Seismic Analysis of Fire Sprinkler Systems

8
ABSTRACT: Post-earthquake functionality of fire sprinkler systems is critical for uninterrupted operation of the building; however, fire sprinklers have experienced extensive damage during past earthquakes. Comprehensive research is required to increase the understanding of the system level seismic performance of fire sprinkler systems subjected to realistic floor motions. In this study, a complete fire sprinkler system including risers, main distribution lines, and pipe branches, and sprinkler heads was analytically modeled using OpenSees. The model was subjected to the tri-axial floor responses of a three-story moment resisting frame obtained from a nonlinear incremental dynamic analysis. The seismic performance of the fire sprinkler system was assessed by statistically evaluating the relative displacement of sprinkler heads and peak forces in the seismic braces as a function of peak ground acceleration. The response of the fire sprinkler system was shown to be effected by the direction of excitation and seismic bracing. In addition, inelastic response of the supporting structure was shown to reduce both sprinkler head displacements as well as seismic brace and wire restraint forces. KEY WORDS: Fire Sprinkler; Seismic Response; Nonstructural Component, Piping, Fragility Functions. 1 INTRODUCTION A building’s ability to remain operational after a significant seismic event is dependent not only on the performance of the structural system but the functionality of several nonstructural systems as well. One of these critical nonstructural systems is the fire protection sprinkler piping system. Extensive damage to fire sprinkler systems in several past earthquakes has resulted in loss of functionality of many structures. Fire sprinkler systems are susceptible to several forms of earthquake damage. Inside the building, vertical pipes (risers) can break under large inter-story drifts in the building. Hangers supporting the weight of the pipe can unseat from their attachment points. Fasteners connecting the hangers to the building structure can pull out under seismic loading. Sprinkler heads can break upon impact with adjacent structural or nonstructural components, such as ceiling panels. Couplings and pipe fittings can break or leak. Piping run crossing separation joints that is not detailed for differential movement can be ruptured, as can pipes that are unintentionally restrained at locations where they pass through walls [1]. Nearly all of these failure modes have been observed in past earthquakes such as the 1964 Alaska Earthquake; the 1971 San Fernando Earthquake; the 1989 Loma Prieta Earthquake; and the 1994 Northridge Earthquake (Fig. 1). Damage to these systems may result in flooding, leaking, loss of functionality, damage to neighboring elements, and most importantly, loss of fire resistance. Criteria for the protection of the sprinkler system first appeared in NFPA 13 [2], Standard for the Installation of Sprinkler Systems, in the 1940s. This guideline has gone through several changes after the evaluation of facilities following major earthquakes. The stringent expected seismic performance of fire sprinklers necessitates special attention to designing and detailing these systems. Comprehensive research is yet required to increase the understanding of the system level performance of fire sprinkler systems subjected to realistic floor motions expected to occur in multistory buildings during earthquakes. Figure 1: Rupture of Sprinkler Pipe at the Elbow that in Sylmar Hospital in 1994 Northridge Earthquake (Photo courtesy of Robert Reitherman). In this study, a complete hospital fire sprinkler system including risers, main distribution lines, branches, sprinkler heads, hangers, and braces was analytically modeled in OpenSees [3] to investigate the influence of the type, quantity, and distribution of seismic braces on the dynamic response of fire sprinkler systems. The effect of nonlinear structural response was included in the investigation as well through the implementation of floor response excitations obtained from an incremental dynamic analysis of a three-story hospital building. Fire sprinkler piping systems are sensitive to both acceleration and deformation [4]. Thus, for a system installed in a multistory building, horizontal pipe runs are necessary to Seismic Analysis of Fire Sprinkler Systems Siavash Soroushian 1 , Arash E. Zaghi 1 , Joe Wieser 1 , E. Manos Maragakis 1 , Gokhan Pekcan 1 , Ahmad Itani 1 1 Department of Civil and Environnemental Engineering, University of Nevada, Reno MS258, 89557 Reno, USA email: [email protected],[email protected], [email protected], [email protected], [email protected], [email protected] Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 Leuven, Belgium, 4-6 July 2011 G. De Roeck, G. Degrande, G. Lombaert, G. M¨ uller (eds.) ISBN 978-90-760-1931-4 411

description

Fire Sprinkler Systems for buildings

Transcript of Seismic Analysis of Fire Sprinkler Systems

  • ABSTRACT: Post-earthquake functionality of fire sprinkler systems is critical for uninterrupted operation of the building;

    however, fire sprinklers have experienced extensive damage during past earthquakes. Comprehensive research is required to

    increase the understanding of the system level seismic performance of fire sprinkler systems subjected to realistic floor motions.

    In this study, a complete fire sprinkler system including risers, main distribution lines, and pipe branches, and sprinkler heads

    was analytically modeled using OpenSees. The model was subjected to the tri-axial floor responses of a three-story moment

    resisting frame obtained from a nonlinear incremental dynamic analysis. The seismic performance of the fire sprinkler system

    was assessed by statistically evaluating the relative displacement of sprinkler heads and peak forces in the seismic braces as a

    function of peak ground acceleration. The response of the fire sprinkler system was shown to be effected by the direction of

    excitation and seismic bracing. In addition, inelastic response of the supporting structure was shown to reduce both sprinkler

    head displacements as well as seismic brace and wire restraint forces.

    KEY WORDS: Fire Sprinkler; Seismic Response; Nonstructural Component, Piping, Fragility Functions.

    1 INTRODUCTION

    A buildings ability to remain operational after a significant seismic event is dependent not only on the performance of the

    structural system but the functionality of several nonstructural

    systems as well. One of these critical nonstructural systems is

    the fire protection sprinkler piping system. Extensive damage

    to fire sprinkler systems in several past earthquakes has

    resulted in loss of functionality of many structures.

    Fire sprinkler systems are susceptible to several forms of

    earthquake damage. Inside the building, vertical pipes (risers)

    can break under large inter-story drifts in the building.

    Hangers supporting the weight of the pipe can unseat from

    their attachment points. Fasteners connecting the hangers to

    the building structure can pull out under seismic loading.

    Sprinkler heads can break upon impact with adjacent

    structural or nonstructural components, such as ceiling panels.

    Couplings and pipe fittings can break or leak. Piping run

    crossing separation joints that is not detailed for differential

    movement can be ruptured, as can pipes that are

    unintentionally restrained at locations where they pass through

    walls [1]. Nearly all of these failure modes have been

    observed in past earthquakes such as the 1964 Alaska

    Earthquake; the 1971 San Fernando Earthquake; the 1989

    Loma Prieta Earthquake; and the 1994 Northridge Earthquake

    (Fig. 1). Damage to these systems may result in flooding,

    leaking, loss of functionality, damage to neighboring

    elements, and most importantly, loss of fire resistance.

    Criteria for the protection of the sprinkler system first

    appeared in NFPA 13 [2], Standard for the Installation of

    Sprinkler Systems, in the 1940s. This guideline has gone

    through several changes after the evaluation of facilities

    following major earthquakes. The stringent expected seismic

    performance of fire sprinklers necessitates special attention to

    designing and detailing these systems. Comprehensive

    research is yet required to increase the understanding of the

    system level performance of fire sprinkler systems subjected

    to realistic floor motions expected to occur in multistory

    buildings during earthquakes.

    Figure 1: Rupture of Sprinkler Pipe at the Elbow that in

    Sylmar Hospital in 1994 Northridge Earthquake (Photo

    courtesy of Robert Reitherman).

    In this study, a complete hospital fire sprinkler system

    including risers, main distribution lines, branches, sprinkler

    heads, hangers, and braces was analytically modeled in

    OpenSees [3] to investigate the influence of the type, quantity,

    and distribution of seismic braces on the dynamic response of

    fire sprinkler systems. The effect of nonlinear structural

    response was included in the investigation as well through the

    implementation of floor response excitations obtained from an

    incremental dynamic analysis of a three-story hospital

    building.

    Fire sprinkler piping systems are sensitive to both

    acceleration and deformation [4]. Thus, for a system installed

    in a multistory building, horizontal pipe runs are necessary to

    Seismic Analysis of Fire Sprinkler Systems

    Siavash Soroushian1, Arash E. Zaghi

    1, Joe Wieser

    1, E. Manos Maragakis

    1, Gokhan Pekcan

    1, Ahmad Itani

    1

    1Department of Civil and Environnemental Engineering, University of Nevada, Reno MS258, 89557 Reno, USA

    email: [email protected],[email protected], [email protected], [email protected], [email protected], [email protected]

    Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011Leuven, Belgium, 4-6 July 2011G. De Roeck, G. Degrande, G. Lombaert, G. Muller (eds.)ISBN 978-90-760-1931-4

    411

  • be modeled along with riser pipes that connect the horizontal

    runs and span between floors. Such a model accounts for the

    effects of differential floor motions of the building. For this

    study, a full 3D model comprising all three floors of the fire

    sprinkler system was investigated.

    The analytical model was subjected to the tri-axial floor

    responses of a three story moment resisting frame hospital

    building that was obtained from a series of nonlinear

    incremental dynamic analyses (IDA) [5]. The response of the

    sprinkler system was obtained in terms of the maximum

    relative displacement of sprinkler heads and forces in the

    seismic braces and wire restraints. Accordingly, probability of

    exceedance from certain demand levels were found at

    different peak ground acceleration (PGA) levels.

    2 METHOD OF ANALYSIS

    To capture the influence of the dynamic response of the parent

    building on the response of the installed fire sprinkler system,

    a three story hospital building was analyzed. The IDA method

    [5] was used to capture the seismic response of the hospital

    building from elastic ranges up to imminence of failure (inter-

    story drifts as large as 7%).

    Afterwards, the floor motions obtained from the analysis

    were introduced to the full model of fire sprinkler system as

    multiple support excitations. The analytical model of the

    sprinkler system comprised all the elements of the system

    such as braces, hangers, main runs and branches, risers, and

    sprinkler heads. Linear analyses were performed on the

    sprinkler system under the same sequence of floor motions

    that were obtained from the IDA analysis of the hospital

    building.

    By this means, the effects of linear and nonlinear response

    of the parent building were realistically accounted for. In this

    method, the dynamic interaction the building and sprinkler

    systems is neglected that is not far from reality, since the mass

    and stiffness of the subsystems is negligible compared to

    those of a building.

    3 COMPUTATIONAL MODEL OF THREE-STORY HOSPITAL BUILDING

    3.1 Specifications of the Building

    A special moment resisting frame (SMRF) hospital building

    was analyzed. Hospital structures are unique because after a

    seismic event, they are required to stay operational with

    minimal interruption. Hospital buildings contain a large

    number of nonstructural components most of which are

    crucial for the functionality of the hospital.

    The plan view and elevation of this three-story hospital is

    presented in Fig. 2. This hospital building was designed as a

    part of this research after SAC three-story office building.

    SAC building had been designed as part of project performed

    by SAC Steel Joint Venture [6]. SAC office building was

    redesigned according to the 2006 International Building Code

    [7] to satisfy the design requirements of the seismic design

    category D. The plan dimensions mimic that of SAC building

    while the elevation was adjusted to be more representative of

    current hospital buildings. It has plan dimensions of 180 ft

    [54.8 m] by 120 ft [36.6 m]. The story heights were adjusted

    to 20 ft [6 m], 16 ft [4.9 m] and 16 ft [4.9 m] making the

    building 52 ft [15.8 m] tall. The lateral load-resisting system

    is comprised of four 3-bay perimeter SMRFs, as shown with

    bold lines in Fig. 2, with simple framing between them.

    Figure 2: Geometry of the Hospital Building

    3.2 OpenSees Model

    The main goal for the analysis of the building was to assess

    the bi-directional floor responses of yielding building

    structures under all three components of seismic excitation by

    accounting for the out-of-plane flexibility of floor decks. With

    these objectives in mind, utilization of three-dimensional

    models with explicitly modeled gravity framing system and

    floor slabs was necessary. The floor decks were modeled

    using two-dimensional area elements. The gravity framing

    system was included for the same reason.

    OpenSees nonlinear analysis software [3] was used for

    including material and geometric nonlinearities (P-Delta

    effects). The beams and columns were modeled using Force-Based Beam-Column frame elements and a Fiber Section object was assigned to these elements. Each Fiber Section object was composed of nine fibers (three on each flange and

    three on the web). Steel02 uniaxial material was used with yield stress of 50 ksi [345 MPa] and strain hardening of 2% of

    the elastic stiffness. Centerline dimensions were used to

    define the model. The fundamental period of the hospital

    building was 1.0 sec. For time history analysis, Rayleigh

    damping was applied and a 5% damping ratio was assigned to

    0.8 and 2.0 times the fundamental period.

    3.3 Input Motions

    The ATC-63 project developed a set of 22 far-field ground motions from recorded seismic events taking place around the

    world [8]. These motions were collected for the purpose of

    performing incremental dynamic analysis to develop the

    collapse margin ratio of a structure [8]. Though this

    investigation is not interested in studying seismic capacity of

    buildings, the set of motions has been deemed appropriate for

    use with incremental dynamic analysis, and as such was

    adopted for this investigation. Table 1 shows the names,years,

    magnitudes, PGAs, and the name of the recording station.

    Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 412

  • Table 1: Ground Motions

    Earthquake Recording Station PGA

    (g) Name Year Mag. Name

    Northridge 1994 6.7 Beverly Hills-Mul. 0.517

    Northridge 1994 6.7 Canyon Country-WLC 0.482

    Duzce, Turkey 1999 7.1 Bolu 0.822

    Hector Mine 1999 7.1 Hector 0.337

    Imperial Valley 1979 6.5 Delta 0.351

    Imperial Valley 1979 6.5 El Centro Array #11 0.380

    Kobe, Japan 1995 6.9 Nishi-Akashi 0.509

    Kobe, Japan 1995 6.9 Shin-Osaka 0.243

    Kocaeli, Turkey 1999 7.5 Duzce 0.358

    Kocaeli, Turkey 1999 7.5 Arcelik 0.219

    Landers 1992 7.3 Yermo Fire Station 0.245

    Landers 1992 7.3 Coolwater 0.417

    Loma Prieta 1989 6.9 Capitola 0.529

    Loma Prieta 1989 6.9 Gilroy Array #3 0.555

    Manjil 1990 7.4 Abbar 0.515

    Superstition Hills 1987 6.5 El Centro Imp. Co. 0.358

    Cape Mendocino 1992 7 Rio Dell Overpass 0.549

    Chi-Chi, Taiwan 1999 7.6 CHY101 0.440

    Chi-Chi, Taiwan 1999 7.6 TCU045 0.512

    San Fernando 1971 6.6 LA-Hollywood 0.210

    Friuli 1976 6.5 Tolmezzo 0.351

    Minimum 1971 6.5 - 0.210

    Maximum 1999 7.6 - 0.822

    Average - 7.0 - 0.424

    Figure 3: Acceleration Spectra the Input Motions

    The 5% damped elastic acceleration spectra of the major,

    minor, and vertical components are presented in Fig. 3,

    respectively. All three components of the ground motion were

    used to excite the structure with the major component of the

    applied ground motion oriented in the long-direction, the

    minor component in the short-direction, and the vertical

    component in the z-direction. This orientation was consistent

    for all 21 events.

    3.4 IDA Analysis Results

    The scaling method developed by Vamvatsikos and Cornell

    [5] was used after minor modification to find a data point

    within the predetermined drift range. The drift window for

    this study was set between 5% drift and 7% drift. Figure 4

    shows the IDA curves that were obtained by fitting SP-Lines

    on data points. Figure 4 shows the IDA curves for the

    maximum inter-story drift and peak roof acceleration. The

    smaller rate of increase in roof acceleration in larger motions

    is a result of yielding of the structure.

    Figure 4: IDA Results

    4 COMPUTATIONAL MODEL OF FIRE SPRINKLER PIPING SYSTEM

    4.1 Specifications of Fire Sprinkler Piping System

    The dimensions and layout of fire protection piping systems

    are individualized for each building and vary based on the

    architecture and occupancy of the parent building.

    Consequently, selecting a generic fire piping system was

    somewhat arbitrary. The fire piping system modeled in this

    study is a portion of the fire sprinkler system in the UCSF

    Medical Center Hospital Building. As part of the fire

    protection sprinkler system of a 15-story hospital structure,

    the selected piping layout is quite robust. It incorporates a

    variety of commonly used components such as distribution

    mains of various diameters, branch lines of various diameters,

    hangers, seismic braces, wire restraints, tee joints, elbow

    joints, pipe reduction joints, and sprinkler heads. It also

    0.0

    1.0

    2.0

    3.0

    Sa (g

    )

    Major

    0.0

    1.0

    2.0

    3.0

    Sa (g

    )

    Minor

    0.0

    1.0

    2.0

    3.0

    0.0 1.0 2.0 3.0

    Sa (g

    )

    Period (sec)

    Events

    +

    +2

    Vertical

    0.0

    1.0

    2.0

    3.0

    4.0

    5.0

    0.00 0.02 0.04 0.06 0.08 0.10

    PG

    A (g

    )

    Maximum Interstory Drift [max]

    0.0

    1.0

    2.0

    3.0

    4.0

    5.0

    0.0 1.0 2.0 3.0 4.0 5.0

    PG

    A (g

    )

    Peak Roof Acceleration [PFAR] (g)

    Events

    Minimum

    16th %

    Median

    84th %

    Maximum

    Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 413

  • contains a sufficient quantity of each of these commonly used

    components to statistically evaluate the seismic performance

    of each type of component in the system.

    Figure 5: Plan View of Fire Piping System

    A plan view of the modeled portion of the sprinkler system

    is shown in Fig. 5. The plan layouts are consistent between

    floors and are connected to a 6 in. [150 mm] riser located at

    the top end in Fig. 5. A penolum height (the height between

    the supporting floor and the ceiling system) of 4 ft [1.2 m]

    was assumed for each floor. The pipe system was suspended

    2.5 ft [762 mm] below the supporting floor and the sprinkler

    drops 1.5 ft [458 mm] to the ceiling height. The 82 ft [25 m]

    main distribution line varies in diameter from 4 in [100 mm]

    at the connection to the riser to 2 in [50 mm]. at the end of the

    main. The main distribution line feeds 16 branch lines ranging

    in diameter from 1-1/2 in. [38 mm] to 1 in. [25 mm]. The

    main and branch lines supply 47 sprinkler heads. The

    sprinkler pipe system is connected to the supporting floor with

    a combination of 12 seismic strut braces, 54 vertical hangers,

    and 80 wire restraints. The seismic braces consist of a 1 in.

    [25 mm] diameter strut oriented in a 45 degree angle from the

    supporting floor such to provide lateral bracing in one

    direction to the pipe Fig. 6. The vertical hangers are 3/8 in.

    [10 mm] threaded rods as required for pipes of 4 in. [100 mm]

    in diameter or less per the NFPA 13 [2]. Lateral restraint at

    the ends of branch lines was provided by a set of 1/8 in [3

    mm] wire restraints Fig. 6. Each of the four wires used in the

    restraint are made of aircraft grade steel with minimum break

    strength of 1700 lb [7.5 kN].

    4.2 OpenSees Model

    The objective of the fire piping system model was to identify

    the relative displacement demands on sprinkler heads and

    force demands in seismic braces and wire restraints of an

    elastic fire sprinkler system under realistic tri-axial differential

    floor motions. This required an elastic model of the three floor

    fire piping system shown in Fig. 7.

    (1)

    (2)

    Figure 6: Pipe Bracing: 1) Seismic Brace, 2) Wire Restraint

    Developed in OpenSees [3], the model consists of two types

    of elements. All of the pipes (riser, mains, and branches) are

    modeled using Force-Based Beam-Column elements with elastic sections. The seismic braces, hangers, and wire

    restraints were modeled with elastic truss elements. All of the

    elements are made of steel with a modulus of elasticity of

    29,000 ksi [200 GPa]. The mass of the piping system was

    determined using the wet weight of the pipes, assuming a

    sprinkler head weight of lb. [0.25 kg]. Both the mass and

    weight were evenly distributed along the length of each pipe

    element and the mass of the sprinkler heads was concentrated

    at the nodal locations of the sprinkler heads. Centerline

    dimensions of the piping system were used to define the

    elements. All pipe to pipe connections were assumed to be

    continuous while the braces, hangers, and wires were assumed

    to be pin-ended truss members. The connection of the riser to

    the ground level was such that no moment demand was

    imposed on the base of the riser pipe. No special connection

    details were used in this basic elastic model of the sprinkler

    piping system. The stiffness of the pipe elements was

    Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 414

  • determined from their respective cross section and does not

    account for the influence of the internal pressurized water.

    The stiffness of the seismic braces and vertical hangers were

    also based on the cross sectional properties, but the stiffness

    of the wire restraints was modified to 10% of their full

    stiffness to compensate for initial slack. This reduction in

    stiffness was based on the recommendation of Zaghi et al. [9].

    A Rayleigh damping matrix was used in the piping model and

    3% damping was assigned to the first and third modes. The

    fundamental period of the piping system was 0.8 seconds and

    the shape of its fundamental mode had a peak amplitude in the

    middle of the second branch from the top on the left side of

    Fig.5.

    4.3 Input Motions

    Once the fire sprinkler model was defined, multiple support

    excitation was used to induce the inertial forces. The input

    motions for each floor level of the piping model coincided with

    the response motion of the previously modeled inelastic three-

    story hospital building. All three orthogonal axes of building

    response were used to excite the piping model. The major

    component of the ground motion and long direction of the

    building were aligned with the axis perpendicular to the main

    distribution lines of the piping system. The vertical floor

    excitation corresponds to the vertical response of a point along

    a gravity load carrying beam in each floor of the three-story

    hospital building.

    In total, 320 time history analyses of the piping system were

    performed, 21 events each scaled to at least 11 different

    intensities. In this manner, a variety of structural responses of

    the hospital building, ranging from elastic to near collapse, were

    introduced to the piping system.

    Figure 7: Three Dimensional View of Fire Piping Model

    5 ANALYTICAL RESULTS

    5.1 Data Analysis Methodology

    Post-processing the data from 141 sprinkler heads, 36 seismic

    braces, and 240 wire restraints required a well defined statistical

    approach. First, all of the peak sprinkler head displacements,

    peak seismic braces, and peak wire restraint forces from each

    analysis were plotted in scatter form and determined to have a

    log-normal distribution. A sample scatter plot is presented in

    Fig.8. Then, using a log-normal distribution cumulative

    probability function the probability of exceeding predetermined

    levels of response were calculated and plotted against the peak

    ground acceleration.

    The results presented explore the influences of excitation

    directivity and inelastic response of the supporting structure on

    the fragility of components of the fire sprinkler system. In order

    to highlight the effect of nonlinear response of the supporting

    structure analysis results from low intensity motions (elastic

    response) were scaled up to higher intensities and compared

    with the results which include inelastic response of the

    structure.

    Figure 8: Dispersion of Sprinkler Head Displacement along

    Short Axis

    5.2 Relative Displacement of Sprinkler Heads

    The geometry of typical fire sprinkler heads is such that the

    center of the sprinkler head is located in the center of a 2 in.

    diameter ring in the ceiling [10]. If the ceiling is assumed to

    move rigidly with the supporting floor slab, the relative

    displacement of the sprinkler head is limited to 1 in. before it

    makes contact with the ceiling and has the potential of

    breaking. This study investigates the characteristics of the

    relative displacement of sprinkler heads during both elastic and

    inelastic building response and quantifies the potential of

    sprinkler head failure due to contact with the ceiling system.

    The time history relative displacement response of similar

    sprinkler heads located on each of the three floors is compared

    in Fig. 9. The response of the sprinkler heads under low

    intensity ground excitation (PGA = 0.055 g) was considerably

    different compared to the same event at a higher intensity (PGA

    = 1.805 g).

    Under the low intensity motion the building responds

    elastically therefore there is a linear correlation between the

    responses of each floor which carries through to the response of

    the elastic piping system. This observation is clear in Fig. 9a, in

    which the responses of similar sprinkler heads at each level

    were in phase with one another and increase in amplitude with

    increasing floor level. The elastic response of the hospital

    building was dominated by the first mode of vibration, thus

    explaining the increasing amplitude in sprinkler head

    displacement due to increased acceleration response of the

    parent building throughout the height.

    When the intensity of the ground excitation exceeded the

    elastic capacity of the hospital building, the linear correlation

    between floor responses was lost. As a result, the floor

    0.00254

    0.0254

    0.254

    2.54

    25.4

    254

    0.0001

    0.001

    0.01

    0.1

    1

    10

    0.0 1.0 2.0 3.0 4.0 5.0

    Sp

    rin

    kle

    r H

    ead

    Dis

    pla

    cem

    en

    t (m

    m)

    Sp

    rin

    kle

    r H

    ead

    Dis

    pla

    cem

    en

    t (i

    n)

    PGA (g)

    Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 415

  • excitations imposed on the piping system no longer were in

    phase with one another due to changing modal response

    characteristics as the building yields. This is observed clearly in

    the sprinkler head response shown in Fig. 9b. Contrary to the

    elastic response, the peak relative displacement of the

    sprinkler head located on the first floor is greater than that on

    the third floor. This is due to the deamplification of the

    ground motion throughout the height of the building as a

    result of yielding.

    Figure 9: Sprinkler Head Displacement History along Short

    Axis (a) Elastic Building Response (b) Inelastic Building

    Response

    In order to quantify the potential of sprinkler head failure

    due to contact with the ceiling, the probability of exceeding

    in. and 1 in. of relative displacement in each horizontal

    direction was plotted against the peak ground acceleration of

    each direction, as shown in Fig. 10.

    There is clearly a directivity influence on the potential of

    making contact with the ceiling. This particular fire sprinkler

    system was more vulnerable to acceleration along its short

    axis then along its long axis. Due to the well braced nature of

    the piping system along its short axis and limited bracing

    perpendicular to several branches, the probability of

    experiencing large sprinkler head displacements along the

    long axis was greater than along the short axis. Several

    sprinkler heads were observed to experience relative

    displacements in the long direction of up to 5 times that of the

    short direction. The location of these peak displacements was

    at sprinkler heads of unbraced branches, the same branches

    that had large amplitudes in the fundamental mode shape.

    These branches that were thought to be susceptible to large

    displacements have been proven to in fact experience

    extremely high displacement demands which may lead to

    failure of sprinkler heads attached to these unbraced branches

    as they contact the ceiling or other surrounding systems.

    Figure 10: Influence of Directivity on Probability of Sprinkler

    Head Displacements Exceeding 1/2 in [12.5 mm] (EL 1) and 1

    in [25 mm] (EL 2)

    Similar plots were developed to assess the influence of

    inelastic response of the supporting structure. Figure 11 plots

    the potential of exceeding 1 in. of relative displacement in the

    major direction for both an elastic and inelastic supporting

    structure. The potential of exceeding a predetermined

    displacement limit was reduced when the supporting structure

    is allowed to yield. The increased level of yielding in the

    supporting structure damps the acceleration response of its

    floors thus reducing the input to the sprinkler piping system.

    Figure 11: Influence of Structural Inelasticity on Probability

    of Short Axis Sprinkler Head Displacements Exceeding 1 in

    [25 mm]

    5.3 Wire Restraint Forces

    Performing a similar analysis of the wire restraints required

    the determination of sample limit states of the wire restraints.

    The breaking force of the wire restraints was specified as

    1700 lb. thus 60% of the breaking force was used to define the

    design force for the wire restraints. Limit states of 15%, 50%,

    and 100% of the design force were set.

    -3.81

    -2.54

    -1.27

    0

    1.27

    2.54

    3.81

    -0.15

    -0.10

    -0.05

    0.00

    0.05

    0.10

    0.15

    Dis

    pla

    cem

    en

    t (m

    m)

    Dis

    pla

    cem

    en

    t (i

    n)

    (a)

    -38.1

    -25.4

    -12.7

    0

    12.7

    25.4

    38.1

    -1.50

    -1.00

    -0.50

    0.00

    0.50

    1.00

    1.50

    2.0 4.0 6.0 8.0 10.0 12.0

    Dis

    pla

    cem

    en

    t (m

    m)

    Dis

    pla

    cem

    en

    t (i

    n)

    Time (sec)

    First Floor Second Floor Third Floor

    (b)

    0%

    10%

    20%

    30%

    40%

    50%

    60%

    70%

    80%

    90%

    100%

    0.0 1.0 2.0 3.0 4.0 5.0

    Pro

    bab

    ilit

    y o

    f E

    xceed

    an

    ce (%

    )

    PGA (g)

    EL 1 (Long) EL 2 (Long) EL 1 (Short) EL 2 (Short)

    0%

    10%

    20%

    30%

    40%

    50%

    60%

    70%

    80%

    90%

    100%

    0.0 1.0 2.0 3.0 4.0 5.0

    Pro

    bab

    ilit

    y o

    f E

    xceed

    an

    ce (%

    )

    PGA (g)

    Elastic

    Inelastic

    Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 416

  • Figure 12: Influence of Structural Inelasticity on Probability

    of Wire Restraint Forces Exceeding 15% (EL 1), 50% (EL 2),

    and 100% (EL 3) of Their Design Force

    The likelihood of exceeding each of these limit states is

    plotted against the peak ground acceleration for each direction

    in Fig. 12. The wire restraint forces, similar to the sprinkler

    head displacement, show a profound influence of directivity.

    The difference, however, is that while the sprinkler head

    displacements in the major direction were significantly less

    than those in the minor direction the opposite is true for the

    wire restraint forces, and vice versa in the alternate direction.

    This shows that while providing seismic restraint limits the

    displacements it increases the force demands on the restraints.

    Figure 13: Influence of Structural Inelasticity on Probability

    of Wire Restraint Forces along the Short Axis Exceeding

    100% of Their Design Force

    Similar to the sprinkler head displacements the influence of

    inelastic response of the supporting structure reduce the

    demands of fire sprinkler system. Figure 13 presents the

    comparison of the probability of exceeding the design force of

    the wire restraints in the minor direction under elastic and

    inelastic response of the supporting structure. Interestingly,

    the data seems to follow a linear trend for the elastic building

    response, as expected, but for the inelastic building response

    the data shows a decreasing trend for very strong ground

    excitations. This means that there comes a point where the

    structure has become so ductile that the acceleration demand

    on the fire sprinkler piping system is less than the ground

    acceleration.

    5.4 Seismic Brace Forces

    The last components examined in this investigation were the

    forces in the seismic braces. For these components the design

    force was determined from the critical buckling capacity of

    the 1 in pipe section used as the seismic braces. The design

    force of 6.3 kips was calculated and similar to the wire

    restraint limit states 15%, 50%, and 100% of the design force

    were defined as the force demands of interest.

    Figure 14: Influence of Structural Inelasticity on Probability

    of Seismic Brace Forces Exceeding 15% (EL 1), 50% (EL 2),

    and 100% (EL 3) of Their Design Force

    Figure 14 illustrates the influence of directivity on the

    potential of exceeding each of the force limits. Similar trends

    to the wire restraint forces were identified. However, the data

    for the seismic brace forces show slightly more scatter due to

    the relatively small number of seismic braces in the fire

    sprinkler plan.

    Figure 15: Influence of Structural Inelasticity on Probability

    of Seismic Brace Forces along the Long Axis Exceeding

    100% of Their Design Force

    0%

    10%

    20%

    30%

    40%

    50%

    60%

    70%

    80%

    90%

    100%

    0.0 1.0 2.0 3.0 4.0 5.0

    Pro

    bab

    ilit

    y o

    f E

    xceed

    an

    ce (%

    )

    PGA (g)

    EL 1 (Long) EL 2 (Long) EL 3 (Long)

    EL 1 (Short) EL 2 (Short) EL 3 (Short)

    0%

    10%

    20%

    30%

    40%

    50%

    60%

    70%

    80%

    90%

    100%

    0.0 1.0 2.0 3.0 4.0 5.0

    Pro

    bab

    ilit

    y o

    f E

    xceed

    an

    ce (%

    )

    PGA (g)

    Elastic

    Inelastic

    0%

    10%

    20%

    30%

    40%

    50%

    60%

    70%

    80%

    90%

    100%

    0.0 1.0 2.0 3.0 4.0 5.0

    Pro

    bab

    ilit

    y o

    f E

    xceed

    an

    ce (%

    )

    PGA (g)

    EL 1 (Long) EL 2 (Long) EL 3 (Long)

    EL 1 (Short) EL 2 (Short) EL 3 (Short)

    0%

    10%

    20%

    30%

    40%

    50%

    60%

    70%

    80%

    90%

    100%

    0.0 1.0 2.0 3.0 4.0 5.0

    Pro

    bab

    ilit

    y o

    f E

    xceed

    an

    ce (%

    )

    PGA (g)

    Elastic

    Inelastic

    Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 417

  • The reduction of force demands due to inelasticity in the

    supporting structure was shown again for the seismic brace

    forces in Fig. 15. A more significant influence of the building

    inelasticity was observed in the likelihood of exceeding the

    seismic brace design force than that shown in the wire

    restraints.

    6 CONCLUSIONS

    This investigation examined the performance of an elastic

    hospital fire sprinkler system under realist tri-axial multi-

    support excitation. Using an uncoupled analysis approach in

    which a three-story hospital building was analyzed using

    nonlinear IDA for a set of 21 far-field ground motions and the

    ensuing floor responses were used to excite three levels of a

    fire sprinkler system, the displacement demands on the

    sprinkler heads along with the force demands in the seismic

    braces and wire restraints were assessed. Significant influence

    of geometry were identified, most critically the effect of

    limited bracing on pipe branches may decrease force demands

    while greatly increasing sprinkler head displacement

    demands. The effect of yielding in the supporting structure

    was assessed. It was concluded that increased levels of

    yielding in the supporting structure reduces both sprinkler

    head displacement demands as well as seismic brace and wire

    restraint forces. Also, the effect of unbraced branches was

    shown to significantly increase the peak sprinkler head

    displacements. In total this investigation has laid the ground

    work for more comprehensive investigations of the seismic

    performance of fire sprinkler systems.

    ACKNOWLEDGMENTS

    This material is based upon work supported by the National

    Science Foundation under Grant No. 0721399. Any opinions,

    findings, conclusions or recommendations expressed in this

    document are those of the investigators and do not necessarily

    reflect the views of the NSF. The input provided by the

    Practice Committee of the NEES Nonstructural Project,

    composed of W. Holmes (Chair), D. Allen, D. Alvarez, and R.

    Fleming; by the Advisory Board, composed of R. Bachman

    (Chair), S. Eder, R. Kirchner, E. Miranda, W. Petak, S. Rose

    and C. Tokas; and by the other members of the Experimental

    Group, A, Filiatrault, G. Mosqueda, A. Reinhorn, and J. T.

    Hutchinson, has been crucial for the completion of this

    research. The authors are especially grateful to Abhinav Gupta

    for providing the piping plan.

    REFERENCES

    [1] SEAOC Seismology Committee. SEAOC Blue Book: Seismic Design Recommendations. Sacramento, CA, 2006.

    [2] NFPA 13 Standard for the Installation of Sprinkler Systems, National Fire Protection Association, 2010 edition, Quincy, MA, 2011.

    [3] OpenSees Ver. 2.2.2f Open System for Earthquake Engineering Simulation, http://opensees.berkeley.edu/, Berkeley, CA, 2010.

    [4] Federal Emergency Management Agency Reducing the Risks of Nonstructural Earthquake Damage, FEMA E-74, U.S. Department of Homeland Security, Washington, DC, 2010.

    [5] Vamvatsikos, D., & Cornell, C. Applied Incremental Dynamic Analysis. Earthquake Spectra , 20 (2), 2004.

    [6] Gupta, A., Krawinkler, H., Seismic Demands for Performance Evaluation of Steel Moment Resisting Frame Structures, John A. Blume Earthquake Engineering Research Center Report No. 132, Department of Civil Engineering, Stanford University, 1999.

    [7] ICC. International Building Code. International Code Council, Falls Church, VA, 2006.

    [8] FEMA P695 Quantification of Building Seismic Performance Factors, Federal Emergency Management Agency, Washington, D.C., 2009.

    [9] Zaghi, A., Maragakis, E., Itani, A., & Goodwin, E. Experimental and Analytical Studies of Hospital Piping Assemblies Subjected to Seismic

    Loading. Earthquake Spectra, In Press, 2011. [10] E 580/E 580M-09a Standard Practice for Installation of Ceiling

    Suspension Systems for Acoustical Tile and Lay-in Panels in Areas

    Subject to Earthquake Ground Motions. West Conshohocken, PA, 2009.

    Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 418