buildings a base plate connection; b concrete encased column base; c embedded column base

275

Number of Buildings 160 200 o 40 80 120

R-R.I'q

R-R-H

R-a-r]

R-B-H

B-B-r]

B-~H

• Column [] Beam Ill Beam-to-Column [] Brace Connection

[ ] Column Base Plate

Figure 8 Damage to structural members with respect to structural type

dice in Japan), and one for the column of the upper story. Two diaphragm plates are inserted between the three separ- ated pieces and shop-welded all around. Then, the entire piece (often called a Cluistmas tree) is transported to the site and connected with the mid-portion of the beam by high tension bolts. Figure 6 shows three typical types of column base connections, including the standard base plate connection, the concrete encased column base connection, and the embedded column base connection. As evidenced in Table 2d, standard b~Lse plate connections were most commonly used.

5.3. Damage statistics Figure 7 shows the correlation between the damage level and structural type, which is further divided by the type of column used. This figure indicates that no significant differences existed in dmnage level with respect to the structural type (R-R, R-B and B-B) and that buildings with wide-flange columns suffered somewhat more serious damage as compared with other column types. The latter may be attributed to the building age because square-tube sections were used in mo~re recent construction. Figure 8 shows the location of d~trnage (columns, beams, beam-to- column connections, braces and column bases) as a func- tion of structural type. Major observation obtained from this figure is as follows: (1) columns in unbraced frames suffered the most damage relative to other frame elements (in terms of the number of buildings), while the braces in braced frames were the most frequently damaged structural element; (2) in unbraced frames, damage to beam-to-column connections and column bases was also siginficant; (3) damage to beam-to-column connections was most significant for unbraced frames having square-tube columns; and (4) damage to columns was most significant for unbraced frames having wide-flange columns.

5.4. Damage to members

5.4.1. Columns: Damage occurred in numerous columns, but most occurred in the vicinity of beam-to-column connections. Damage to columns themselves included plastification near column ends, excessive bending and local buckling, and fractures in base metals and column splices. In particular, many wide-flange columns sustained excessive bending about their weak-axis. Figure 9 shows a fracture at the base metal of a cold-formed square tube having dimensions of 450 (width) x 450 (depth) × 25 (thickness) mm. The fracture surface was mostly brittle, but significant local buckling appeared above the fractured section, suggesting that the column was loaded beyond its yield stress.

A cluster of modem high-rise residential steel buildings, located near the seashore, exhibited over 50 fractured columns and braces. The fractured columns were made of square tubes, with the depth of 500-550 mm and the thickness of 50-55 mm. Fracture occurred at base metals (Figure lOa), at welded column splices and at beam-to- brace connections (Figure lOb). In the fracture, surfaces were rather rough, involving shear lips and tear ridges, which confirmed that fracture was brittle involving a small amount of plastification. Extensive studies are underway to identify the causes of the fracture from the perspective of forces exerted on the columns, material properties, welding properties, loading-rate effects, temperature effects and others.

5.4.2. Braces: Damage to braces was found to be more severe in relatively smaller cross-sections (rods, angles and fiat plates). Although this was not quantified, the size of cross-sections was correlated strongly with the

0% 20% 40% 60% 80% 100%

Figure 7 tural type

• Collqace • Severe [ ] Moderate [ ] Minor

Distribution of damage level with respect to struc- Figure 9 Fracture at a cold-formed square tube column


Figure 11 Damage to brace connections: a breakage of bolts; b beam web buckling and out-of-plane deformation

Number of Buildings 0 20 40 60 80 100 120

(l)

Figure 10 Brittle fracture of square tube jumbo columns: a fracture at base metal; b fracture at brace-to-column connection

building age, with small cross-sections used more frequently in older buildings. Damage to braces with larger cross-sections was concentrated mostly on their connections with adjoining beams or columns. Figure 11 shows examples of such damage, in which connection bolts were broken (Figure l la) and a beam connected to a pair of braces sustained siginficant web-plate buckling and out-of- plane distortion (Figure llb). On the other hand, not a few braced frames built in severely shaken areas sustained minimal damage and in these cases the braces were firmly connected to the adjoining beams and columns. In summary, damage to braces having large cross-sections was strongly correlated with the connection details, and poor connection details suffered more severe damage.

5.4.3. Column bases: Most of the damage to column bases was observed for standard base plate connection type. Figure 12 shows the damage level with respect to the location of damage in the standard base plate connections, indicating that the majority of damage occurred at anchor bolts. In present seismic design practice, standard base plate

(2)

(3)

(1) Damage to Anchor Bo l t s (2) Failure o f Welds at Base Plates (3) E x c e s s i v e D e f o r m a t i o n o f Base Plates • Conapse II Severe ~ Moderate [] Minor

Figure 12 Damage location and level of standard column base connections

connections are designed with the pin-supported assumption, meaning no moment transfer at the column base. Regardless of the assumptions, the column bases must sustain shear forces securely under load reversals. The design profession now recognizes that the present practice is somewhat inadvertent and more careful design is required for such connections.

5.4.4. Beam-to-column connections: As stated earlier, many fractures were observed in beam-to-column welded connections. Fractures to beam-to-column connections were essentially divided into two types. Figure 13 shows the first type of fracture, which occurred in beam-to-column (shop-welded) connections where columns, beam and connection panels were fillet-welded of rather small sizes. At a glance, it was understood that such small welds could hardly transfer the stresses exerted to the connections. In fact, they fractured without any observed plastification in the neighboring columns and beams.

Damage to steel buildings in the 1995 Hyogoken-Nanbu earthquake: M. Nakashima et al. 277

Figure 13 Fracture at beam-to-column connections with fi l let weld ing of small sizes: a fracture at column top; b fracture at beam end

The second type of fracture was observed in many beam- to-column connections with the full-penetration welding. Fractures were mostly brittle and occurred at weld metals, heat-affected zones, base metals (initiated from the toe of weld access holes), and diaphragm plates (Figure 14). In many instances of such fractures, the following obser- vations were also disclosed: ( 1 ) residual story drift was not significant; (2) damage to interior and exterior finishes were minimal; (3) fractures occurred mostly at beam bottom flanges only; (4) significant yielding, plastification and local buckling of beam bottom flanges were observed, indicating that the beams dissipated some energy before fracture; and (5) such plastification occurred only in the beams while the adjoining colurans remained almost elastic. The reason for observation (5) may be that many designers adopted weak-beam concepts, or that real yield strengths of square tube columns were significantly higher than their nominal values due to cold-forming in the making of square tube sections. Figure 15 shows a set of results of tests conducted for the base metal of a fractured beam. The Chalpy V-notch test shows that the material had more than 50 Joule of energy for the temperature at 0°C and the base metal in the vicinity of the fractured surface was significantly hard- ened. A rough correlation between the hardness and strain indicated that the base metal might have sustained over 15% of cumulative plastic strain before fracture.

6. Potential causes of damage to beam-to- column connections

The damage to steel beam-to-column welded connections has created one of the most serious problems that have exposed following the Hyogoken-Nanbu Earthquake. Speculation abounds about possible causes of the damage,

Figure 14 Fracture at beam-to-column connections with full penetration groove welding: a fracture at base metal init iated from toe of weld access hole; b fracture involving yielding and local buckling

200

6 1 5 0 L .

.=

100

L .

.~ s o <

250 _=

;>

¢= 2001 I , =

¢

I , , =

.~. 150

i i i

• I 1 4 W [ _+_ - - - 0 214W . . . . . . . . . . . . . . . -,-

i o

' e i i

- . + = - . . . . . . . ~. . . . . . . . ~- . . . . . . . - t . . . . . . . - t - - -

, i

e '

. . . . . . . . . . O - . . . . . ° . . . . . . . . . . . . . . . . . .

i

-40 40 -20 0 20 Temperature (c)

(a) Chalpy V-Notch Test Weld Metal

. , z Base Matal

o

l e ',,',,w I

. . . . . . . . . . . . . . . . . . . . .

- ' q l

J I I I 0 50 100 150 200 250

Distance from Fracture Surface (mm) Co) Hardness Test

Figure 15 Material propert ies of a base metal near a fractured surface: a Chalpy V-notch test; b hardness test


and various research efforts are underway to locate the true reasons. The following summarizes possible causes that were under consideration.

6.1. Materials Japan uses both the integrated casting and scrap-based (mini-mill) steels for beam members. The market share of scrap-based steels is higher for smaller, thinner sections (commonly 450 mm deep and 40 mm thick or smaller). Some concerns have been raised about the fracture toughness of such steels particualry at the flange-web junction, but no plans were under consideration for restricting the use of the steels. Futher, it is notable that large construction companies establish in-house regulations for quality assur- ance for their use of continuous casting steels.

Japan is in a process of introducing new types of struc- turai steels, called SN steels, lit should be emphasized that this action initiated long before the Hyogoken-Nanbu earthquake.] Both upper and lower limits are specified for the yield and maximum strengths, and an upper bound of 0.8 is specified for the ratio of the yield to the maximum strength. These steels have already been commercially available and, particularly after the Hyogoken-Nanbu earthquake, designers have been more encouraged about the use of these steels.

6.2. Connection details Present Japanese beam-to-column connections are charac- terized as follows: (1) square tube (often cold-formed) sections are used for columns in recent construction; (2) the through-diaphragm connection (Figure5a) is by far the most popular; and (3) in this type of connections welding is accomplished in shop, and welding robots have been introduced lately for automatic welding at some loations.

A constructor conducted a detailed survey on several damaged steel buildings in which shop-welded through-diaphragm connections were used. Out of 2396 connections surveyed, 79 were found to have damaged including complete fractures and partial cracks. The damage percentage is 3.3%. Figure 16 shows the fracture and crack distribution with respect to the damage location, indicating that 20.5% of the damaged connections were fractured at the base

40 35 30 25 20 15 10 5 0

(%)

(1) (2) (3) (4) (5) (6)

(1) Complete Fracture at Base Metal (2) Complete Fracture at Weld Metal (3) Crack at Crater (4) Crack from Run-off Tab (5) Crack to Web Initiated from Weld Access Hole (6) Cra.ek to Diaphragm Plate Initiated from Weld

Figure 16 Distribution of damage to beam-to-column connections with respect to type and location (79 connections surveyed)

metal. In these instances, the fracture initiated from the toe of the weld access hole. It is speculated that this type of fracture was attributed primarily to a combined effect of stress concentration at the toe and low fracture toughness of the material at the web-flange junction. After the Hyogoken-Nanbu earthquake, extensive studies have been undertaken to avoid this type of fracture and various modified connection details have been proposed. Most of them attempted to reduce the size of weld access holes, aiming to mitigate the stress concentration at the toe.

In the through-diaphragm connection, the beam web is shop-welded directly to the column flange, but normally the column is supplied without any vertical stiffener or diaphragm at the back of the beam web (Figure 17a). Because of the flexibility of the column flange in out-of-plane deformation, the moment resisted by the beam web was reduced, causing stress concentration in beam flanges. Concerns were raised about this issue after the earthquake and efforts are underway to quantify the degree of reduction in the web's moment reistance.

As stated earlier, most of the fracture in beam-to-column connections occurred at beam bottom flanges only, which is very similar to what had been observed from the 1994 US Northridge earthquake. In the US, the coincidence of the weld root, the backing bar, and the most stressed fiber has been said to be a primary source of bottom flange fractures (Figures 17b). On the other hand, Japanese shop welding enables welding of the beam bottom flange from the opposite side, as shown in Figure 17a, where the weld root is located on the interior side of the bottom flange. The observation that many Japanese shop welded beam-to- column connections also fractured only at bottom flanges indicates that the root location is not the sole cause of bottom flange fractures. Composite action with floor slabs, another source being addressed, remains suspect because in the Hyogoken-Nanbu earthquake some beam-to-column connections sustained fractures in both the top and bottom flanges when floor slabs were not present.

6.3. Welding In recent construction in Japan, semi-automatic Co2- (or sometimes argon) shielded welding has been used almost exclusively for the welding of beam flanges. This practice has been common not only in shop welding, but in field welding as well. In early 1970s, self-shielded flux-cored electrodes were introduced in Japan, but because of low toughness observed in the weld metals its use faded quick- ly.

d

(a) Tl~ough-Diapl~gm (b) Web-Bolted Flange-Welded Connoetion Connection

Figure 17 Details of beam-to-column connections: a through- diaphragm connection; b web-bolted flange-welded connection

Damage to steel buiMings in the 1995 Hyogoken-Nanbu earthquake: M. Nakashima et al. 2 7 9

B4 B4 B4 B4 B4

(24 B4C4 B4(24 B4C4 B4C4 B4C4 3.8

(24 B4(24 B4C4 B4C4 B4C4 B4C4 3.8

(24 B3 C4 B3(24 B3C4[ B3C4 B3C4 3.8

(24 B3 C4 B3C4 B3C4 B3C4 B3C4 3.8

C3 B2 C3 B2 C3 B2 C3 B2C3 B2C2 3.8

C1 BI C'2 BI C2 BIC2 B1C2 B1CI 3.8

e l 4.2

I 8.1 I 7.2 I 7.2 I 7.2 16.7551

Figure 18 Seven story, five bay steel unbraced frame analyzed

As shown in Figure 1,5, 24.4% of the damaged connections had complete fracture along the weld metal, 10.3% had cracks at craters and 37.2% had cracks initiated from run-off tabs. This large percentage of damage associated with welding was striking, and serious concerns were raised about the present weldLag practice. Higher voltages and larger deposition rates than those stipulated in regulations and excessive weaving are thought to be the likely causes.

6.4. Plastic rotations exerted on beam-to-column connections Immediately after the Hyogoken-Nanbu earthquake, two questions arose: (1) when beam-to-column connections (with what plastic rotations) fractured; and (2) how much plastic rotation capacities would have been needed for them not to fail during the earthquake? The answer to these questions is yet very difficult. Unfortunately, no steel buildings in the Kobe area was heavily instrumented, or no ground motions was recorded in the most severely damaged areas. It is also difficult to estinaate the degree of plastic rotations for the damaged beam-to-column connections.

The following data were obtained from a preliminary inelastic dynamic response analysis conducted by the first writer for the purpose of estimating the degree of plastic rotations that damaged beam-to-column connections might have sustained in this earthquake. The building analyzed was a seven story, five bay steel moment frame whose major structural properties are shown in Figure 18 and Table 3 and simple plastic hinge method was employed to analyze the building. One spring was inserted at each end of a member and it was assumed that the hinge behaves rigid-elastically. Unlike the conventional plastic hinge, a

limit plastic rotation was specified to simulate the fracture at the bottom flange. Once the plastic rotation reached the specified limit plastic rotation, the hinge stiffness degrades and the hinge behaves as a slip model for succeeding loading. The building was analyzed for its dynamic response by the direct integration method and two ground motion records were used as input motion; namely the JMA Kobe record and the predicted record (given in Hayashi and Kawase 3 and designated as Point-B-20), both introduced earlier. The building had 1.05 s for the elastic fundamental natural period and a base shear coefficient of 0.41 for the ultimate strength. The building was hypothetical, but its dimensions closely followed a steel building constructed in Kobe. For each ground motion, two limit plastic rotations (same values for all beam members) were assigned: i.e. infinite and 0.004 rad. The former was adopted to see how much plastic rotations would be needed if the building responded without beam fracture, and the latter represented a rather small plastic rotation capacity.

The results obtained are summarized in Table 4 and two displacement response histories and two story shear versus story displacement relationships are shown in Figure 19. A summary of observation was as follows:

(1) The total energy input to the building (with the assumption of no fracture) was 2.34m/s and 3.87 m/sec in terms of the equivalent velocity for JMA Kobe and Point-B-20 records, respectively, and the maximum plastic rotation achieved is about twice for Point-B-20 than for JMA Kobe. These values indicate a significant difference in the damage potential between the two records.

(2) The maximum plastic rotation required if the beam members would not fail was 0.014 rad. for JMA Kobe record. This value of plastic rotation was considered achievable for beam-to-column connections designed with the present practice 5. On the other hand, 0.029 rad. of plastic rotation is required for Point-B- 20, which may not be easy to achieve even with the modem practice.

(3) Early beam fractures indeed increase maximum story drifts, maximum plastic rotations and particularly maximum cumulative plastic rotations. The reduction in stiffness caused by fractures is also evident and the period of vibration is enlarged significantly (from 2.5 to 4 times the period of vibration that is achieved with no fracture assumed).

The above observation was very rudimentary since it was based upon the analysis of only one fictitious building. Fur- thermore, numerous simplifications were made in the analy-

Table 3 Member properties of seven story, five bay steel unbraced frame

Member ID Dimension Cross-sectional Moment of inertia Plastic moment number (mm) area (ram 2) (mm 4) (kN × ram)

C1 BOX: 550x550x22 65 050 2 527 000 000 3 016 000 C2 BOX: 550x550×19 59 150 2 281 000 000 2 629 000 C3 BOX: 500×500x19 50 790 1 665 000 000 2 160 000 C4 BOX: 450x450x14 45 580 1 477 000 000 1 838 000 B1 H: 700x300×14x25 24 100 2 030 000 000 1 539 000 B2 H: 700×300x14x22 22 380 1 847 000 000 1 406 000 B3 H: 700×250x12x22 18 870 1 547 000 000 1 181 000 B4 H: 700×200x12x22 16 670 1 294 000 000 1 005 000


Table 4 Results of dynamic response analysis of seven story, five bay steel unbraced frame

Ground motion Limited plastic rotation specified Vt (m/s) Vp (m/s) Max. story drift angle Max. residual story drift angle Max. plastic rotation (rad.) Max. cumulative plastic rotation (rad.)

J MA-Kobe J MA-Kobe Point-B-20 Point-B-20 Infinite 0.004 rad. Infinite 0.004 rad. 2.34 2.46 3.87 3.15 1.92 1.83 3.29 2.51 1/50.8 1/31.4 1/27.9 1/20.6 1/189 1/327 1/388 1/46.9 0.0142 0.0359 0.029 0.0546 0.0785 0.497 0.255 0.574

Disp. (m) 0.6 [ Point-B-20 I

0 " 4 ~ 2 0 0.2 0 -0.2 -0.4

-0.6 [ Time (sec)

(a) Time History (7th story) (no fracture) Story Shear (MN)

I Point.B.20 [ 6

-0.15 5 -0.1

-~ [ Story Disp. (m) -6

Co) Story Shear vs. Stow Disp. Relationship (4th Story) (no fracture)

Disp. (m) [ Point.B.20l

0.9 ~ 2 0 0.6 0.3

0

-0.3 | w Time (sec) -0.6 [ -0.9

(c) Time History (7th story) (early fracture)

[Point-B-2{~ 6 Story Shear (MN)

. ~ 0.2

-6 Story Disp. (m)

(d) Stow Shear vs. Stow Disp. Relationship (5th Stow) (early fracture)

Figure 19 Examples of response time histories and story shear vs story displacement relationships obtained from analysis

sis, including the rigid base assumption, neglect of P-A effects, neglect of floor slabs, neglect of interaction between the frame and attached interior and exterior finishes, neglect of three-dimensional effects, etc. Therefore, the comment given here serves just as a clue for the plastic rotations that beam-to-column connections in steel buildings built in severely shaken regions might have experienced. Various research efforts have been undertaken to predict the responses of damaged steel buildings, but as mentioned previously, with no strong ground motion record or data on the actual building response available, estimation of plastic rotations of beam-to-column connections will remain difficult.

7. Classification of damage with respect to previous awareness and difficulty in providing solutions In the previous section, the damage to steel buildings was described with respect to the structural components. The damage may also be classified in light of the previous awareness and the difficulty of providing solutions. TYPE- I is the damage that can be described as, "We knew they were seismically vulnerable; but we did not take action, and they failed." The extensive damage to old steel buildings, mentioned earlier, falls into this category. TYPE II is the damage that can be described as, "We did not know they were so bad (because their weaknesses were hidden); upon seeing the damage, we could locate reasons." Fractures at beam-to-column connections with fillet welding of small sizes are typical examples of this type of damage. TYPE III is the damage that can be described as "They were designed and constructed according to present parctice; nevertheless they failed." Here, our present design and construction practice is called into question and action by reasearchers and engineers has to be initiated. TYPE III may be classified further into TYPE Ill-a, stated as "After examining the damage, we located the true reasons and now know how to improve", and TYPE III-b stated as, "After one year, we still debate." Damage to column base connections and brace-to-column and brace-to-beam connections may be categorized as TYPE Ill-a, whereas fractures of jumbo columns and cold-formed square tube columns and fractures of beam-to-column connections are TYPE III-b.

Then, how do we try to solve these types of problems? For Type I: old steel buildings, retrofitting and redevelop- ment is no doubt the ultimate solution, but it is far beyond a technical issue, rather a serious problem that requires socio- economic considerations. For Type II: quality oriented problem, for many years the engineering community has provided numerous manuals, implemented quality assur- ance programs, organized seminars for training, etc. It should not be overlooked, however, that quality is after all

Damage to steel buildings in the 1995 Hyogoken-Nanbu earthquake: M. Nakashima et al. 281

in the hands of those who really touch upon the building and evenutally we have to rely on their morale and con- science. Just asking them such emotional acts is not fair, either. The entire construction industry has to show its respect to these workers ~ad it can be done through reason- able rewards. The Japanese construction industry is organized with a multiple stratum system; that is, large gen- eral contractors at the top, followed by sub contractors, sub- sub contractors, etc. Proper distribution of capital among these sectors is a must for good quality control, but the reality is now not necessarily this way. For Type Ill-a, the engineering community has already enforced new, modified design and construction procedures, and various research efforts have begun for Type III-b.

8. Conclusions

A summary of this paper is as follows:

(1) Damage to steel buildings observed in the 1995 Hyogoken-Nanbu earthquake was reviewed. Extensive damage to old steel buildings that had not conformed to the present seismic performance requirements was reported.

(2) Columns, braces, column bases and beam-to-column connections were identified as typical locations of damage, and representative damage patterns were discussed.

(3) Damage to beam-to-column connections was detailed.

(4)

Two types of damage: one fracture of connections with fillet welding of small sizes and the other fracture of connections with full penetration groove welding, were identified, with the latter case involving significant yielding and local buckling. Plausible causes of the damage were discussed from the persepectives of materials, welding, connections details and plastic rotation demands. Damage to steel buildings was classified with respect to the previous awareness and the difficulty in providing solutions.

References 1 'Preliminary Reconnaissance Report of the 1995 Hyogoken-Nanbu

Earthquake,' The Architectural Institute of Japan, March 1995, 197pp. (in Japanese)

2 'English Edition of Preliminary Reconnaissance Report of the 1995 Hyogoken-Nanbu Earthquake,' Nakashima, M. and Bruneau, M. (eds) The Architectural Institute of Japan, April 1995, 215pp.

3 Hayashi, Y. and Kawase, H. 'Strong motion evaluation in Chuo Ward, Kobe, during the Hyogoken-Nanbu earthquake of 1995,' J. Struct. Construct. Engng, the Architectural Institute of Japan, 1996, (481), pp. 37-46 (in Japanese)

4 'Reconnaissance Report on Damage to Steel Building Structures Observed from the 1995 Hyogoken-Nanbu earthquake,' Steel Com- mittee of the Kinki Branch of the Architectural Institute of Japan, May 1995, 167pp. [in Japanese, attached with Abridged English Ver- sion, Nakashima, M. (ed.)]

5 'Report by Weld Access Hole Research Sub-Committee,' Weld Access Hole Research Sub-Committee, Kozai Club, Japan, October, 1993, 213pp. (in Japanese)

本文献由“学霸图书馆-文献云下载”收集自网络，仅供学习交流使用。

学霸图书馆（www.xuebalib.com）是一个“整合众多图书馆数据库资源，

提供一站式文献检索和下载服务”的24 小时在线不限IP

图书馆。

图书馆致力于便利、促进学习与科研，提供最强文献下载服务。

图书馆导航：

图书馆首页文献云下载图书馆入口外文数据库大全疑难文献辅助工具

http://www.xuebalib.com/cloud/

http://www.xuebalib.com/

http://www.xuebalib.com/cloud/


http://www.xuebalib.com/vip.html

http://www.xuebalib.com/db.php

http://www.xuebalib.com/zixun/2014-08-15/44.html


classification of damage to steel buildings observed in the...

Documents