seismic performance of elevated steel silos during van earthquake, october 23, 2011

23
ORIGINAL PAPER Seismic performance of elevated steel silos during Van earthquake, October 23, 2011 Eren Uckan Bulent Akbas Jay Shen Rou Wen Kenan Turandar Mustafa Erdik Received: 23 December 2013 / Accepted: 4 July 2014 Ó Springer Science+Business Media Dordrecht 2014 Abstract Silos are commonly used industrial structures to store dry/granular materials like cement or wheat. A typical silo consists of a vertical steel tank supported by a braced steel frame which rests on concrete support. Due to unloading purposes, the tank is gen- erally located at an elevated position. This makes the structure vulnerable to axial loads in columns due to excessive overturning moments generated at the base of the structure. During the October 23, 2011 Van earthquake in Turkey, many silos collapsed either due to column buckling or foundation problems. In this paper, the field observations regarding the seismic performance of silos after the Van earthquake are first summarized. Then, the seismic performances of two steel-elevated silos located in the earthquake region are studied. One of the silos survived the earthquake by some minor damages in the form of buckling (at bottom horizontal brace) and spalling of concrete support, while the other silo remained undamaged. Nonlinear dynamic time history analyses are performed to evaluate the seismic performances of both silos. As the input ground motion, the recorded ground motion from a temporary aftershock station (about 2 km away from the silos) in the second earthquake is used. Analyses indicate that design and construction quality of elevated silos determine the seismic performance. Finally, recommendations are given to improve the seismic performance of new constructions. E. Uckan (&) Á K. Turandar Á M. Erdik Kandilli Observatory and Earthquake Research Institute (KOERI), Bogazici University, Cengelkoy, Istanbul 34684, Turkey e-mail: [email protected] B. Akbas Department of Earthquake and Structural Engineering, Gebze Institute of Technology, Kocaeli, Turkey J. Shen Department of Civil, Construction and Environmental Engineering, Iowa State University, Ames, IA, USA R. Wen Sharma & Associates, Inc., Countryside, IL, USA 123 Nat Hazards DOI 10.1007/s11069-014-1319-9 Author's personal copy

Upload: mfhfhf

Post on 17-Sep-2015

25 views

Category:

Documents


6 download

DESCRIPTION

بلاىيفتثقر

TRANSCRIPT

  • ORIGINAL PAPER

    Seismic performance of elevated steel silos during Van

    earthquake, October 23, 2011

    Eren Uckan Bulent Akbas Jay Shen Rou Wen Kenan Turandar

    Mustafa Erdik

    Received: 23 December 2013 / Accepted: 4 July 2014

    Springer Science+Business Media Dordrecht 2014

    Abstract Silos are commonly used industrial structures to store dry/granular materials

    like cement or wheat. A typical silo consists of a vertical steel tank supported by a braced

    steel frame which rests on concrete support. Due to unloading purposes, the tank is gen-

    erally located at an elevated position. This makes the structure vulnerable to axial loads in

    columns due to excessive overturning moments generated at the base of the structure.

    During the October 23, 2011 Van earthquake in Turkey, many silos collapsed either due to

    column buckling or foundation problems. In this paper, the field observations regarding the

    seismic performance of silos after the Van earthquake are first summarized. Then, the

    seismic performances of two steel-elevated silos located in the earthquake region are

    studied. One of the silos survived the earthquake by some minor damages in the form of

    buckling (at bottom horizontal brace) and spalling of concrete support, while the other silo

    remained undamaged. Nonlinear dynamic time history analyses are performed to evaluate

    the seismic performances of both silos. As the input ground motion, the recorded ground

    motion from a temporary aftershock station (about 2 km away from the silos) in the second

    earthquake is used. Analyses indicate that design and construction quality of elevated silos

    determine the seismic performance. Finally, recommendations are given to improve the

    seismic performance of new constructions.

    E. Uckan (&) K. Turandar M. Erdik

    Kandilli Observatory and Earthquake Research Institute (KOERI), Bogazici University, Cengelkoy,

    Istanbul 34684, Turkey

    e-mail: [email protected]

    B. Akbas

    Department of Earthquake and Structural Engineering, Gebze Institute of Technology, Kocaeli, Turkey

    J. Shen

    Department of Civil, Construction and Environmental Engineering, Iowa State University, Ames, IA,

    USA

    R. Wen

    Sharma & Associates, Inc., Countryside, IL, USA

    123

    Nat Hazards

    DOI 10.1007/s11069-014-1319-9

    Author's personal copy

  • Keywords Elevated silos Non-building structures Van earthquake

    1 Introduction

    Elevated silos are one of the most widely used non-building structures in industry to store

    dry materials such as wheat or cement. The center of gravity of these structures is generally

    high which makes the structure extremely vulnerable to ground shaking due to the

    increased bending moments and axial loads. Collapse of a silo may cause loss of material,

    cleaning up of the contamination, replacement costs, and most importantly possible injury

    or loss of life. Therefore, they need to be designed to redistribute the forces within the

    structure when a local failure occurs. This will also help decrease the insurance premiums

    of the companies.

    The studies on the seismic performance of silos have gained great attention in recent

    years. Silvestri et al. (2012) investigated the lateral loads induced by effective mass of the

    grain on ground-supported (flat-bottom) silos in detail. Silo failures were widely observed in

    the past earthquakes. In a more recent study, they also carried out shaking table tests of

    ground-supported (flat-bottom) silos (Silvestri et al. 2014). In the December 7, 1988 Spitak,

    Armenia earthquake, a part of the granary in the flour mill complex was damaged (Dogangun

    et al. 2009). In the August 17, 1999 and November 12, 1999 Duzce, Turkey earthquakes at

    the SEKA Paper Factory, three reinforced concrete silos containing wastewater completely

    collapsed (Erdik and Uckan 2013). In the September 21, M = 7.6, 1999 ChiChi, Taiwan

    earthquake, a concrete factory silo collapsed. In May 21, 2003 Zemmouri,M = 6.8, Algeria

    earthquake, some storage facilities and equipment were severely damaged. In the recent

    M = 7.2 Van Earthquake, 2011, many small-to-medium size companies suffered from

    similar problems and also continuing legal issues between the silo manufacturers and plant

    owners (Akbas and Uckan 2012; Uckan 2013). In addition to such studies, vibration control

    methods in industrial facilities were also studied by Paolacci et al. (2013). This paper deals

    with the performance of the silo structures in the Van earthquake.

    On October 23, 2011, 1:41:21 p.m. local time, the city of Van in eastern Turkey was hit

    by a major earthquake causing heavy loss of human lives and properties. The epicenter of

    the earthquake, Tabanli Village, was 30 km north of Van (Fig. 1) in the eastern region of

    Turkey by Lake Van (Erdik et al. 2012).

    The earthquake is considered to be a very strong damaging earthquake with a magnitude

    of Mw = 7.2 and a depth of 19.2 km. It was associated with a reverse faulting mechanism

    dipping toward north (Erdik et al. 2012; Atzori et al. 2011; METU/EERC 2011; EERI

    2012), caused significant damage in the cities of Van, Ercis as well as in many villages,

    caused deaths of nearly 600 people, and injured 2,000, and about 15,000 buildings were

    damaged in the region (Bayrak et al. 2013; Taskin et al. 2013). About 58 buildings are

    reported to have totally collapsed, 52 of them are located in Ercis, while 6 of them are

    located in Van. Earthquake-affected areas are mainly located within a circle of 30 km

    radius and around the epicenter (Fig. 1). Peak ground acceleration (PGA) of the main

    shock, measured at the Muradiye station, was 0.18 and 0.17 g in NS and EW directions,

    respectively, and 0.08 g in vertical direction (Table 1) (AFAD 2011). The number of

    aftershocks with M = 4.04.9 reached up to 114, the biggest number of aftershocks ever

    recorded after any major earthquake in Turkey (Fig. 1) (Bayrak et al. 2013). Significant

    aftershock has been associated with this earthquake.

    Nat Hazards

    123

    Author's personal copy

  • Two weeks after the main shock, another earthquake with a moderate magnitude

    (M = 5.6) hit the town of Edremit, located 10 km south of Van. It was possibly triggered

    by the first event and caused by a strike-slip fault. About 30 buildings already damaged in

    the main shock collapsed. The water and electric networks were damaged again. No

    damage was observed in Ercis.

    Many reconnaissance reports and papers have been published after the earthquake, but

    none of them specifically investigated the damage to industrial facilities, especially to silos

    Location of the silo (red) and the aftershock station of KOERI (white)

    Strong motion station (white) (AFAD)

    Fig. 1 Main shock, big aftershocks, and the second (Edremit) earthquake [red stars: M C 5.0, pink circles:

    4.0 B MB5.0, yellow stars: the main shock (M = 7.2) and Edremit Earthquake (M = 5.6)] (Kalafat et al.

    2012)

    Table 1 Characteristics of the Muradiye and Semsibey earthquake records

    Date Record name Location Depth

    (km)

    Magnitude

    (Mw)

    PGA

    (g)

    23 October 2011 Muradiye NS

    (main shock)

    Van Muradiye

    Meteorology Directorate

    19.02 7.2 0.179

    9 November 2011 Semsibey (second

    earthquake)

    Semsibey Primary

    School

    5.6 0.09

    Nat Hazards

    123

    Author's personal copy

  • (Bayrak et al. 2013; Ozden et al. 2011; Taskin et al. 2013; Onen et al. 2012; Guney 2012).

    This paper aims at investigating the damage at silos which collapsed during the main

    shock. The seismic performances of two steel-elevated silos located in the earthquake

    region with concentrically braced frames are studied. One of them survived the earthquake

    by some minor damages in the form of buckling (at bracings) or spalling of concrete due to

    insufficient seating widths of the supporting concrete, while the other silo remained

    undamaged. Nonlinear dynamic time history analyses are performed to evaluate the

    seismic performances of the silos. As the input ground motion, the acceleration records

    from the nearest station (Semsibey) are used. The ground motion is scaled by a factor of 2

    in order to match the calculated bias adjusted PGA value of PGA = 0.2 g.

    2 Seismicity, site observations, and strong motion properties of the Van earthquake

    Seismicity of the region mainly consists of continent-to-continent collision of Anatolian

    and Arabian plates (Fig. 2). The earthquake is reported to have occurred on a blind oblique

    trust fault, namely Van fault, oriented toward the NESW direction (Erdik et al. 2012;

    KOERI 2011; Bayrak et al. 2013) and is not marked on the active fault map of Turkey

    (Emre et al. 2012).

    2.1 Seismicity and site observations

    Part of the convergence between these two plates takes place along the BitlisZagros fold

    and thrust belt at a rate of approximately 2.4 cm/year (Erdik et al. 2012). The focal region

    Fig. 2 Active tectonic plates, major faults, and their movement directions (Gulen et al. 2002) [from KOERI

    Report (KOERI 2011)]

    Nat Hazards

    123

    Author's personal copy

  • of the earthquake and much of easternmost Turkey lie toward the southern boundary of the

    complex zone of continental collision between the Arabian Plate and the Eurasian Plate.

    The geology of the surrounding area is composed of lake, river, and land sediments having

    layers of loose sand, gravel, and clay (Ozden et al. 2011). Groundwater level is known to

    be high especially for the areas close to Lake Van.

    Geotechnical observations after the earthquake indicated that there were mainly liq-

    uefaction and liquefaction-based lateral spreading and displacement, landslides, slope

    failures, and rockfalls. Water transmission/distribution systems suffered from ground

    shaking and soil failure problems. In Ercis, water supply system has been interrupted for a

    couple of days due to pipe breaks in the main water supply system. Local authorities

    indicated that as a result of pressure reduction in the water supply system, it was likely that

    it be polluted by the neighboring sewage system due to potential fluid flow. Due to tensile

    and compressive soil deformations, pipe crashing and pullout damages were also observed.

    However, after minor repairs and reconfigurations, all services were fully functional

    (Akbas and Uckan 2012). The heavy precast concrete frames with precast roof beams

    suffered from connection problems. Noticeable displacements and deformations were

    observed at the beamcolumn connections and footings of the structures with precast

    concrete frame systems located around the VanErcis highway. More severe cases were

    observed at the industrial zone of Van. The precast concrete beams slipped off from their

    seats and collapsed because of inadequate steelconcrete bondage.

    2.2 Strong ground motion

    In the main shock, the strong motion stations in Van were not operating. Nearest station

    was the one in Muradiye (Fig. 1). Following the preliminary field investigation conducted

    by KOERI (2011) on October 24, 2011, the deployment of 10 additional seismic instru-

    mentation began on October 25, 2011. Buildings with minor damage seemed to have

    experienced only non-structural damage due to mostly separation of infill walls from the

    structural frame. This type of damage was also a manifestation of long duration and rather

    low amplitude (believed to be\0.2 g).

    Distribution of PGA associated with the main shock was obtained by KOERI (2011) by

    utilizing the software ELER. The unadjusted estimates of a PGA on the order of 0.6 g and

    peak ground velocity (PGV) of 50 cm/s may have been experienced in the epicentral

    region. However, there was no evidence from the field to support these levels of ground

    motion.

    The bias adjusted ground hazard in Van was estimated by (Erdik et al. 2012) from the

    nearest stations in Muradiye and Bitlis. The distribution of revised PGAs is shown in

    Fig. 3 where the PGA is only about 0.2 g.

    The damage observations in Van and Ercis relate to a building stock that has been

    exposed only to about 50 % of the reference (or design) acceleration of 0.30.4 g level. As

    such, any nearby aftershock (M C 5.5) that can create ground motion above 0.150.2 g

    level can cause additional damage and collapses as evidenced in Van city by the November

    9, 2011, M5.7 Edremit earthquake (Erdik et al. 2012).

    3 Earthquake damage

    Almost all elevated cement and wheat silos that were fully loaded during the earthquake

    collapsed or seriously damaged. Some suffered from rupture at their base due to excessive

    Nat Hazards

    123

    Author's personal copy

  • bending and bearing stresses. Local buckling and anchorage failures were also observed.

    The tall twin silos collided with each other. Figs. 4 through 11 show some of the heavily

    damaged or fully collapsed silos during the earthquake. Figs 4 and 5a show collapsed

    cement silos in Van Industrial Zone due to supporting concrete failure.

    These silos were immediately retrofitted with tie beams after the earthquake. The

    buckling load for the gusset plate at the connection (Fig. 5b) is estimated as 190.0 kN for

    the two gusset plates with dimensions of 20 mm 9 50 mm 9 200 mm and with yield

    stress, Fy, of 250 MPa. A simple analytical study has shown that the axial load to be

    transferred by the connection was on the order of 210 kN during the earthquake. Thus,

    buckling is the expected behavior for this connection. The deformed shape in Fig. 5b

    indicates buckling due to uniaxial compression at the two parallel vertical gusset plates and

    bending deformation at the base plate (horizontal) due to axial load in the silo leg. The

    cement silos near Ercis in Figs. 6 and 7 are other examples of supporting concrete failure.

    A number of elevated wheat silos collapsed due to inadequate design in Van Industrial

    Fig. 3 Distribution of the bias adjusted PGA (%) associated with the main shock obtained by the ELER

    software (PGA = 0.2 g) (Erdik et al. 2012)

    Fig. 4 Construction of a tie beam for the collapsed cement silo in Van Industrial Zone due to supporting

    concrete failure

    Nat Hazards

    123

    Author's personal copy

  • Zone (Fig. 8). Brittle welded connection failure caused failure of some silos in Van

    (Fig. 9). A minor shell buckling is observed at an elevated gravel silo in Van (Fig. 10).

    Fig. 11 shows an undamaged ground-supported and a collapsed elevated wheat silo in the

    industrial zone of Van.

    a b

    Silo leg

    Silo body

    Fig. 5 A collapsed cement silo in Van: a construction of tie beams for the damaged supporting concrete

    foundation, b gusset plate buckling due to inadequate design at the leg-to-silo body connection

    Fig. 6 A collapsed cement silo near Ercis due to failure of the supporting concrete

    Nat Hazards

    123

    personal copy

  • Fig. 7 A collapsed silo due to supporting concrete failure in Ercis

    Fig. 8 Collapsed elevated wheat silos due to inadequate design in Van Industrial Zone

    Fig. 9 A collapsed silo due to brittle welded connection failure in Van

    Nat Hazards

    123

    Author's personal copy

  • 4 Structural framing and design of non-building structures in ASCE 7-10 (2010)

    For non-building structures not similar to buildings (elevated tanks, vessels, bins, flat-

    bottom ground-supported tanks, concrete chimneys, cooling towers, etc.), a structural

    framing system is selected and assigned appropriate values of seismic load reduction

    coefficient (R), over strength factor (X0), and displacement amplification factor (Cd)

    (Table 15.42 in ASCE 7-10 2010). Seismic response of a building-like or a non-building

    structure is most significantly affected by the following parameters:

    a. Earthquake input: location and soil condition dependent. No difference for any

    structure.

    b. Initial structural dynamic properties: initial stiffness and damping.

    a b

    Fig. 10 A minor shell buckling at an elevated gravel silo in Van

    Fig. 11 An undamaged ground-supported (left) and collapsed elevated wheat silo in Van Industrial Zone

    Nat Hazards

    123

    Author's personal copy

  • c. Structural configurations: plan and elevation configurations.

    d. Seismic load resisting system (SLRS).

    e. Interaction between structural and non-structural components.

    Concentrically braced frames (CBFs) are probably one of the most efficient seismic load

    resisting systems (SLRSs) for building-like industrial structures. It has certainly become

    the most popular SLRS in buildings after 1994 Northridge earthquake. The seismic per-

    formance of CBFs in past earthquakes has been good to excellent, but some brittle damage

    patterns were frequently observed including fracture of brace members due to excessive

    buckling deformation and fracture of connections. Buckling deformation is desirable;

    however, fracture under cyclic load is not. Seismically compact section must be used in

    cases where inelastic deformation is expected. Fracture of a connection that is a brittle limit

    state has to be prevented by imposing the maximum force possible for inelastic design

    (capacity design approach). Special CBF is expected to have significant inelastic defor-

    mation supply, and thus, designed with relatively lower strength than otherwise elastic

    response would require (on the order of three times). Ductile behavior and corresponding

    reduction in design lateral load come from the yielding of the tension brace (diagonal). The

    resulting frame deformation necessitates buckling of the compression brace (diagonal).

    ASCE 7-10 (2010) gives the design requirements of CBFs. However, ASCE 7-10

    (2010) has been primarily developed on seismic performance of building-like structures.

    Building is defined as any structure whose intended use includes shelter of human occu-

    pants (ASCE 7-10 2010). This definition is in terms of function, but does not mention

    anything related to structural characteristics, which include the following:

    a. Diaphragms or other horizontal elements as integral parts of lateral load resisting

    system.

    b. Non-structural components consisting of significant portion of total seismic weight,

    affecting actual stiffness and damping.

    The requirements given for building-like structures in ASCE 7-10 (2010) might be

    applicable to industrial structures with non-essential modifications. Structures not similar

    to buildings have fundamentally different response and need special attentions by indus-

    trial standards in addition to the requirements in ASCE 7-10 (2010). Non-building struc-

    tures similar to a building have the following features:

    a. Non-buildings structures are designed and constructed in a manner similar to

    buildings.

    b. They will respond to strong ground motions in a fashion similar to buildings.

    c. They have basic load carrying systems similar to those used for buildings.

    d. Lateral forces are usually transferred with diaphragms or other elements.

    The similarities between the requirements for building-like and non-building structures

    similar to buildings allow applying building code provisions directly whenever applicable.

    Differences between the building and industrial structures similar to buildings are also

    listed as follows (ASCE 7-10 2010):

    a. Non-structural components in buildings mainly serve the human occupants, whereas

    they serve mainly specific functions of the structures of particular industry in non-

    building structures that might have substantial impact on actual damping, stiffness, and

    strength. For example, the presence of partition walls and exterior enclosure walls in

    conventional buildings may result in large damping and initial stiffness than in non-

    building structures.

    Nat Hazards

    123

    Author's personal copy

  • b. Building-like structures might have irregularities in mass (affecting distribution of

    lateral forces if equivalent lateral force procedure is used and might require response

    spectrum procedure) and stiffness (which might cause significant torsional

    irregularity).

    c. Differences are reflected in the design process. For instance, the fundamental period of

    an industrial structure shall be evaluated by a rational method other than the

    approximate period for building-like structures, whereas for conventional buildings,

    the fundamental period can be obtained through empirical relations.

    d. Drift limitation for building-like structures (generally 2 % 9 story height) does not

    apply to non-building structures, and if analysis indicates, it is not critical for structural

    stability.

    5 Analytical study

    Two silos are used for an analytical study (Figs. 12, 13). Silo-1 in Fig. 12 is a typical silo

    in the region. These silos did not experience any significant damage during the earthquake.

    Silo-2 experienced some damage at horizontal braces and supporting concrete footing

    (Fig. 13). The bottom horizontal brace member has been buckled, and the bearing strength

    of the supporting concrete was exceeded and the base plate had some damage. Both silos

    are located in northern Van, 10 km to the city center (Fig. 14).

    The steel grade for the silos is classified as Fe37 with Fy = 235 MPa for the braces and

    rings, whereas FE52 steel grade with Fy = 345 MPa is used for the legs. Modulus of

    elasticity is taken as 200,000 MPa. The legs, rings, and braces (Figs. 15a, 16a) are modeled

    as steel frame elements, and body of the silo is modeled as thin plate shell. Cross-sectional

    properties are given in Table 2. Structural framing of the silos consisted of horizontal legs,

    braces, horizontal braces, silo body, and rings inside the body. The total height of the Silo-

    1 and Silo-2 is 17.884 and 14.398 m, respectively. The rings inside Silo-1 are located at the

    levels of 8.134, 9.634, 11.134, 12.634, 14.134, 15.724, and 16.384 m, whereas the rings

    inside Silo-2 are located at the levels of 4.730, 6.230, 7.730, 9.230, 10.730, 12.259, and

    12.919 m. For both silos, 11.0 tons of mass was assigned on center of each ring level

    except the upper ring, where 6.0 tons of mass is assigned on the center of upper ring levels

    to represent the mass and weight of the dry materials inside the silo, i.e., silo is assumed to

    be fully loaded. Thus, the total weight of each silo is assumed to be fully loaded with

    (7 9 11.0 tons ? 6.0 tons) 9 9.81 m/sec2 = 830 kN of dry material. The masses are

    connected to the rings with infinitely rigid, weightless, and mass-less fictitious frame

    members. The silo is assumed to be simply supported to the ground. Moment releases are

    assigned on both ends of braces and horizontal braces.

    Dynamic properties of the silos are given in Figs. 15b, c and Figs. 16b, c. The first

    mode of vibrations of Silo-1 and Silo-2 are found to be 1.09 and 0.63 s, respectively.

    Modal mass participating ratio for both silos is about 86 % for the first mode, where it adds

    up to 99.5 % for the first two modes.

    There was only one ground motion record for the Van earthquake recorded at Muradiye

    (Fig. 14). There was another ground motion station in the city center of Van, but the night

    watch shut down the electricity the night of the earthquake for some unknown reason

    according to local authorities. The distance between the location of the recorded ground

    motion station (Muradiye) and the silo location is about 70 km (Fig. 14). On November 9,

    2011, a second earthquake hit the region again with a magnitude of 5.6. Fortunately, this

    Nat Hazards

    123

    Author's personal copy

  • earthquake was recorded at many station around the earthquake region, and one of the

    stations, Semsibey, was within a few km of the Silos (Fig. 14). Characteristics and

    acceleration time histories of the Muradiye and Semsibey ground motions are given in

    Table 1 and Fig. 17, respectively. For comparison, uniform hazard spectra for 5 % critical

    damping are constructed for three earthquake levels, namely EQ-1, EQ-2, and EQ-3, based

    on RHA (2007). Short-period spectral acceleration, Ss, and 1-sec period spectral acceler-

    ation, S1, values corresponding to these earthquake levels are given in Table 3. Effective

    PGAs corresponding to these earthquake levels are determined as 0.24, 0.42, and 0.62 for

    EQ-1, EQ-2, and EQ-3, respectively. Fig. 18 shows the response spectra corresponding to

    Muradiye and Semsibey ground motions. For comparison, design spectra corresponding to

    10 % probability of exceedance in 50 years as described in Turkish Earthquake Code (TEC

    2,165 mm

    3@1,

    952

    mm

    2,27

    8 m

    m9,

    750

    mm

    17,8

    84

    mm

    Fig. 12 A typical undamaged

    silo in Van (Silo-1)

    Nat Hazards

    123

    Author's personal copy

  • 2007) are also given in Fig. 18 for different site conditions. Response spectrum definition

    in TEC (2007) is based on effective peak ground acceleration which is given as 0.4 g for

    the earthquake site.

    a b

    Fig. 13 Framing of a minor damaged silo due to horizontal brace buckling and concrete support failure in

    Van (Silo-2)

    Lake Van

    Turkey

    Fig. 14 Muradiye (AFAD) and Semsibey Primary School (KOERI) stations and silo locations (source:

    Google Earth)

    Nat Hazards

    123

    Author's personal copy

  • To determine the seismic response of the silos, nonlinear dynamic time history analyses

    are carried out. Damping ratio is assumed to be 2 % for dynamic analyses. P-Delta effect is

    always included in the analyses. Axial plastic hinges are assigned to the middle of the

    braces. Post-buckling behavior of the braces under compression is accounted for (Fig. 19).

    In Fig. 19, Ry is ratio of the expected yield stress to the specified minimum yield stress, Fy;

    Fcr is the critical stress; Ag is gross area; Pcr is maximum axial buckling load; dyc is the

    yield displacement under compression; dyt is the yield displacement under tension; Py is

    the maximum tension force; and Presidual is the residual load after buckling of the member.

    Dead and live loads are assumed as the initial conditions of the nonlinear dynamic time

    history analysis. Semsibey record is used for the nonlinear dynamic time history analyses

    and scaled to 0.20 g as justified in Sect. 2.2. To further investigate the nonlinear behavior

    of the silos, Semsibey ground motion was scaled to 0.40 g as defined in TEC (2007) for the

    site.

    Peak axial compression forces and peak drift ratios are summarized in Table 4. Both

    silos did not experience any inelastic action when subject to Semsibey ground motion

    scaled to 0.2 g, i.e., they remained elastic (Figs. 20, 21). Drift ratios were about 0.57 and

    0.41 % for Silo-1 and Silo-2, respectively (Table 4). Axial force in the most critical brace

    in Silo-1 was about 180 kN corresponding to a demand/capacity ratio of 0.6. This silo did

    not experience any damage during the earthquake, and the analyses indicated the same

    way. Axial force in the most critical brace in Silo-2 was about 150 kN corresponding to a

    demand/capacity ratio of 0.5 (Table 4). This silo suffered minor damage during the

    earthquake in the form of bottom horizontal brace buckling and minor damage in

    a b c

    First ring

    Upperrings

    Brace

    Leg

    Horizontalbrace

    Fig. 15 Model details and dynamic properties of Silo-1: a structural model, b Mode 1 (T = 1.09 s),

    c Mode 2 (T = 0.20 s)

    Nat Hazards

    123

    Author's personal copy

  • supporting concrete. However, dynamic analyses indicated that axial force level for the

    horizontal braces in Silo-2 was way lower than their capacity and axial load demand/

    capacity ratio was about 0.10 (Fig. 21). This indicates that damage occurred in the hori-

    zontal braces due to some other reason. It should be pointed out that Silo-2 had no tie

    beams connecting the supporting concrete.

    When subjected to Semsibey ground motion scaled to 0.4 g, the seismic response of

    both silos was moderate to heavy (Figs. 22, 23). Drift ratios increased to 1.01 and 0.71 %

    a b c

    First ring

    Upperrings

    Brace

    Leg

    Horizontalbrace50

    0 m

    m1,

    952

    mm

    9,66

    8 m

    m

    14,3

    98

    mm

    2,22

    78

    mm

    Fig. 16 Model details and dynamic properties of Silo-2: a structural model, b Mode 1 (T = 0.63 s),

    c Mode 2 (T = 0.13 s)

    Table 2 Silo properties

    Component Cross-sectional

    Silo-1 Silo-2

    Legs /300 9 6 mm hollow pipe section TS178 9 178 9 6.4 mm

    Braces 80 9 80 9 5 mm angle 80 9 80 9 5 mm angle

    First ring UPN120 UPN120

    Upper rings UPN80 UPN80

    Horizontal braces 80 9 80 9 5 mm angle 80 9 80 9 5 mm angle

    Body of silo 5 mm thin plate 5 mm thin plate

    Nat Hazards

    123

    Author's personal copy

  • for Silo-1 and Silo-2, respectively (Table 4). Braces in Silo-1 yielded, whereas only bottom

    short legs in Silo-2 experienced inelastic action, but not considered to be significant

    (Fig. 23).

    6 Comments

    This study investigated the effect of Van earthquake, 2011 (M = 7.2) in Turkey on

    industrial structures, especially on elevated silos. During the earthquake, some industrial

    structures with precast concrete framing systems, water and electric utilities, and elevated

    silos in concrete plants were either damaged or fully collapsed. Moreover, most of the silos

    were not insured. Therefore, legal issues also arised between the silo providers and owners.

    The equipment in these structures moved and/or collapsed, causing major business inter-

    ruption, loss of market, and stock damage. The loss of capacity in small- and

    a Muradiye

    -0.2

    -0.15

    -0.1

    -0.05

    0

    0.05

    0.1

    0.15

    0.2

    0 10 20 30 40 50 60Ac

    cele

    ratio

    n (g)

    Time (sec)

    -0.20

    -0.15

    -0.10

    -0.05

    0.00

    0.05

    0.10

    0.15

    0.20

    0 10 20 30 40 50 60

    Acce

    lera

    tion

    (g)

    Time (sec)

    EW - November 9, 2011

    Semsibeyb

    Fig. 17 Acceleration time history of the ground motions: a Muradiye Station NS component, b Semsibey

    Primary School EW component

    Table 3 Ssand S

    1values (RHA

    2007)Earthquake level Ss (g) S1 (g) PGA (g)

    EQ-1 (50 % PE in 50 years) 0.59 0.16 0.24

    EQ-2 (10 % PE in 50 years) 1.05 0.31 0.42

    EQ-3 (2 % PE in 50 years) 1.57 0.52 0.62

    Nat Hazards

    123

    Author's personal copy

  • 00.2

    0.4

    0.6

    0.8

    1

    1.2

    0 0.5 1 1.5 2 2.5 3

    Sae

    (g)

    T (sec)

    Stiff Soil (TEC)Stiff Clay (TEC)Soft Clay (TEC)Soft Soil (TEC)Muradiye NSSemsibey EW

    Fig. 18 Response spectra corresponding to Muradiye and Semsibey stations and TEC (2007) (5 % critical

    damping)

    a b

    P

    PCR=Ry FcrAg

    Py =Ry Fy Ag

    Presidual 0.3 Pcr

    P

    PCR

    Py

    0.3 Pcr

    ytyc

    3yc

    Fig. 19 Force-deformation behavior of a typical brace member: a hysteresis with post-buckling behavior,

    b idealized behavior

    Table 4 Peak axial compression

    force and drift ratios

    Pmax Axial compression strength

    Silo no. PGA Peak axial force (P) P/Pmax Peak drift ratio

    (g) (kN) (%)

    Silo-1 0.2 180 0.60 0.57

    0.4 300 1.00 1.01

    Silo-2 0.2 149 0.50 0.41

    0.4 235 0.78 0.71

    Nat Hazards

    123

    Author's personal copy

  • microenterprises had additional adverse socioeconomic effects due to loss of employment

    and production. The earthquake is estimated to have caused a total of 12 billion USD in

    total economic losses (Erdik et al. 2012). All these facts pointed out the need for con-

    structing safer new building and non-building structures and retrofitting thousands of

    Fig. 20 Seismic response of Silo-1 subject to Semsibey EW components scaled to 0.2 g

    Nat Hazards

    123

    Author's personal copy

  • similar existing vulnerable ones. The uneven damage distribution on silos in the earthquake

    posed a new question why not this one?. Damage to infrastructure and utility brought

    more reasons for erecting safer non-building structures in industrial facilities.

    Fig. 21 Seismic response of Silo-2 subject to Semsibey EW components scaled to 0.2 g

    Nat Hazards

    123

    Author's personal copy

  • 7 Conclusions

    In this study, seismic performances of the silos in the earthquake based on site observations

    are first investigated in detail. Seismic analyses on silos are also carried out to determine

    the seismic response of typical silos in the region through nonlinear dynamic time history

    analyses. The analytical studies are carried out on two elevated silos, and the results are

    Fig. 22 Seismic response of Silo-1 subject to Semsibey EW components scaled to 0.4 g

    Nat Hazards

    123

    Author's personal copy

  • discussed. The nonlinear behaviors of the members are modeled by considering the tension

    (yielding) and compression (buckling) hysteresis curves. As the input ground motion, the

    acceleration records from the nearest station (Semsibey) are used. The response spectrum

    of the ground motion is far below the TEC (2007) design spectrum. Therefore, inelastic

    behavior is not observed during the analysis. In general, the numerical results are observed

    to be consistent with the site observations. This is probably due to the low level of ground

    motion. However, when the input motion is scaled up to match the design earthquake,

    significant inelastic behavior is observed in one of the silos.

    The main outcomes of this study are summarized as follows:

    Fig. 23 Seismic response of Silo-2 subject to Semsibey EW components scaled to 0.4 g

    Nat Hazards

    123

    personal copy

  • 1. The failure of silos were brittle, sudden and generally due to weak detailing,

    inadequate design of supporting concrete footings.

    2. Field observations revealed only one cement silo collapsed due to inadequate design of

    braces and legs, whereas others collapsed because of supporting concrete failure due to

    lack of tie beams between the supporting concrete footings.

    3. Demand to capacity ratios remained well below 1.0 for the silos; that is, Van

    earthquake was not considered to be a strong earthquake for the non-building

    structures in the region.

    4. Overturning is a major problem for silos. Aspect ratio for the silos in this study was 8.3

    for Silo-1 and 6.65 for Silo-2.

    5. In current practice, supporting concrete area is generally equal to the base plate area.

    In these cases, tie beams connecting the supporting concrete footings should be used to

    prevent failure of the supporting concrete.

    6. Bearing strength of the supporting concrete is exceeded in many cases, and supporting

    concrete has no confinement effect when tie beams are not used.

    7. Construction quality and detailing are key issues in enhancing the seismic performance

    of the silos.

    8. The silos that survived the earthquake without any damage were designed, detailed,

    and constructed well.

    9. The scope of the future studies should be to investigate simple, practical, and efficient

    retrofitting methods. This is also expected to have an influence on the earthquake

    insurance premiums of likely facilities.

    Acknowledgments This research was supported by the Directory of Scientific Research Project of Bos-

    phorus University (Grant # BAP5715) and Directorate of Kandilli Observatory and Earthquake Research

    Institute, Bosphorus University (KOERI).

    References

    AFAD (2011) Prime Ministry Disaster and Emergency Management Agency. http://www.afad.gov.tr

    Akbas B, Uckan E (2012) Seismic and structural observations from Van earthquakes of October 23 and

    November 9, 2011. Gebze Institute of Technology, Department of Earthquake and Structural Engi-

    neering, Report No. 2012/01-1

    ASCE 7-10 (2010) Minimum design loads for buildings and other structures. American Society of Civil

    Engineers, ASCE

    Atzori S, Tolomei C, Salvi S, Zoffoli S (2011) Co-seismic ground displacement and preliminary source

    models for the 10/23/2011, Mw 7.2, Van earthquake, Eastern Turkey. SiGRiS activation report

    Bayrak Y, Yadav RBS, Kalafat D, Tsapanos TM, Cinar H, Singh AP, Bayrak E, Yilmaz S, Ocal F, Koravos

    G (2013) Seismogenesis and earthquake triggering during the Van (Turkey) 2011 seismic sequence.

    Tectonophysics 601:163176

    Dogangun A, Karaca Z, Durmus A, Sezen HM (2009) Cause of damage and failures in silo structures.

    J Perform Constr Fac 23:6571

    EERI (2012) Learning from earthquakes Mw 7.1 Ercis-Van, Turkey earthquake of October 23, 2011.

    Special Earthquake Report, Earthquake Engineering Research Institute

    Emre O, Duman TY, Ozalp S, Elmaci H (2012) Site observations and preliminary evaluation of source fault

    of 23 October 2011 Van earthquake. Active Tectonics Research Group, MTS Publications, Ankara in

    Turkish

    Erdik M, Uckan E. (2013) Earthquake damage and fragilities of the industrial facilities. International

    Conference on Seismic Design of Industrial Facilities, RWTH, Aachen University

    Erdik M, Kamer Y, Demircioglu M, Sesetyan K (2012) 23 October 2011 Van (Turkey) earthquake. Nat

    Hazards. doi:10.1007/s11069-012-0263-9

    Nat Hazards

    123

    Author's personal copy

  • Gulen L et al (2002) Surface fault break, aftershock distribution, and rupture process of the 17 August 1999

    Izmit, Turkey, earthquake. Bull Seismol Soc Am 92:230244

    Guney D (2012) Van earthquakes (23 October 2011 and 9 November 2011) and performance of masonry

    and adobe structures. Nat Hazards Earth Syst Sci 12:33373342

    Kalafat D, Suvarikli M, Ogutcu Z, Kekovali K, Yilmazer M, Ocal MF, Gunes Y (2012) A recent example of

    continentcontinent collision: October 23, 2011 Van Earthquake (Mw = 7.2) : Southeastern Turkey,

    American Geophysics Union AGU 2012 Fall Meeting, S51B-2420, 3-7 December 2012, San Fran-

    cisco, CA, USA

    KOERI (2011) KOERI Report. http://www.koeri.boun.edu.tr/depremmuh/deprem-raporlari/Van_Eq_SM_

    31102011.pdf

    METU/EERC (2011) Site observation of seismic and structural damage on Van earthquake of Mw = 7.2,

    October 23. Middle East Technical University, Earthquake Engineering Research Center, Report No.

    METU/EERC 2011-04

    Onen YH, Dindar AA, Cosgun C, Seckin E (2012) Reconnaissance and on-site evaluation report of Van

    Earthquakes: October 23 and November 9, 2013. Istanbul Kultur University

    Ozden S, Akpinar E, Atalay H (2011) Reconnaissance report on Van earthquake of October 2011. Kocaeli

    University

    Paolacci F, Giannini R, De Angelis M (2013) Seismic response mitigation of chemical plant components by

    passive control systems. J Loss Preven Proc 26(5):879948

    RHA (2007) Seismic code for the construction of harbor and coastal structures, railways, airports. General

    Directorate for Construction of Railways, Harbors and Airports, Ankara

    Silvestri, S, Ivorra S, Chiacchio LD, Trombetti T, Foti D, Gasparini G,Dietz M, Taylor C (2014) Shaking-

    table tests of flat-bottom silos containing grain-like material, Earthq Eng Struct D (in press)

    Silvestri S, Gasparini G, Trombetti T, Foti D (2012) On the evaluation of the horizontal forces produced by

    grain-like material inside silos during earthquakes. Bull Earthq Eng 10:15351560

    Taskin B, Sezen A, Tugsal UM, Erken A (2013) The aftermath of 2011 Van earthquakes: evaluation of

    strong motion, geotechnical and structural issues. Bull Earthq Eng 11:285312

    Uckan E (2013) Lifeline damage caused in the 23 October (Mw = 7.2) 2011 and 9 November (M = 5.6)

    2011, Van earthquakes in eastern Turkey. International efforts in Lifeline earthquake Engineering,

    Technical Council on Lifeline Earthquake Engineering, Monograph No: 38, ASCE Edited by: Davis C,

    Miyajima M., Yan L

    Nat Hazards

    123

    Author's personal copy