ArticlePDF Available

Hybrid Simulation of Structure-Pipe-Structure Interaction within a Gas Processing Plant

September 2020
Journal of Pipeline Systems Engineering and Practice 12(2)

DOI:10.1061/(ASCE)PS.1949-1204.0000526

Authors:

Ziliang Zhang

University of Southampton

Jamin Park

Myongji University

Oh-Sung Kwon

University of Toronto

Anastasios Sextos

University of Bristol

Show all 7 authorsHide

Though often overlooked, the impact of seismic transient ground deformation on natural gas (NG) pipes can be highly adverse. Particularly, pipe elbows may undergo excessive in-plane bending demand and buckling. In this paper, a critical scenario of a pipe coupling two industrial structures typically found in an NG processing plant is studied. High strain and cross-sectional ovalization on the elbows are probable during an earthquake due to the out-of-phase oscillation of the two structures imposing asynchronous displacement demands at the two pipe-ends. A parametric study was first performed to investigate various structure-pipe-structure configurations which increase seismic demands to pipe elbows. Simultaneous mobilisation of divergent oscillation between two supporting structures at the low-frequency range, a lower pipe-structure stiffness ratio, a shorter length of straight pipe segments in the linking pipe element and a higher pipe internal pressure have led to the onset of critical strain demands in pipe elbows.

Illustration of acceleration time histories of Ricker wavelets with varying predominant frequencies.

…

, and , responses with respect to variations of structural frequency ratio / and predominant frequency of excitation (Ricker wavelets for sub-figure (a) and (b); earthquake ground motions for sub-figure (c) and (d)). Note the variation of / from 0.30 to 2.73 corresponds to ≡ 3.3 Hz and varying from 1.0 Hz to 9.0 Hz.

…

Acceleration time histories (left) and their Fast Fourier Transform amplitudes (right) used as input ground motions. PF: Predominant Frequency.

…

Substructuring scheme and key components of the HS. NICON = network interface for the controller; UTNP = University of Toronto networking protocol; and EQ = earthquake.

…

() responses (left) and force-displacement curves (right) obtained from the HS and the corresponding FE prediction. The () response of a no-pipe case is also plotted to reflect the impact of structure-pipe-structure interaction.

…

Figures - uploaded by the authors

Content may be subject to copyright.

Content uploaded by Jamin Park

Content may be subject to copyright.

Content uploaded by Anastasios Sextos

Content may be subject to copyright.

Hybrid Simulation of Structure-Pipe-Structure

Interaction within a Gas Processing Plant

Ziliang Zhang1; Jamin Park, Ph.D.2; Oh-Sung Kwon, Ph.D., P.Eng., M.ASCE3;

Anastasios Sextos, Ph.D., M.ASCE4; Elias Strepelias, Ph.D.5;

Nikolaos Stathas, Ph.D.6; and Stathis Bousias, Ph.D.7

Abstract: Though often overlooked, the impact of seismic transient ground deformation on natural gas (NG) pipes can be highly adverse.

Particularly, pipe elbows may undergo excessive in-plane bending demand and buckling. In this paper, a critical scenario of a pipe coupling

two industrial structures typically found in an NG processing plant is studied. High strain and cross-sectional ovalization on the elbows are

probable during an earthquake due to the out-of-phase oscillation of the two structures imposing asynchronous displacement demands at the

two pipe ends. A parametric study was first performed to investigate various structure-pipe-structure configurations that increase seismic

demands to pipe elbows. Simultaneous mobilization of divergent oscillation between two supporting structures at the low-frequency range, a

lower pipe-structure stiffness ratio, a shorter length of straight pipe segments in the linking pipe element, and a higher pipe internal pressure

have led to the onset of critical strain demands in pipe elbows. To validate this observation, an experimental campaign was developed in which

a full-scale linking pipe element was physically tested by means of hybrid simulation (HS). The study shows that the seismic interaction of the

structures coupled with the pipe is nonnegligible and can even be critical for the integrity of the coupling pipe. The finding depends on the

structural system’s dynamic and geometrical properties as well the frequency content of the earthquake excitation. DOI: 10.1061/(ASCE)

Introduction

Natural gas (NG) constitutes a significant percentage of current

global energy consumption. Its demand has increased over the last

decade and is expected to proliferate into the future with increased

global interest in clean energy (DOE 2017;Sextos et al. 2018).

Among many factors, transport and supply play an essential role

in the NG industry, which includes transmission, storage, gas

liquefaction, and regasification (GIE 2015). Since NG reserves

are commonly distant from consumer markets, the need for deliv-

ering NG to end-users has led to the world’s mass construction

of complex lifeline systems with numerous integrated components

and processes. In 2019, the annual regasification capacity of

large-scale liquefied natural gas (LNG) terminals in Europe came

to 241 billion m3ðNÞ=year and capacity expansion of another

140 billion m3ðNÞ=year was planned (GIE 2019). Along with their

clear economic and strategic importance, NG facilities are often

associated with high natural and man-made risks. As a result,

gas infrastructure security and safety has always been the core value

of NG transmission facilities such as LNG terminals, compres-

sion stations, peak sheaving stations, pressure let-down stations,

and blending stations, which are vulnerable to natural hazards

such as earthquakes. In the last decade, natural hazard triggering

technological accidents (Na-Tech), and more specifically seismic

Na-Tech, has been slowly accepted as a fundamental contribution

to the overall risk assessment figures calculated by considering

solely industrial accidents (Lanzano et al. 2015). Past Na-Tech

disasters have displayed such devastating consequences in causing

substantial social, economic, and environmental loss that the future

prevention of loss of containment events in critical infrastructures is

clearly an issue of major importance (Nakashima et al. 2014).

Pipe elbows are critical components to the safety of pipe within

NG processing plants. Compared to straight pipe segments with the

identical cross section specification and material properties, the el-

bows are more flexible and associated with significantly higher

stresses, strains, and cross-sectional ovalization (Karamanos 2016).

To date, code prescriptions of seismic design for pipelines are gen-

erally scarce. In European Committee for Standardization (CEN)

Eurocode 8, for example, principal guidelines were provided for

above-ground pipeline in accordance with generic seismic design

approaches (CEN 2006). In terms of research, much effort has been

devoted on dynamic analysis of above-ground pipe elbows, mostly

in the form of cyclic bending analysis of individual elbow members

in which their failure mode under extreme loading conditions were

heavily investigated both numerically and experimentally. The

most reported elbow damage pattern was the axial development

1Ph.D. Candidate, Dept. of Civil Engineering, Univ. of Bristol, Queen’s

Bldg., University Walk, Bristol BS8 1TR, UK. Email: zz17635@bristol

.ac.uk

2Postdoctoral Research Fellow, Dept. of Civil and Mineral Engineering,

Univ. of Toronto, Toronto, ON, Canada M5S 1A4. Email: jamin.park@

utoronto.ca

3Professor, Dept. of Civil and Mineral Engineering, Univ. of Toronto,

Toronto, ON, Canada M5S 1A4. ORCID: https://orcid.org/0000-0002

-3292-9194. Email: os.kwon@utoronto.ca

4Professor, Dept. of Civil Engineering, Univ. of Bristol, Queen’s Bldg.,

University Walk, Bristol BS8 1TR, UK (corresponding author). ORCID:

https://orcid.org/0000-0002-2616-9395. Email: a.sextos@bristol.ac.uk

5Postdoctoral Research Fellow, Structures Laboratory, Dept. of Civil

Engineering, Univ. of Patras, Patras 26504, Greece. Email: ilstrepelias@

upatras.gr

6Postdoctoral Research Fellow, Structures Laboratory, Dept. of Civil

Engineering, Univ. of Patras, Patras 26504, Greece. ORCID: https://

orcid.org/0000-0002-5694-8779. Email: stathas@upatras.gr

7Professor, Director of Structures Laboratory, Dept. of Civil Engineer-

ing, Univ. of Patras, Patras 26504, Greece. Email: sbousias@upatras.gr

Note. This manuscript was submitted on May 17, 2020; approved on

September 8, 2020; published online on December 24, 2020. Discussion

period open until May 24, 2021; separate discussions must be submitted for

individual papers. This paper is part of the Journal of Pipeline Systems

Engineering and Practice, © ASCE, ISSN 1949-1190.

J. Pipeline Syst. Eng. Pract., 2021, 12(2): 04020073

of through-wall cracks near elbow flanks due to low cycle fatigue

accompanied by ratcheting effect (Hasegawa et al. 2008;Hassan

et al. 2015;Jeon et al. 2017;Karvelas et al. 2019;Nakamura

and Kasahara 2016,2017;Varelis et al. 2012;Watakabe et al.

2017), while evidence of local buckling followed by crack develop-

ment has also been mentioned (Hasegawa et al. 2008).

Investigations also went into complex pipe systems coupled

within industrial structures and plants. Reza et al. (2013) investi-

gated, by means of hybrid simulation (HS), the seismic perfor-

mance of a full-scale pipe system coupled within a single

industrial building. The pipe system is located on the top level

of a 3-story steel structure and is connected to a number of storage

tanks and devices inside the structure. Test results showed that even

with the maximum earthquake input under their investigation, the

pipeline system remained below the yield limit at all locations. The

authors of this paper believed that excessive strain development on

pipe was inhibited because all supporting points of the pipe was

contained on the same floor of the same structure, which is likely

to respond synchronously during the earthquake. We stress that this

would not necessarily be the case if different support points of the

pipe (in a single structure, between multiple structures, or between

a structure and the free field) vibrated out-of-phase.

For above-ground pipes, Sakai et al. (2013) evaluated the safety

of a piping system using HS, in which a 90° elbow was physically

tested and the remainder of the pipeline system was simulated in a

coupled numerical model. Permitting the assumption that one end

of the elbow specimen was fully fixed onto the laboratory floor, and

therefore had zero motion, the test concluded that the 8-in. diameter

uniform wall-thinning elbow can fail in the form of low cyclic

fatigue under certain conditions. Vathi et al. (2017) simulated the

seismic performance of a pipe system and its associated pipe rack

and liquid storage tank within an industrial plant. After undergoing

seismic excitation, it was found that the critical component of the

pipe system was the upper elbow located at the top of a pipe rack,

where local strain value exceeded the limit of severe plastification.

The differential motion between the ground surface and the struc-

tural response on top of the pipe rack where the pipe was elevated

made possible the asynchronous displacements at two ends of the

elbow, hence the conspicuous elbow in-plane bending. Sextos et al.

(2017) examined numerically the seismic performance of mechani-

cal subsystems within a nuclear power plant containment structure

using refined finite-element models. It was found that under certain

circumstances, elbows were susceptible to significantly increased

seismic demand if geometrical nonlinearities introduced by the ef-

fects of structure rocking and sliding with uplift are considered.

Wenzel et al. (2018) analyzed the nonlinear behavior of a coupled

foundation tank–pipeline system using HS, in which a liquid stor-

age tank and its base-isolated foundation were simulated numeri-

cally and a small portion of pipe connected directly to the tank was

tested physically. Under the assumption that the far end of the

physical pipe specimen was fully fixed to the laboratory floor, a

significant displacement time history was exerted onto the physical

pipe specimen during the HS. The result showed that the critical

component was one of the elbows located near heavy auxiliary

masses on pipe. Bursi et al. (2018) numerically evaluated the non-

linear response of a whole LNG plant under moderate seismic load-

ing. The study found elbows on top of the tall LNG storage tank

were the critical components in the loop and can exhibit a high

degree of vulnerability during transient ground motions. This

was because of the high differential displacement between the pipe

rack and the pump columns located over the dome of the stor-

age tank.

Similarly, for buried NG pipelines, the impact of out-of-phase

oscillation induced by differential earthquake inputs has been

highlighted previously. Psyrras et al. (2019,2020) numerically

and experimentally investigated the seismic risk of buried NG pipe-

lines when subjected to spatially varying transient ground deforma-

tions. Results showed that even for straight buried pipelines, the

seismic vibrations at the vicinity of laterally inhomogeneous sites

can produce differential movements on different locations of a long

pipeline due to kinematic soil-pipe interaction. As a result, appre-

ciable axial stress concentration can be observed in the critically

affected pipeline segment near the soil material discontinuity, high

enough to trigger coupled buckling modes into the plastic range.

Notwithstanding the aforementioned advancements, there are

still several clear limitations in the existing literature. While many

studies investigate the seismic demand of pipe coupled to its sur-

roundings, rarely had a set of realistic boundary conditions been

successfully adopted to a realistic seismic scenario in which differ-

ential displacement between the two pipe ends of an above-ground

pipe was significant. It was not uncommon in previous studies that

researchers employed overbold assumptions regarding boundary

conditions of the pipe supports, such as full constraint (zero mo-

tion) at one end of the investigated pipe, whereas in reality any pipe

support connected to another entity should have the corresponding

motion at the boundary. For example, the base of a pipe support

anchored onto the ground surface would be subjected to the

same excitation as the foundation input motion exerted to nearby

structures. This level of boundary condition accuracy is the (bare)

minimum that should be adopted in any modeling practices of

above-ground pipes, regardless of considering or neglecting any po-

tential coupling effect. Further more, the importance of structure

coupling in industrial NG plants due to the existence of pipes ex-

tending between them has not yet been addressed. It is unclear that

to what extent the negligence of structural coupling introduced by

bridging NG pipes can affect their design prospect. Should the in-

teraction leads to detrimental effects, whether buckling failure or

other forms of damage can occur on the pipes or the coupled struc-

tures during transient ground motion, awaits investigation. Finally,

how the coupling effect is influenced by the various properties of the

pipe and those of the supporting industrial structures, as well as the

characteristic of the input earthquake excitation remains in doubt.

Along these lines, the objectives of this paper are:

•To identify key parameters of the coupling problem within the

proposed structure-pipe-structure configuration, and illustrate the

sensitivity of both the global structural response and the induced

local elbow demand to these parameters and their most critical

combination, by means of finite-element analysis (FEA); and

•To experimentally examine the damage potential of pipe when it

is subjected to the differential displacement between two pipe

ends by means of HS, given that the difficulties associated with

numerically modeling geometrical nonlinearities of pressurized

pipe with buckling potential, the effect of nonnegligible struc-

ture-pipe-structure interaction, and the scale of the industrial

structures involved in the proposed scenario.

Problem Studied

The scenario examined herein consists of two realistic industrial

building configurations: supporting Structure A is a three-story

NG compressor house, and supporting Structure B is an exposed

platform topped with two tall and heavy reliquefication condensers

on its deck [Fig. 1(a)]. Both structures are steel moment-resisting

frames with reinforced concrete slabs and are assumed to behave

elastically. They have first-mode natural frequencies of fA¼

3.3Hz and fB¼2.3Hz, respectively; hence, a structural fre-

quency ratio is fB=fA¼0.7. There is an NG pipe behaving as

J. Pipeline Syst. Eng. Pract., 2021, 12(2): 04020073

a linking element between the two structures with a cross-sectional

diameter of 219.6 mm, wall thickness of 6.3 mm, and two 90° el-

bows with bend radius of 302.0 mm located in the middle. The

elbow bend factor is therefore h¼Rt=ðD=2Þ2¼0.16. The two

structures are laterally separated at a perpendicular distance of

7.56 m and the length of the intermediate straight pipe segment

between two 90° elbows is 1.38 m. A slender steel column supports

the pipe near the middle, merely providing vertical resistance. As

the overall pipe-structure system is subjected to ground excitation

along the x-axis, differential displacement between the two pipe

ends imposes compression or tension to the bridging pipe as the

two supporting structures vibrate out-of-phase, bending the pipe

elbows in-plane. This is the result of different dynamic responses

of the two structures even though the earthquake ground motion

they were subjected to is actually identical given their short sepa-

rating distance and the common foundation and underlying soil

profile.

Existing design criteria require that for the case in which a sec-

ondary system is attached to a primary system, the evaluation of the

coupling effect can only be neglected if the total mass of the inter-

acting secondary system is less than 1% of the primary supporting

structure (Fouquiau et al. 2018;Taghavi and Miranda 2008). How-

ever, it has also been pointed out that if the secondary system is

extended and supported at two or more locations, the coupling ef-

fect shall be investigated regardless of any mass percentage value

(Firoozabad et al. 2015). The authors believe cautiousness is even

more indispensable for the structure-pipe-structure configuration

proposed herein, in which the secondary system (i.e., the linking

NG pipe) is attached to two dynamic systems with divergent dy-

namic characteristics, and is therefore excited by the out-of-phase

oscillation between the latter. Fundamentally, if a decoupled analy-

sis is to be carried out for a partial structure, it is vital to ensure that

the decoupling does not significantly affect the frequencies and the

response of the primary system (Gupta and Tembulkar 1984). From

preliminary numerical analyses of the proposed structure-pipe-

structure system, it was observed that while the natural frequencies

of the two structures were not altered dramatically by the presence

or removal of the linking pipe element, a clear deviation of the

Fig. 1. Proposed structure-pipe-structure configuration: (a) detailed three-dimensional (3D) model; (b) simplified finite-element model of the

reference case; (c) illustration of varying Hp; (d) illustration of varying Lz; and (e) illustration of varying Lx. DOF = degree of freedom.

J. Pipeline Syst. Eng. Pract., 2021, 12(2): 04020073

structural response between the holistic case and the no-pipe case

was noted; thus, coupled analysis for the proposed scenario is ap-

propriate and necessary.

Identification of Key Problem Parameters and

Maximum Pipe Demand

Analysis Outline

For an in-depth study of the problem and to find realistic conditions

under which the seismic demand on the pipe becomes critical,

a parametric analysis scheme was established using the general-

purpose finite-element analysis software ABAQUS version 6.14.

To parameterize the proposed structure-pipe-structure scenario, the

two supporting structures were simplified to equivalent single-

degree-of-freedom (SDOF) oscillators topped with lumped mass

[Fig. 1(b)]. Due to the fact that the emphasis of the structure-

pipe-structure scenario investigated herein is on the eventual dam-

age of pipe elbows when the linking pipe element is subjected to

differential displacement between its two ends, the equivalent

SDOF oscillators were assumed to behave linear elastically and

were modeled using the ABAQUS Two-node Linear Beam In

Space Element, B31. The simplification preserved the elevation

of the pipe anchor point, the first-mode natural frequency of the

structures, the elastic swaying stiffness of the structures, and the

structural mass concentrated at the elevation of gravity center of

the corresponding detailed three-dimensional (3D) models. Fur-

thermore, given that the prototypes of the two supporting structures

are moment-resisting frames with axially stiff slab, which are ex-

pected to deform in shear during earthquakes, the stiff structure

floors onto which the linking pipe are extended and attached are

assumed parallel to the flat ground surface throughout the duration

of ground excitation. Therefore, the two pipe ends in the simplified

model were fixed to the oscillator at first five degrees of freedom

(DOF) and were free on the sixth DOF (i.e., rotation about z-axis).

Note that the simplified pipe connection may introduce a certain

degree of error in terms of the state of strain on pipe, as demon-

strated by Guarracino et al. (2009) both numerically and experi-

mentally for a four-point bent pipe. It was also assumed that the

ground excitation is limited to x-axis only and that the vibration

of the equivalent SDOF supporting structures were restricted in

the x,y-plane, which is the vibration direction of the dominant

first-mode response of the corresponding detailed 3D models.

Finally, we assumed that the base of the two equivalent SDOF sup-

porting structures were fully fixed to the ground and were always

subjected to the identical input ground excitation. On the other

hand, the linking pipe was modeled in a greater detail to capture

its potential buckling and nonlinear hysteretic response under

dynamic loading. The ABAQUS Four-node Reduced-Integration

Shell Element, S4R, was utilized for modeling the pipe geometry,

assigning plastic material properties with a linear kinematic hard-

ening rule. The mesh density on the elbows was set to 54 elements

around the cylinder circumference and 3,510 shell elements in total

for each 90° elbows. Coarser mesh was chosen for the straight pipe

segments as the excessive strain development and nonlinearities are

expected to concentrate on and around the elbows. The selected

type and size of shell element have been widely used in previous

pipe elbow modeling practices (Varelis et al. 2011;Vazouras et al.

2010) and were proven reliable through our preliminary analyses.

To verify the model simplifications, we define Ddiff ðtÞas the time

history of x-directional differential displacement between the two

pipe ends [Eq. (1)]

Ddiff ðtÞ¼uBðtÞ−uAðtÞð1Þ

where uAðtÞ;uBðtÞ= time variation of x-axis positions of pipe-end

Points A and B. Validation of the equivalent SDOF simplification

in its ability of reliably reproducing the structural displacement re-

sponses was demonstrated by a comparison of Ddiff ðtÞresults be-

tween a detailed 3D model, as shown in Fig. 1(a), and a simplified

model, as shown in Fig. 1(b). The Ddiff ðtÞresult produced by the

simplified model using equivalent SDOF oscillators compared well

with the corresponding Ddiff ðtÞobtained from detailed 3D model.

The parameters examined in the numerical parametric study are

identified in Table 1. In particular, fB=fAis the ratio of the first-

mode natural frequencies of Structures B and A, respectively. The

parameter fgis the predominant frequency of input excitation,

determined at the frequency where the highest peak occurs in its

fast Fourier-transform diagram. They jointly describe the funda-

mental dynamic mechanism of the out-of-phase oscillation between

the two supporting structures. Parameter Hpaims to capture the

Table 1. List of variables examined in the parametric study

Symbol Parameter description Range of variation

fB=fAThe ratio of the natural frequencies of supporting Structures B and A 0.30 to 2.73 (fA≡3.3Hz, fB¼1.0Hz to 9.0 Hz)

fgPredominant frequency of the input ground excitation 1.0–9.0 Hz

HpElevation of pipe-end attachment points on the structures 2.0–8.0 m

KA,KBLinear elastic equivalent SDOF swaying stiffnesses of the two

supporting structures

50%–90% of the reference case. At 100%:

KA¼52670 kN=m, KB¼16670 kN=m

LzLength of pipe between the two 90° elbows (Fig. 1) 1.5–3.0 m

LxPerpendicular distance between two structures (Fig. 1) 4.0–9.0 m

PPipe internal pressure 0–12.5 MPa

D(constant) Pipe cross-sectional diameter 219.6 mm

t(constant) Pipe wall thickness 6.3 mm

R(constant) Bend radius of the 90° elbows 302.0 mm

ρ(constant) Pipe material density 7.85 t=m3

E(constant) Pipe material elastic modulus 2.10 ×108kPa

ν(constant) Pipe material Poisson’s ratio 0.3

σy;1;ϵp;1;σy;2;ϵp;2

(constant)

Pipe material bilinear nonlinarity: yield stresses and plastic strains σy;1¼275000 kPa, ϵp;1¼0

σy;2¼650000 kPa, ϵp;2¼0.15

ζ(constant) Rayleigh damping 2%

Hs(constant) Elevation of the mass centers of the two SDOF supporting structures

in simplified models

5.3 m

ag(constant) Peak ground acceleration (PGA) of input ground motions 1.0 g

J. Pipeline Syst. Eng. Pract., 2021, 12(2): 04020073

amplification of pipe-end differential displacement due to higher

pipe elevation given the same structure, pipe, and excitation proper-

ties. The influence of pipe-structure stiffness ratio is also examined

through the reflecting variable of linear elastic equivalent SDOF

swaying stiffnesses of the two supporting structures KA,KB.

The length of the straight pipe segment between the two 90° elbows

Lzand the perpendicular distance between the two supporting

structures Lxare further varied to assess the impact of different

geometry of the structure-pipe-structure configuration. Finally,

internal pressure of the NG pipe Pis examined to consider the dif-

ferent operation conditions of a pressurized NG pipe. Other param-

eters, including pipe cross-sectional specifications, the geometry of

the pipe elbows, pipe material properties, structural damping ratio,

structure shape characterized by the elevation of its mass center,

and the peak ground acceleration (PGA) of the input excitation,

are taken as constants as the impacts of their variations are not

unique for the problem presented herein.

The value of variables in the reference finite-element (FE)

model is summarized herein as well as in Fig. 1(b). It has structural

frequencies fA¼2.2Hz and fB¼3.3Hz, hence fB=fA¼0.7.

The input ground motion is selected from the 1972 Nicaragua

earthquake recorded at Managua ESSO station, with a predominant

frequency fg¼2.2Hz (Input motion 4 in Fig. 4). The equivalent

SDOF swaying stiffness of the supporting structures are KA¼

52,670 kN=m and KB¼16,670 kN=m. Given the natural fre-

quency and the stiffness, the mass of the two supporting structures

in the reference case are MA¼122.5t and MB¼79.8t, respec-

tively. The geometry of the structure-pipe-structure system is

described with the parameters Hp¼5.30 m, Lz¼1.38 m, and

Lx¼7.56 m. The pipe internal pressure is P¼3.0MPa. Note that

in each of the following variations in the parametric study, only the

examined parameter(s) will be deviated from the reference case in

each section.

Structural and pipe response quantities examined throughout

the parametric study are Ddiff;max and ϵh;max . The peak differential

displacement between two pipe ends, Ddiff ;max, is defined as the

absolute value of the largest in time pipe-end differential displace-

ment Ddiff ðtÞthat occurred during the excitation

Ddiff;max ¼maxðjDdiff ðtÞjÞ;t¼0→tgð2Þ

where tg= total length of the input ground excitation. Ddif f;max

is therefore a nonnegative scalar derived for each analysis case,

providing insight into the level of global response of the coupled

structure-pipe-structure dynamic system. Similarly, the maximum

hoop strain on the elbows, ϵh;max, defined as the largest amplitude

of elbow hoop strain ϵhðtÞobtained during the excitation within the

90° bent elbows, represents the level of local seismic elbow strain

demand. Like any peak values, the Ddiff;max and the ϵh;max neither

reflect the time variation nor the potential cumulative character of

the response quantities. Nonetheless, as scalars, they provide

straightforward indications of the level of seismic demand to the

pipe and can be comprehended within the context of parametric

study.

Effect of Structural and Ground Motion Frequencies

To gain understanding into the frequency-dependency of the peak

differential displacement between two pipe ends Ddiff ;max and the

induced maximum elbow hoop strain ϵh;max, we examine the ratio

of first-mode natural frequencies of the two supporting structures

fB=fAand the frequency content of the input excitation character-

ized by its predominant frequency fg. Given that the purpose of the

study was not to explore fatigue or ground motion duration impact

on nonlinear response of the elbows, but to identify the effect of

ground motion frequency content on the developed hoop strains,

wavelet pulses were employed for analysis having an amplitude

of 1.0 g and a predominant frequency fgvarying from 1.0 Hz

to 9.0 Hz. More precisely, Ricker wavelets (Ricker 1943) were used

to excite a series of models with varying structural frequency ratio

fB=fAranging from 0.30 to 2.73 (corresponding to a variation of

fBranging from 1.0 Hz to 9.0 Hz while fAwas kept equivalent to

3.3 Hz), representing the pulse-like waveforms of acceleration in-

puts with a narrow frequency bandwidth (Fig. 2).

Inspection of Ddiff;max [Fig. 3(a)] and ϵh;max [Fig. 3(b)] results

over the variation of structural frequency ratio fB=fAreveals non-

zero responses in all cases except when fB=fA¼1.0. For these

cases, a minimum Ddiff;max response of around 25 mm and a mini-

mum ϵh;max response of around 0.1% exist even when none of the

structure natural frequencies fAand fBare close to the predomi-

nant frequency of excitation fg. This indicates that the out-of-phase

oscillation between two supporting structures can occur as long as

their first-mode natural frequencies of vibration are not identical.

On the pipe elbows, hoop strain develops accordingly during the

excitation as the elbows bend due to the differential motion exerted

between the two pipe ends.

Inspection of Ddiff;max and ϵh;max results over the variation of

input predominant frequency fgshows the fact that the responses

will reach local maximum values when resonance to the input ex-

citation occurs for at least one of the supporting structures. For all

models with a fB=fAvalue other than 1.0, wavelet excitation with

predominant frequencies fg¼3.0Hz and fg¼3.5Hz have led to

Ddiff;max response greater than 58 mm and ϵh;max response higher

than 0.51%, which can be attributed to the resonance of Structure A

(whose natural frequency fA≡3.3Hz) to these inputs. We note

that the responses observed in these cases are nearly constant as

long as Structure A is the only resonant supporting structure. A

series of peaks goes diagonal across the 3D plots [i.e., from the

point (fg¼1.0Hz, fB=fA¼0.3) to the point (fg¼9.0Hz,

fB=fA¼2.73)] reflect the resonance of Structure B (whose natural

frequency fBvaries between 1.0 Hz and 9.0 Hz) to the correspond-

ing wavelet excitation. These diagonal peaks are higher as the natu-

ral frequency of Structure B fBand the predominant frequency of

excitation fgare lower. Moreover, we notice that the Ddiff;max and

the ϵh;max responses further amplify when the natural frequencies of

the two supporting structures, fAand fB, are both close to the pre-

dominant frequency of excitation fg. Within the scheme of this

parametric study, the phenomenon is observed at the analysis case

fg¼3.0Hz and fB=fA¼0.92 (i.e., fA≡3.3Hz, fB¼3.04 Hz),

which results in Ddiff;max ¼120 mm and ϵh;max ¼0.78%. Notice

that this Ddiff;max value is almost exactly twice as much as the

Ddiff;max values observed in cases in which Structure A is the sole

supporting structure in resonance with the fg¼3.0Hz wavelet in-

put, whereas the ϵh;max value is intensified by around 50%.

Fig. 2. Illustration of acceleration time histories of Ricker wavelets

with varying predominant frequencies.

J. Pipeline Syst. Eng. Pract., 2021, 12(2): 04020073

Looking through all analysis cases, the global maximum values

of Ddiff;max and ϵh;max occur at the analysis case fg¼1.0Hz and

fB=fA¼0.30 (i.e., fA≡3.3Hz, fB¼1.0Hz), resulting in

Ddiff;max ¼274 mm and ϵh;max ¼2.32%. The aforementioned ob-

servations indicate that a combination of supporting structures with

low first-mode natural frequencies (either a single or both support-

ing structures) and ground excitation with a low predominant fre-

quency can lead to the onset of higher out-of-phase vibrations

between the two supporting structures, hence a higher seismic el-

bow strain demand.

Of course, real earthquake ground motions are typically rich in a

broader range of frequency contents. A selection of five earthquake

accelerograms with different predominant frequencies (Table 2and

Fig. 4) and their PGA scaled to ag¼1.0g were used to excite the

FE models with varying fB=fAvalues. Note that the intention

of this practice was not to extensively explore the impact of differ-

ent ground motions to the proposed structure-pipe-structure sce-

nario but to provide a proof that the observations gained from the

wavelet cases are also conceptually applicable for real ground mo-

tions. While similar trends can be qualitatively confirmed by in-

terpreting Figs. 3(c and d),Ddif f;max and ϵh;max values obtained

using ground motion inputs carry greater randomness. Overall,

a higher magnitude of responses can be observed due to the

much-longer duration of excitation, in which global maximum

values Ddiff;max ¼391 mm and ϵh;max ¼5.26% are indicated at

the analysis case fg¼1.2Hz (i.e., Input motion 5), fB=fA¼

0.30 (i.e., fA¼3.3Hz, fB¼1.0Hz). The reference case of

the parametric study, fg¼2.2Hz (i.e., Input motion 4) and

Fig. 3. Ddiff;max and ϵh;max responses with respect to variations of structural frequency ratio fB=fAand predominant frequency of excitation fg:

(a and b) Ricker wavelets; and (c and d) earthquake ground motions. Note the variation of fB=fAfrom 0.30 to 2.73 corresponds to fA≡3.3Hz and

fBvarying from 1.0 to 9.0 Hz.

J. Pipeline Syst. Eng. Pract., 2021, 12(2): 04020073

fB=fA¼0.70 (i.e., fA¼3.3Hz, fB¼2.3Hz), which is also the

case later tested in HS, is marked on the figures where Ddiff;max ¼

189 mm and ϵh;max ¼3.13% are predicted.

Effect of the Pipe-End Attachment Point Elevation

Given the same NG pipe, the same structural dynamic properties,

and the same ground motion input, the dynamic response of the

structure-pipe-structure system can be different depending on the

specific location; in particular, the elevation of the attachment points

where the two ends of the linking pipe element are connected to the

supporting structures. In the parametric study, the two pipe-end at-

tachment points have identical elevation and their variations are

assumed simultaneous so that a single parameter (i.e., pipe-end at-

tachment elevation Hp) is sufficient for describing the phenomenon.

A reasonable variation of Hpin the range of 2.0–8.0 m was consid-

ered to account for NG pipes connected at different heights between

two typical industrial structures in NG plants. Note that because the

two supporting structures are represented by equivalent SDOF os-

cillators, the simplified FE models have the limitation in replicating

the true profile of structural lateral deformation along their elevation

during the excitation. This means the analysis accuracy reduces

when the pipe-end attachment elevation Hpis far off from the eleva-

tion of mass centers of the SDOF supporting structures Hs, which is

equivalent to 5.3 m. Nevertheless, the impact of varying Hpto the

global and local system responses can be reflected.

Table 2. List of earthquake records used as input ground motions

Identifier Event Station and recorded direction Unscaled PGA (g) Predominant frequency fg(Hz)

1 San Fernando (1971) Santa Felita Dam (Outlet), 262 0.15 9.0

2 Northridge (1994) Lake Hughes #9, 90 0.26 4.9

3 San Fernando (1971) Castaic - Old Ridge Route, 21 0.32 3.0

4 Nicaragua (1972) Managua ESSO, 90 0.26 2.2

5 San Fernando (1971) Palmdale Fire Station, 120 0.11 1.2

Fig. 4. Acceleration time histories (a) and their fast Fourier-transform amplitudes; and (b) used as input ground motions. PF = predominant frequency.

J. Pipeline Syst. Eng. Pract., 2021, 12(2): 04020073

Within the range of examined pipe-end attachment elevation Hp

values, analysis result shows an linearly correlated relationship be-

tween the Hpand the Ddiff;max response [Fig. 5(a)]. The Ddiff;max

maxes out at 296 mm and reaches its minimum of 43 mm when the

pipe-end attachment elevation is Hp¼8.0m and Hp¼2.0m, re-

spectively. On the other hand, while the ϵh;max response [Fig. 5(b)]

also becomes larger as the Hpis larger, the correlation is not linear.

Compared to the reference case in which Hp¼5.3m, increasing

the Hpvalue by 2 m intensifies the ϵh;max output by no more than

16%, whereas decreasing the Hpvalue by the same amount results

in around 40% of reduction on the ϵh;max response.

Effect of the Structural Stiffness

An important property of the coupled structure-pipe-structure sys-

tem is characterized by the relative ratio between the stiffness of the

Fig. 5. Ddiff;max and ϵh;max responses with respect to variations of Hp,KAand KB,Lz,Lx, as well as P.

J. Pipeline Syst. Eng. Pract., 2021, 12(2): 04020073

linking pipe element and those of the supporting structures. Con-

sidering the fact that a total of two structures are involved in the

scenario with different swaying stiffness values and the nonlinear

behavior expected for the linking pipe element, we herein examine

only the variation of the linear elastic equivalent SDOF structural

stiffness and use it to conceptualize the variation of pipe-structure

stiffness ratio, instead of defining the ratio explicitly. A higher input

of structural stiffness indicates a lower pipe-structure stiffness ratio

and vice versa. In the parametric study, the stiffness of the two

structures, KA,KB, are assumed to vary simultaneously so that their

values in different analysis cases can be expressed using a single

percentage value with regard to the reference model. To give an

idea, the linear elastic equivalent SDOF swaying stiffnesses of

Structure A and Structure B in the reference case are KA¼

52670 kN=m and KB¼16670 kN=m, while the x-axis initial stiff-

ness of the linking pipe element is equivalent to 751.5kN=m. Ad-

ditionally, in order to reflect the case when pipe-structure stiffness

ratio is zero, we also analyzed the aforementioned models with the

linking pipe element being removed. Note that the variation of KA,

KBis always accompanied by a corresponding change of equiva-

lent SDOF masses of the two structures, so that the natural frequen-

cies of the two supporting structures are kept invariant in this

section.

The Ddiff;max [Fig. 5(c)] and ϵh;max [Fig. 5(d)] responses are

plotted against structure stiffness KA,KB. The Ddiff;max response

is higher for cases when structural stiffness is higher (i.e., lower

pipe-structure stiffness ratio), indicating a weakened coupling ef-

fect between the two supporting structures introduced by the

weaker linking pipe element. Also, we observe constant numerical

outputs of Ddiff;max ¼212 mm for all analysis cases when the

linking pipe element is removed, represented by the dotted line

on Fig. 5(c). The Ddiff;max curve continually approaches the dotted

line but does not meet it within the range of examined KA,KB. One

might perceive that the curve and the dotted line will never meet at

any finite value of structural stiffness as long as the linking pipe

element keeps its presence, hence its stiffness. A similar trend ap-

plies to the ϵh;max response as well. As the structure-pipe-structure

interaction is weakened due to a higher KA,KBinput, the elbow

strain demand correspondingly raises and approaches a theoretical

maximum value as the pipe-structure stiffness ratio approaches

zero. For the proposed structure-pipe-structure configuration, the

Ddiff;max result exceeds 95% of the no-pipe cases when structural

stiffness percentage exceeds 180% of the reference model, in which

case KA¼105340 kN=m and KB¼33330 kN=m. In such a case

when the pipe-structure stiffness ratio is lower than a certain level,

hence the Ddiff;max response does not clearly deviate from the cor-

responding no-pipe case, a coupled analysis is rendered unneces-

sary. This means predetermined structural responses from a no-pipe

case can be used as inputs to predict seismic demand of the linking

pipe element with acceptable accuracy. However, we stress that

a coupled analysis is always recommended in the preliminary

stage of any pipe-related research in which similar structure-pipe-

structure configurations are involved, so that the boundary condi-

tion of the pipe can be made realistic. Additionally, when deciding

whether a coupled analysis can be neglected, it would be good prac-

tice to inspect not only the peak value Ddiff;max but the quantity’s

full time variation Ddiff ðtÞwhen possible.

Effect of the Straight Pipe Length between Two Elbows

and the Perpendicular Distance between Two

Supporting Structures

The length of the straight pipe segment between the two 90° elbows

Lzand the perpendicular distance between the two supporting

structures Lxare variables describing the different geometry lay-

outs of the linking pipe element. The examined range of Lzwas

selected as 1.0–3.0 m and the range of Lxwas selected as 4.0–

9.0 m. Within these ranges the overall length and shape of the link-

ing pipe element is realistic for typical bridging pipes within NG

plants, while further complicating the parametric study by adding a

pipe rack is avoided. The two 90° elbows are assumed to always

locate exactly at the middle, forming the linking pipe element in a

symmetrical shape.

The Ddiff;max response are found to have a very weak depend-

ency on the variation of Lz[Fig. 5(e)] as the deviation of responses

between all cases are less than 10%. On the other hand, the ϵh;max

response reduces significantly as the Lzis larger [Fig. 5(f)], there is

a 50% reduction on elbow strain demand as the Lzincreases from

the reference case of 3.0 m. The similar global displacement re-

sponses of the structures with respect to different Lzinputs is be-

cause the x-axis stiffness of the linking pipe element remains

almost unchanged within the range of Lzvariation. Meanwhile,

a shortened Lzmeans that the elbows are subjected to a larger bend-

ing angle, hence a much higher local elbow strain demand. On the

other hand, the variation of Lxhas led to almost constant Ddif f;max

[Fig. 5(g)] and ϵh;max [Fig. 5(h)] responses, in which the deviation

between all cases is less than 3% and 10%, respectively.

Effect of Pipe Internal Pressure

NG is usually highly compressed for its transmission through

pipes. In the parametric study, pipe internal pressure Pin the range

from zero to 80% of the reference pipe’s nominal yield stress,

80%py¼80%ð2σy;1t=DÞ≈12.5MPa, was examined. Note that

the uniqueness of variable Pfrom the rest of the variables examined

in the parametric study is that the internal pressure affects both the

demand and the capacity of the pipe. The collapse moment of pipe

elbows is known to increase with higher internal pressure up to a

certain threshold value, and then decrease with further increasing of

internal pressure. Previous research showed that for the 90° elbows

with bend factor h≈0.16, this threshold value is around P¼

10.5MPa when the elbows are subjected to closing bending mo-

ment (Shalaby and Younan 1998).

The peak differential displacement between two pipe ends

Ddiff;max is lower as a result of higher Pinput [Fig. 5(i)]. This

is because an increased pipe internal pressure can lead to higher

x-axis stiffness of the linking pipe element, hence a more pro-

nounced structure-pipe-structure interaction to mitigate the differ-

ential displacements between two supporting structures. However,

although the Ddiff;max response tends to reduce with higher Pso

that effectively the 90° elbows are less bent during the excitation, its

benefit to the alleviation of seismic elbow strain demand is com-

pletely suppressed by the presence of the higher pipe internal pres-

sure itself. The maximum elbow hoop strain ϵh;max soars as the P

input increases [Fig. 5(j)].

Experimental Setup: Hybrid Simulation

System Substructuring Scheme

Considering the nonnegligible structure-pipe-structure interaction

and the size of the interactive industrial structures, HS is believed

a necessary and efficient way for experimentally investigating the

buckling potential and the detailed nonlinear hysteretic behavior of

the linking pipe element as well as the interactive response of the

overall system. We developed a HS scheme based on the reference

model used in the parametric study, in which a physical specimen

of the linking pipe element, consisting of three straight pipe

J. Pipeline Syst. Eng. Pract., 2021, 12(2): 04020073

segments and two 90° elbows, was tested at the Structures Labo-

ratory of the University of Patras, Greece, whereas the complemen-

tary part involving the two supporting structures was solved

numerically (Park et al., forthcoming). A total of two substructure

modules, as illustrated in Fig. 6and that will be discussed in the

following sections, were therefore configured for the HS in order to

investigate the coupled response of the structure-pipe-structure

system.

The generalized HS framework, UT-SIM (second edition), de-

veloped by the University of Toronto research group (Huang and

Kwon 2018;Mortazavi et al. 2017), was used for integrating the

numerical and experimental substructures. The UT-SIM framework

employs the University of Toronto Networking Protocol (UTNP)

for communication while a software library provides useful func-

tions in exchanging data between diverse numerical and experi-

mental models. The generalized nature of UT-SIM framework

assigns each substructure module with an interface of communica-

tion. In this HS scheme, OpenSees computational platform version

2.4.3 (rev 5645) (Mazzoni et al. 2006) was selected to perform the

analysis tasks of solving both the numerical substructure and the

main integration algorithm. Therefore, an OpenSees user defined

element termed SubStructure was featured to the numerical model

to collect the required restoring force through UTNP. On the other

hand, a software called the network interface for the controller

(NICON) (Zhan and Kwon 2015), based on the LabView program-

ming environment and National Instrument hardware, allows

communication, coordinate conversion, analog voltage generation,

data transmission, and acquisition of the physical substructure

module.

During the entire HS, the numerical integration algorithm cal-

culates a set of command displacements (uA;uB) at each analysis

time step. This digital information is passed from the numerical

integration module to the physical substructure module via the

UTNP-utilizing communication interfaces featured in the UT-SIM

framework, consisting of the SubStructure OpenSees element for

the former and the NICON software for the latter. The calculated

structural displacements are received by the NICON and go

through a simple calculation which turns them into the differential

displacement to be imposed to the physical pipe specimen

[Ddiff ðtÞ¼uAðtÞ−uBðtÞ], before the command is then converted

into an analog voltage signal employing National Instrument data

acquisition digital-to-analog conversion hardware. The generated

displacement command is subsequently interpreted by a modular

actuator controller unit, which drives the unidirectional hydraulic

actuator using proportional–integral–derivative (PID) control. Given

an accurate actuation control and the fact that strain rate effect due to

seismic motion is believed to have little influence on the material

stress-strain behavior of pipe (Yoshizaki et al. 2000), the overall

structure-pipe-structure system is subjected to deformations and

damage equal to those a real earthquake would generate. The struc-

tural response parameters (actuator force and actually imposed

displacement) are not known in advance and are measured and re-

corded during the HS via the actuator load cell (243.45 Actuator,

MTS Systems, Eden Prairie, Minnesota) and a high-resolution

Temposonics transducer (MTS Systems, Eden Prairie, Minnesota)

attached on the actuator. The restoring force and displacement

responses of the physical substructure acquired at the current time

step are converted to digital form and fed back to the numerical

Fig. 6. Substructuring scheme and key components of the HS. NICON = network interface for the controller; UTNP = University of Toronto

networking protocol; and EQ = earthquake.

J. Pipeline Syst. Eng. Pract., 2021, 12(2): 04020073

integration algorithm. Eventually the current time step is completed

and HS progresses to the next one until full completion of the ex-

citation record.

Numerical Integration Module

The numerical integration module of the proposed HS scheme

contains the FE model representing the two supporting structures

in the proposed structure-pipe-structure scenario and handles the

main integration algorithm for dynamic time history analysis.

The linear elastic equivalent SDOF modeling approach used in

the parametric study for the two supporting structures was trans-

plant to the HS in the OpenSees numerical integration module.

By employing the modeling approach and applying the identical

assumptions and boundary conditions to the numerical substruc-

ture, we ensure the compatibility and equilibrium condition at

the numerical-physical coupling DOF (i.e., x-axis differential mo-

tion between the two pipe ends) so that a single unidirectional ac-

tuator is sufficient for applying the appropriate boundary condition

to the physical substructure module during HS. Note that in the

OpenSees model, the equivalent SDOF columns were modeled us-

ing the Elastic Timoshenko Beam-Column Element, which employs

the same beam theory as the ABAQUS B31 element used in the

parametric study models. A dedicated SubStructure element was

defined, acting as the interface of communication for the numerical

substructure. Furthermore, supporting structure properties em-

ployed in the numerical integration module were identical to those

of the reference case in the parametric study. Structure A had a

natural frequency of fA¼3.3Hz and mass of MA¼122.5t;

Structure B had a natural frequency of fB¼2.3Hz and mass of

MB¼79.8t. The numerical model was assigned with 2% Ray-

leigh damping and alpha-operator splitting method (Combescure

and Pegon 1997) was used as the integration algorithm. The total

analysis period of the HS was set to 10 s, corresponding to a total of

1,000 time steps as the step size was selected to be 0.01 s. The first

7 s of the accelerogram from the 1972 Nicaragua Earthquake

recorded at the Managua ESSO station (i.e., Input motion 4 as

shown in Fig. 4), with a predominant frequency of 2.2 Hz and the

PGA scaled to 1.0 g, was used as the input ground motion to the

numerical integration module in HS.

The effectiveness of the numerical integration module was veri-

fied by a series of OpenSees-ABAQUS multiplatform simulations

prior to the actual HS, in which the physical substructure module

containing the linking pipe specimen was represented by an

ABAQUS numerical replacement. Accuracy and stability of the in-

tegration process, effectiveness of the OpenSees numerical model,

and the smooth operation of the associated UT-SIM framework

components which cooperate with the numerical substructure were

double-checked.

Physical Substructure Module

The physical substructure module is composed of the full-scale

physical specimen of the linking NG pipe and relevant accessories

including a hydraulic actuator, an actuator controller, the HS in-

terface software NICON that links the actuator controller to the

numerical model, the measuring instruments, and a data acquisition

system.

The physical NG pipe specimen residing at the Structures Lab-

oratory of the University of Patras, Greece (Fig. 6), includes three

straight pipe segments welded in situ on two 90° elbows with a

cross-sectional diameter (D) of 219.6 mm and a pipe wall thickness

(t) of 6.3 mm. The length of the straight pipe segment located be-

tween the two elbows (Lz) is 1.38 m, the length of the other two

straight pipe segments is 3.48 m and the bend radius of the 90°

elbows (R) equals 302.0 mm. Therefore, the perpendicular distance

between the two structures (Lx) is 7.56 m, the pipe nondimensional

geometry parameters are R=D¼1.38,D=t¼34.86 and the elbow

bend factor is h¼Rt=ðD=2Þ2¼0.16. Before the HS, a water

pumping system applied pipe internal pressure (P) of 3.0 MPa

to the pipe specimen, accounting for the compressed NG inside

the pipe. In the laboratory, the pipe specimen was rigidly clamped

onto the strong laboratory floor through a triangular connector at

one end, while its other end was attached to a unidirectional actua-

tor. A low-friction guiding device was set up around the straight

pipe segment near the actuator side, limiting the actuator’s move-

ment along the x-axis. This was to ensure the SDOF equilibrium

condition at the numerical-physical coupling node so that the errors

introduced at the interface between two HS substructures were

minimized. On the other hand, because the pipe specimen was

in contact with the guiding device, a contact force with unknown

magnitude was inevitably included as part of the specimen restor-

ing force since potential horizontal or vertical pipe inclination

might occur during the HS. As a result, a small error may still exist

in the force feedback to the numerical integration module at every

analysis time step, which could harm the accuracy of HS result.

While all contacting surfaces between the pipe and the guiding de-

vice were covered with polytetrafluorethylene sheets and were

highly lubricated to reduce friction, the upper-half of the guiding

device was also left rather loose to further reduce the impact of

contact/pipe inclination to its minimum. Two strip supports with

polytetrafluorethylene and lubricated flat surfaces were placed

under the middle of the pipe specimen to provide pipe constraint

in the vertically downward direction, simulating the single-column

pipe support in the proposed structure-pipe-structure configura-

tion. By doing so, initial pipe flexure due to its self-weight was

prevented.

The effectiveness of the laboratory setup was validated to ensure

that the presence of the auxiliary gears do not obstruct the validity

of our model assumptions. Evidence obtained during and after pre-

liminary nondamaging HS showed a minor effect of the contacting

force originated from the restraint device and the strip supports. We

noted that the pipe specimen started to deform elastically at a very

small applied actuator displacement, indicating a very small un-

wanted contacting force in the laboratory setup. The force is exper-

imentally estimated at less than 2% of the maximum restoring force

that the linking pipe specimen would experience during the full-

amplitude HS. Given the previous discussion, it is concluded

that experimental results obtained from the HS laboratory setup

are valid.

Still, if not tuned properly, the HS setup can generate erroneous

results for various other reasons. These may include the working

frequencies and amplitudes of the physical setup, the condition of

the tested specimen, the actuator, the control device, the type of

chosen control algorithm and its parameter setting, as well as

the selected size of the ramp and hold periods for each analysis

time step (Molina et al. 2011). Hence, the effectiveness of the

physical substructure module was further optimized and verified

by a series of nondestructive HS prior to the actual HS. Firstly,

through trial and error, appropriate parameters of the PID controller

as well as the allowable velocity of the actuator were determined so

as to confront the noise in the reference (analog) signal due to ana-

log-to-digital (A/D) conversion and to minimize control error. By

the same token, low actuator responsiveness yields the need for

longer stabilization period at the end of the ramp and an appreciable

hold period for averaging an adequate number of restoring force

sample values (analog). As a result, testing wall-clock time in-

creased considerably. In the trade-off between HS accuracy and

J. Pipeline Syst. Eng. Pract., 2021, 12(2): 04020073

time efficiency, an appropriate maximum actuator speed was se-

lected equal to 1mm=s, and a waiting period of 5 s was used after

each execution of command displacement so as to reduce the

undesirable fluctuation of forces. An averaged measurement of

reaction force was designed to be taken in a period of 2 s after

the 5-s waiting period, so that the noise-to-signal ratio in the meas-

urement can be further reduced. Such a configuration ensures the

numerical integration module to get representative force and dis-

placement feedback from the physical substructure module, while

producing a relatively reasonable 6-h HS execution time. More-

over, the initial stiffness of the physical NG pipe specimen, which

is required as an input variable for the alpha-operator splitting

numerical integration algorithm, was experimentally measured.

Instrumentation

Strain Gauges

A 16-channel data acquisition system for strain measurement

was used for the test. The strain gauges were installed as shown in

Fig. 7(a). The four locations with significant strains on a half-elbow

were identified based on numerical analyses. Because of the pos-

sible out-of-plane deformation of the pipe and the existence of the

restraint device, the pipe specimen may behave unsymmetrically

despite its symmetrical geometry. Thus, all four half-elbows were

instrumented with strain gauges at the same locations to ensure the

measurement of maximum strains on the elbows.

Ovalization Measuring Devices

Two special-purpose ovalization measuring devices (one per el-

bow) with LVDTs were used in order to measure the development

of cross-sectional ovalization on the elbows [Fig. 7(b)]. The main

body of the ovalization measurement device is a light steel

frame that is in contact with the elbow at four points along the

perimeter of a single cross section; the frame is welded to the

elbow at its intrados, while displacement measurements are taken

at the elbow’s extrados and two flanks. The steel frame itself is

considered rigid, allowing the LVDTs to be pressed against the

elbow wall, thus obtaining the correct measurement of elbow

cross-sectional diameter change, or flattening (Varelis et al. 2012),

at two perpendicular pipe diameters. The devices are installed

in the middle (45°) section of the elbows, where maximum elbow

flattening was predicted. Preliminary numerical analysis also

proofed a negligible impact to the pipe responses brought by

the welded ovalization devices.

Observations and Results from Hybrid Simulation

The examined linking NG pipe element showed a favorable perfor-

mance under the specific structure-pipe-structure configuration and

the input earthquake ground motion, during which no leakage was

observed. The HS confirmed the minor influence of the contacting

force originated from the auxiliary restraint device and the strip

supports, and no out-of-plane deformation of the pipe specimen

was observed during the HS.

Differential Displacement Time History and

Force-Displacement Relationship

Time history of differential displacement between the two pipe

ends, Ddiff ðtÞ, and force-displacement relationship of the linking

NG pipe (Fig. 8) obtained from the HS (HS holistic) and its cor-

responding numerical model in ABAQUS (FEA holistic) are com-

pared to gain insight into the hysteretic response of the linking

pipeline element. The Ddiff ðtÞresponse represents the system re-

sponse on a global level, whereas the force-displacement curve re-

veals evident hysteresis behavior of the pipe. The DdiffðtÞresponse

of the HS is very similar to its finite-element analysis (FEA)

counterpart in the first-half of the ground motion, leading to iden-

tical peak differential displacement between two pipe ends at

Ddiff;max ¼189 mm. As the input excitation gradually dies down,

more discrepancy between the two curves are observed. It is also

noted that while the FEA holistic model of the structure-pipe-

structure system predicted well the amplitude of relative displace-

ment time history, the result obtained from HS showed a slightly

higher vibration frequency. Additionally, the Ddiff ðtÞresponse

of standalone structures without being coupled by the linking

pipe element (FEA no pipe) is presented, showing the impact of

structure-pipe-structure interaction.

Fig. 7. Instrumentation: (a) strain gauges; and (b) ovalization device.

J. Pipeline Syst. Eng. Pract., 2021, 12(2): 04020073

Strains on Elbows

Four groups of strains were monitored at the top and bottom sur-

faces of the two elbows to account for the possible unsymmetrical

behavior of the pipe specimen during the HS due to the presence of

the constraint device. Despite this concern, test results were found

to be similar on the two elbows. Critical hoop strain measurement

from the HS, ϵhðtÞ, sampled physically at the locations correspond-

ing to where the maximum elbow hoop strain was observed in

numerical predictions, is plotted in Fig. 9. We note that measure-

ment on the bottom side of Elbow 1 shows zero strain and the meas-

urement on the top side of Elbow 2 saturates after 5 s into the HS.

These errors can be attributed to a detached or a damaged gauge.

Ratcheting effect of strain development is clear from both HS re-

sults and FE prediction. The maximum elbow hoop strain during

the HS occurred on the top side of Elbow 1 (i.e., the one close to the

actuator) at ϵh;max ¼3.49%, whereas the FE predicts ϵh;max ¼

3.13%. Moreover, the ratchet effect of elbow hoop strain develop-

ment recorded during the HS was only approximately captured by

FEA, in which the time spot when ϵh;max occurs and the general

trend of strain development was quite different.

Cross-Sectional Ovalization

Cross-sectional ovalization is quantified and visualized in the form

of cross-sectional flattening, i.e., the change of elbow diameter in a

certain direction [Fig. 10(a)]. The horizontal and vertical cross-

sectional flattening on both elbows were compared against the

numerical prediction. The hybrid simulation result shows a less

significant permanent cross-sectional flattening at the vertical di-

rection when compared to the numerical prediction: at around

2.5–3.5 s on the time history, the center line of the FEA-vertical

curve shifted upward with a magnitude of about 5 mm, while main-

taining a similar level of vibration compared to the HS result. The

same trend was also observed through an inspection of the corre-

sponding flattening-Ddiff curve in Fig. 10(b). Because the ovali-

zation results are similar on both elbows, only those from one of

the elbow are shown.

Concluding Remarks

In this paper, the seismic performance of a coupled structure-pipe-

structure system typically found in an NG processing plant was

Fig. 8. (a) Ddiff ðtÞresponses; and (b) force-displacement curves obtained from the HS and the corresponding FE prediction. The DdiffðtÞresponse of

a no-pipe case is also plotted to reflect the impact of structure-pipe-structure interaction.

Fig. 9. ϵhðtÞresponses on four half-elbows of the HS and the corresponding FE prediction, obtained at the significant strain location where maximum

hoop strain occurred in FE prediction.

J. Pipeline Syst. Eng. Pract., 2021, 12(2): 04020073

assessed by means of hybrid simulation (HS). A parametric study

was firstly performed based on simplified FE models and the HS

was conducted in which the holistic system was simulated as two

coupled substructures so the linking NG pipe can be modeled

physically in full-scale. Reliable boundary conditions of the linking

NG pipe were established in our investigation by modeling

explicitly the two supporting structures and analyzing the coupled

structure-pipe-structure system as a holistic integer. Although the

x-axis stiffness of the linking pipe element is much smaller than the

swaying stiffness of the supporting structures, simulation results

show a clear coupling effect introduced by its presence. Globally,

differential displacement between the two supporting structures re-

duces up to 40% through the duration of the ground excitation for

the reference analysis case due to the presence of the linking pipe

element, indicating that the structure-pipe-structure interaction

should not be overlooked for the proposed scenario and an HS

is necessary for capturing the interaction experimentally. Locally,

the connection of the linking pipe element to a total of two support-

ing structures with distinguishing dynamic properties contributes

to the differential displacement between two pipe ends, hence the

pronounced out-of-phase oscillation. As a result, the two pipe el-

bows can be bent severely into their nonlinear range.

The triggering factors of critical seismic demand for the linking

NG pipe in the proposed structure-pipe-structure scenario are sum-

marized as follows:

•The simultaneous mobilization of divergent structural oscilla-

tion at the low-frequency range between the two supporting

structures: It leads to high responses both globally and locally,

which can be attributed to the adverse combination of variable

fB=fA,fg, and Hp(i.e., natural frequency ratio of the support-

ing structures, predominant frequency of excitation, and the

elevation of pipe-end attachment points). Peak differential dis-

placement between two pipe ends, Ddiff;max , and maximum lo-

cal elbow hoop strain, ϵh;max, are at their highest when the two

supporting structures have different natural frequencies both in

the low-frequency range and are both resonant to the input

ground motion. Meanwhile, the elevation of pipe connecting

points directly affects how much the linking pipe element

can be exposed to the generated differential displacements be-

tween the two supporting structures, given the same input earth-

quake, the same structures, and the same pipe properties.

•A lower relative stiffness of the linking pipe element with re-

spect to that of the structures: In general, a lower pipe-structure

stiffness ratio leads to a lower structure-pipe-structure interac-

tion, hence higher Ddiff;max and ϵh;max responses.

•Adverse geometry characteristic of linking pipe element: As the

length of the straight pipe segments varies, the stiffness of the

linking pipe element is barely affected. However, given the sim-

ilar Ddiff;max responses, a linking pipe element with shorter

straight pipe segments means that the two 90° elbows are

more susceptible to bending, hence a higher seismic elbow

strain demand.

•A higher pipe internal pressure P: The existence of pipe internal

pressure naturally introduces a static load on the pipe, and

hence, increases the elbow strain demand. Its potential benefit

in mitigating excessive Ddiff;max response, thanks to the simul-

taneous pipe stiffness increase, is completely overshadowed by

the increased pressure load itself.

Adjusting these variables to the unfavorable side can lead to

significantly increased seismic demand to the pipe elbows. It is

particularly true for the variables fB=fAand fg,Hp, as well as

P. Even when configuring only one of these variables from

the reference scenario, which is a typical industrial site and is there-

fore deemed as a probable scenario, to a worst-case scenario, in

which maximum responses were observed from the parametric

study, can easily result in at least a 30% increase of the elbow strain

demand.

Compared with the corresponding FE predictions, HS results

show a peak differential displacement response between the two

pipe ends of 189 mm and a maximum elbow hoop strain of 3.49%

for the reference model. Pipe yielding, material plasticity, and strain

ratcheting were observed on the elbows together with a clear asym-

metric hysteretic behavior of the linking pipe element, which is

mainly due to the nonlinear geometry of the linking pipe element

Fig. 10. Pipe cross-sectional ovalization response: (a) horizontal and vertical cross-sectional flattening versus time curves on Elbow 2 from the HS

and the corresponding FE prediction; and (b) horizontal and vertical cross-sectional flattening versus Ddiff curves on Elbow 2 from the HS and the

corresponding FE prediction.

J. Pipeline Syst. Eng. Pract., 2021, 12(2): 04020073

and the ovalization of pipe cross section. On the other hand, pipe

buckling or a loss of containment event did not occur under this

level of elbow strain demand generated during the 7-s earthquake

excitation of the HS.

Overall, the present study shows that the structure-pipe-structure

interaction should not be overlooked, and we recommend a coupled

analysis to be considered at least in the preliminary stage of future

pipe-related studies in which similar structure-pipe-structure con-

figurations are involved in a time history analysis, so that the im-

posed displacements at the boundary of the pipe segment are

considered appropriately.

While numerical observations and the HS results were satisfac-

tory, the study has some limitations, and thus, future works can be

devoted to the following four directions. First, due to limited labo-

ratory resources, the authors were only able to execute a single HS,

which corresponds to the reference case of the parametric study.

The limited stroke of the available hydraulic actuator also meant

that an HS for any analysis case with a potentially higher dynamic

structural response was not possible. Second, the presented numeri-

cal and experimental studies did not account for structural nonli-

nearity, soil-structure interaction, or a sophisticated modeling of

pipe-end connection. Also, only a basic constitutive model was em-

ployed in the FE analyses to simulate the cyclic nonlinearity of the

pipe steel material. Efforts can be put into those directions in the

future, making use of refined FE models, to examine in greater de-

tail the proposed structure-pipe-structure scenario. Third, investiga-

tions involving an extensive collection of earthquake ground

motions can be made to derive fragility curves for the proposed

scenario and gain deeper insight to the dynamic nature of the sys-

tem in a probabilistic manner. Finally, given that the strain develop-

ment on a steel pipe elbow is cumulative with regard to the total

number of excitation cycles it undergoes, future work can be done

to account for the effect of multiple earthquake events on the

seismic elbow strain demand and its possible damage modes. This

will address the potential seismic threat to steel NG pipe elbows

in which they fail because of low cycle fatigue during their entire

service life span after experiencing a number of strong ground

motions, despite an immediate loss of pipe integrity in the form

of buckling failure may not happen during a single earthquake

event.

Data Availability Statement

Some or all data, models, or code that support the findings of this

study are available from the corresponding author upon reasonable

request. Items included are the ABAQUS finite-element models

and scripts used in the parametric study, the numerical data from

the parametric study, the OpenSees finite-element model used

in the hybrid simulation, files and software associated with the

UT-SIM framework used in the hybrid simulation, and the exper-

imental data from the hybrid simulation.

Acknowledgments

This work was funded by the Horizon 2020 Programme of the

European Commission through Grant MSCA-RISE-2015-691213-

EXCHANGE-Risk (Experimental & Computational Hybrid

Assessment of Natural Gas Pipelines Exposed to Seismic Risk,

www.exchange-risk.eu). The first author also expresses his grati-

tude to the China Scholarship Council for financially supporting

his doctoral studies (Grant No. 201808060061).

References

Bursi, O. S., R. di Filippo, V. La Salandra, M. Pedot, and M. S. Reza.

2018. “Probabilistic seismic analysis of an LNG subplant.”J. Loss Prev.

Process Ind. 53 (May): 45–60. https://doi.org/10.1016/j.jlp.2017.10

.009.

CEN (European Committee for Standardization). 2006. Eurocode 8:

Design of structures for earthquake resistance—Part 4: Silos, tanks

and pipelines eurocode. EN 1998-4:2006. Brussels, Belgium: CEN.

Combescure, D., and P. Pegon. 1997. “α-Operator splitting time integration

technique for pseudodynamic testing error propagation analysis.”Soil

Dyn. Earthquake Eng. 16 (7–8): 427–443. https://doi.org/10.1016

/S0267-7261(97)00017-1.

DOE. 2017. Understanding natural gas and LNG options. Washington,

DC: DOE.

Firoozabad, E. S., B.-G. Jeon, H.-S. Choi, and N.-S. Kim. 2015. “Seismic

fragility analysis of seismically isolated nuclear power plants piping

system.”Nucl. Eng. Des. 284 (1): 264–279. https://doi.org/10.1016/j

.nucengdes.2014.12.012.

Fouquiau, P.-V., F. Barbier, and C. Chatzigogos. 2018. “New dynamic de-

coupling criteria for secondary systems.”In Proc., 16th European Conf.

on Earthquake Engineering. Istanbul, Turkey: European Association

for Earthquake Engineering.

GIE (Gas Infrastructure Europe). 2015. Gas infrastructure Europe security

risk assessment methodology. Brussels, Belgium: GIE.

GIE (Gas Infrastructure Europe). 2019. Gas infrastructure Europe LNG

map 2019. Brussels, Belgium: GIE.

Guarracino, F., A. Walker, and A. Giordano. 2009. “Effects of boundary

conditions on testing of pipes and finite element modelling.”Int. J.

Press. Vessels Pip. 86 (2–3): 196–206. https://doi.org/10.1016/j.ijpvp

.2008.09.009.

Gupta, A. K., and J. M. Tembulkar. 1984. “Dynamic decoupling of secon-

dary systems.”Nucl. Eng. Des. 81 (3): 359–373. https://doi.org/10.1016

/0029-5493(84)90282-6.

Hasegawa, K., K. Miyazaki, and I. Nakamura. 2008. “Failure mode and

failure strengths for wall thinning straight pipes and elbows subjected

to seismic loading.”J. Pressure Vessel Technol. 130 (1): 1–8. https://doi

.org/10.1115/1.2826425.

Hassan, T., M. Rahman, and S. Bari. 2015. “Low-cycle fatigue and

ratcheting responses of elbow piping components.”J. Pressure Vessel

Technol. 137 (3): 1–12. https://doi.org/10.1115/1.4029068.

Huang, X., and O.-S. Kwon. 2018. “A generalized numerical/experimental

distributed simulation framework.”J. Earthquake Eng. 24 (4):

682–703. https://doi.org/10.1080/13632469.2018.1423585.

Jeon, B.-G., S.-W. Kim, H.-S. Choi, D.-U. Park, and N.-S. Kim. 2017. “A

failure estimation method of steel pipe elbows under in-plane cyclic

loading.”Nucl. Eng. Technol. 49 (1): 245–253. https://doi.org/10

.1016/j.net.2016.07.006.

Karamanos, S. A. 2016. “Mechanical behavior of steel pipe bends: An

overview.”J. Pressure Vessel Technol. 138 (4): 1–15. https://doi.org/10

.1115/1.4031940.

Karvelas, C., N. Stathas, E. Strepelias, X. Palios, G. Chatzopoulou, S.

Bousias, and S. Karamanos. 2019. “Strong cyclic loading of an indus-

trial piping system.”In Proc., 2nd Int. Conf. on Natural Hazards &

Infrastructure. Athens, Greece: National Technical Univ. of Athens.

Lanzano, G., F. Santucci de Magistris, G. Fabbrocino, and E. Salzano.

2015. “Seismic damage to pipelines in the framework of Na-Tech risk

assessment.”J. Loss Prev. Process Ind. 33 (Jan): 159–172. https://doi

.org/10.1016/j.jlp.2014.12.006.

Mazzoni, S., F. McKenna, M. H. Scott, and G. L. Fenves. 2006. Open

system for earthquake engineering simulation (OpenSees)—OpenSees

command language manual. Berkeley, CA: Univ. of California,

Berkeley.

Molina, F. J., G. Magonette, P. Pegon, and B. Zapico. 2011. “Monitoring

damping in pseudo-dynamic tests.”J. Earthquake Eng. 15 (6): 877–

900. https://doi.org/10.1080/13632469.2010.544373.

Mortazavi, P., X. Huang, O.-S. Kwon, and C. Christopoulos. 2017. Exam-

ple manual for University of Toronto simulation (UT-SIM) framework.

An open-source framework for integrated multi-platform simulations

for structural resilience. 2nd ed. Toronto, Canada: Univ. of Toronto.

J. Pipeline Syst. Eng. Pract., 2021, 12(2): 04020073

Nakamura, I., and N. Kasahara. 2016. “Trial model tests with simulation

material to obtain failure modes of pipes under excessive seismic loads.”

In Proc., ASME 2016 Pressure Vessels and Piping Conf.,1–11.

New York: ASME. https://doi.org/10.1115/PVP2016-63422.

Nakamura, I., and N. Kasahara. 2017. “Excitation tests on elbow pipe

specimens to investigate failure behavior under excessive seismic

loads.”J. Pressure Vessel Technol. 139 (6): 1–11. https://doi.org/10

.1115/1.4037952.

Nakashima, M., O. Lavan, M. Kurata, and Y. Luo. 2014. “Earthquake en-

gineering research needs in light of lessons learned from the 2011

Tohoku earthquake.”Earthquake Eng. Eng. Vibr. 13 (1): 141–149.

https://doi.org/10.1007/s11803-014-0244-y.

Park, J., Z. Zhang, O. Kwon, S. Bousias, and A. Sextos. Forthcoming.

“Hybrid testing for assessing the coupled seismic response of a natural

gas pipeline linking two structures within a pressurising station.”In

Proc., 17th World Conf. on Earthquake Engineering. Hoboken, NJ:

Wiley. http://www.exchange-risk.eu/uploads/4/3/9/4/439426/pipe_hybrid

_testing_-_conference_paper_finalized1.pdf.

Psyrras, N., O. Kwon, S. Gerasimidis, and A. Sextos. 2019. “Can a buried

gas pipeline experience local buckling during earthquake ground shak-

ing?”Soil Dyn. Earthquake Eng. 116 (Jan): 511–529. https://doi.org/10

.1016/j.soildyn.2018.10.027.

Psyrras, N., A. Sextos, A. Crewe, M. Dietz, and G. Mylonakis. 2020.

“Physical modeling of the seismic response of gas pipelines in laterally

inhomogeneous soil.”J. Geotech. Geoenviron. Eng. 146 (5): 04020031.

https://doi.org/10.1061/(ASCE)GT.1943-5606.0002242.

Reza, M. S., O. S. Bursi, G. Abbiati, and A. Bonelli. 2013. “Pseudo-

dynamic heterogeneous testing with dynamic substructuring of a piping

system under earthquake loading.”In Vol. 8 of Proc., ASME 2013

Pressure Vessels and Piping Conf. New York: ASME.

Ricker, N. 1943. “Further developments in the wavelet theory of seismo-

gram structure.”Bull. Seismol. Soc. Am. 33 (3): 197–228.

Sakai, M., S. Matsuura, and F. Inada. 2013. “Hybrid seismic response

evaluation method for wall thinning piping system.”In Vol. 1 of Proc.,

21st Int. Conf. on Nuclear Engineering. New York: ASME.

Sextos, A., et al. 2018. “Exchange-risk: Experimental & computational

hybrid assessment of natural gas pipelines exposed to seismic risk.”

In Proc., 16th European Conf. on Earthquake Engineering,1–12.

Istanbul, Turkey: European Association for Earthquake Engineering.

Sextos, A., G. Manolis, N. Ioannidis, and A. Athanasiou. 2017. “Seismi-

cally induced uplift effects on nuclear power plants. Part 2: Demand on

internal equipment.”Nucl. Eng. Des. 318 (Jul): 288–296. https://doi.org

/10.1016/j.nucengdes.2016.12.036.

Shalaby, M. A., and M. Y. A. Younan. 1998. “Limit loads for pipe elbows

with internal pressure under in-plane closing bending moments.”

J. Pressure Vessel Technol. 120 (1): 35–42. https://doi.org/10.1115/1

.2841882.

Taghavi, S., and E. Miranda. 2008. “Effect of interaction between primary

and secondary systems on floor response spectra.”In Proc., 14th World

Conference on Earthquake Engineering. Tokyo: International Associ-

ation for Earthquake Engineering.

Varelis, G. E., S. A. Karamanos, and A. M. Gresnigt. 2012. “Pipe elbows

under strong cyclic loading.”In Vol. 8 of Proc., ASME 2012 Pressure

Vessels and Piping Conf.,45–54. New York: ASME.

Varelis, G. E., P. Pappa, and S. A. Karamanos. 2011. “Finite element analy-

sis of industrial steel elbows under strong cyclic loading.”In Vol. 8

of Proc., ASME 2011 Pressure Vessels and Piping Conf., 109–119.

New York: ASME.

Vathi, M., S. A. Karamanos, I. A. Kapogiannis, and K. V. Spiliopoulos.

2017. “Performance criteria for liquid storage tanks and piping systems

subjected to seismic loading.”J. Pressure Vessel Technol. 139 (5):

1–12. https://doi.org/10.1115/1.4036916.

Vazouras, P., S. A. Karamanos, and P. Dakoulas. 2010. “Finite element

analysis of buried steel pipelines under strike-slip fault displacements.”

Soil Dyn. Earthquake Eng. 30 (11): 1361–1376. https://doi.org/10.1016

/j.soildyn.2010.06.011.

Watakabe, T., I. Nakamura, A. Otani, M. Morishita, T. Shibutani, and M.

Shiratori. 2017. “Seismic qualification of piping systems by detailed

inelastic response analysis: Part 4—Second round benchmark analyses

with stainless steel piping component test.”In Vol. 8 of Proc., ASME

2017 Pressure Vessels and Piping Conf. New York: ASME.

Wenzel, M., F. Basone, and O. S. Bursi. 2018. “Novel metamaterial-

based foundation concept applied to a coupled tank-pipeline system.”

In Vol. 5 of Proc., ASME 2018 Pressure Vessels and Piping Conf.,

1–12. New York: ASME.

Yoshizaki, K., M. Hamada, and T. O’Rourke. 2000. “Large deformation

behavior of low-angle pipeline elbows subjected to in-plane bending.”

In Vol. 1508 of Proc., 12th World Conf. on Earthquake Engineering.

Tokyo: Association for Earthquake Engineering.

Zhan, H., and O.-S. Kwon. 2015. “Actuator controller interface program

for pseudo- dynamic hybrid simulation.”In Proc., 2015 World

Congress on Advances in Structural Engineering and Mechanics,1–13.

Daejeon, Korea: International Association of Structural Engineering

and Mechanics.

J. Pipeline Syst. Eng. Pract., 2021, 12(2): 04020073

Hazard, exposure, fragility, and damage state homogenization of a virtual oil refinery testbed for seismic risk assessment

Article

Sep 2024
EARTHQ SPECTRA

A virtual mid-size oil refinery, located in a high-seismicity region of Greece, is offered as a testbed for developing and testing system-level assessment methods due to direct impact from seismic shaking and without considering geohazards, such as liquefaction and surface faulting. Its characterization is offered in a dedicated repository (https://doi.org/10.5281/zenodo.11419659) and it comprises (a) a comprehensive probabilistic treatment of seismic hazard tied to an open-source seismological model; (b) a hazard-consistent set of ground motion records; (c) a full geolocated exposure model with all pertinent critical assets, namely tanks, pressure vessels, process towers, chimneys, equipment-supporting buildings, and a flare; (d) the corresponding record-wise asset demands and summarized fragilities derived via nonlinear dynamic analyses on reduced-order numerical models. Background information is provided on all refinery assets to delineate their role in the refining process. Furthermore, an explicit homogenization of the damage states is proposed, translating them from the asset level to the refinery system level considering the importance of each asset on the overall operational and structural integrity of the refinery. The results can form the basis of any follow-up study that seeks to characterize the effects of cascading failures (fires, explosions), mitigation measures, seismic sequences, and operational constraints on the functionality, risk, and resilience of refining facilities.

Network Interface Program for Actuator Controllers for Pseudo-Dynamic Hybrid Simulation

Article

Apr 2023

Experimental Hybrid Simulation of Severe Aftershocks Chains on Buildings Equipped with Curved Surface Slider Devices

Article

Full-text available

Aug 2022

In this research work the outcomes of a hybrid experimental campaign are analyzed, in order to evaluate the influence of aftershock events on the frictional response of sliding-based isolation devices for buildings. To achieve this, a hybrid testing framework was accordingly defined, by considering a numerical substructure, in terms of a simplified analytical model of a case study structure, and a physical substructure, as a full-scale Curved Surface Slider device, tested within the Bearing Tester System of the EUCENTRE Foundation Laboratory in Pavia (Italy). The tested isolator was equipped with a special sliding material, made up of a Poly-Tetra-Fluoro-Ethylene-based compound (PTFE), filled with carbon fibers and with a solid lubrication. The hybrid tests were performed, in terms of earthquake simulations, and the response of the base-isolated structural system was computed, by applying single-events, rather than aftershock chains. Results lead to a better understanding of the behavior of sliding-based seismic isolation systems, characterized by medium-to-high tribological properties, in terms of peak and residual displacements for both the single-event and the mean responses. Specifically, this work provides hybrid experimental evidence of the influence of an initial displacement offset on the overall behavior of the considered structural system.

Seismic fragility assessment of building‑type structures in oil refineries

Article

Full-text available

Aug 2022
B EARTHQ ENG

A seismic fragility assessment methodology is presented for equipment-supporting reinforced concrete and steel buildings that are typically encountered in oil refineries. Using a suite of hazard-consistent ground motions and reduced-order models, incremental dynamic analysis is performed to obtain the seismic demand of the structural systems examined. Appropriate drift- and floor acceleration-sensitive failure modes are considered to define the limit state capacities of the supporting structure and the nested non-structural process equipment. Special care is exercised on the demand and capacity representation of structural and non-structural components, offering a transparent roadmap for undertaking analytical fragility assessment for equipment-supporting buildings typical to an oil refinery. The findings and the proposed methodology can be exploited by designers and facility managers for mitigating the risk of failure prior to the occurrence of an earthquake event, for designing the pertinent structures and their non-structural components by means of a risk-aware performance-based methodology, or as feed data in early warning systems.

Seismic fragility assessment of high-rise stacks in oil refineries

Article

Full-text available

Jul 2022
B EARTHQ ENG

The seismic fragility is assessed for typical high-rise stacks encountered in oil refineries, namely process towers, chimneys, and flares. Models of varying complexity were developed for the structures of interest, attempting to balance computational complexity and accuracy regarding the structural dynamic and strength properties. The models were utilized along with a set of hazard-consistent ground motions for evaluating the seismic demands through incremental dynamic analysis. Demand/capacity-related uncertainties were explicitly accounted for in the proposed framework. Damage states were defined for each of the examined structure considering characteristic serviceability and ultimate limit states. Τhe proposed resource-efficient roadmap for the analytical seismic fragility assessment of typical high-rise stacks, as well as the findings of the presented research work are available to be exploited in seismic risk assessment studies of oil refineries.

Network Interface Program for Actuator Controllers for Pseudo-Dynamic Hybrid Simulation

Article

Full-text available

Apr 2023
J EARTHQ ENG

A network interface program for controllers (NICON) for pseudo-dynamic hybrid simulations is presented. The program is implemented as an independent program residing between the simulation model and an actuator controller with various functions necessary for hybrid simulations, such as communication, coordinate transformation, error compensation. These functions are generalized to allow for a generic configuration of actuator setup theoretically with any number of control points and control degrees of freedom. Application examples of NICON to hybrid simulations of civil infrastructures subjected to earthquakes or fire are presented.

Physical Modeling of the Seismic Response of Gas Pipelines in Laterally Inhomogeneous Soil

Article

Full-text available

May 2020

This paper reports on results from a series of 1-g, reduced-scale shake table tests of a 216-m-long portion of an onshore steel gas transmission pipeline embedded in horizontally layered soil. A set of first-order dynamic similitude laws was employed to scale system parameters appropriately. Two sands of different mean grain diameter and bulk density were used to assemble a compound symmetrical test soil consisting of three uniform blocks in a dense-loose-dense configuration. The sand-pipe interface friction coefficients were measured as 0.23 and 0.27. Modulated harmonics and recorded ground motions were applied as table excitation. To monitor the detailed longitudinal strain profiles in the model pipe, bare Fiber Bragg Grating (FBG) cables were deployed. In most cases, the pipe response was predominantly axial while bending became significant at stronger excitations. Strain distributions displayed clear peaks at or near the block interfaces, in accord with numerical predictions, with magnitudes increasing at resonant frequencies and with excitation level. By extension to full scale, peak axial strain amounted to 10−3, a demand half the yield strain, but not negligible given the low in situ soil stiffness contrast and soil-pipe friction.

NEW DYNAMIC DECOUPLING CRITERIA FOR SECONDARY SYSTEMS

Conference Paper

Full-text available

Jun 2018

This article aims to propose new criteria for dynamic decoupling between primary and secondary systems, supplementing existing criteria in the French nuclear industry. The criteria developed are based on frequency and response decoupling of both systems and can be directly used by engineers to complement existing criteria. They provide information on the types of interactions between the systems (inertial or kinematic interactions) and on the types of modeling to be adopted (lumped masses or complete model).

EXCHANGE-RISK: EXPERIMENTAL & COMPUTATIONAL HYBRID ASSESSMENT OF NATURAL GAS PIPELINES EXPOSED TO SEISMIC RISK

Conference Paper

Full-text available

Jun 2018

Historically, a number of major catastrophic earthquakes have resulted in a real loss of life and property worldwide. During the last decades, in particular, the overall exposure to seismic risk has increased not only due to the higher population density but also due to the more challenging construction methods and the multilayered connection between the various urban socioeconomic activities. Within this complex built environment, the energy transportation systems, despite their structural simplicity, are seemingly one of the weakest links in terms of seismic safety. Currently, energy production, preservation and safe transportation is one of the top priorities. This demands the need to eliminate the probability of occurrence of a potential seismically induced failure (i.e., related to explosion, fire, leakage etc) that would not only have devastating environmental impact in the affected areas, but could also cause operation disruptions with equally significant socioeconomic consequences throughout Europe. EXCHANGE-Risk is an Intersectoral/International, Research and Innovation transfer scheme between academia and the industry in Europe and North America focusing on mitigating Seismic Risk of buried steel pipeline Networks subjected to ground-imposed permanent and co-seismic deformations. This paper discusses the challenges addressed by EXCHANGE-Risk and the recent advancements made in this topic. It also paves the road for further discussion on the methodologies, experiments and tools that need to be developed to mitigate seismic risk of natural gas pipelines at a European level and beyond.

Example Manual for the University of Toronto Simulation Framework. An Open-source Framework for Integrated Multi-Platform Simulations for Structural Resilience (Second Edition)

Technical Report

Full-text available

Jul 2017

The performance assessment of civil infrastructure, such as buildings, bridges, subway tunnels, power plants, etc., under extreme loading conditions still represents a formidable challenge for engineers. With current advancements in modeling techniques and computing power, increasingly complex and realistic models of structures are being developed and refined, primarily in a single modeling package. However, the scientific and engineering community has not yet achieved complete models that can capture the entire response of complex structural systems, in order to fully assess their performance under extreme loading conditions. Multiple challenges still remain as most complex systems incorporate very different components, each requiring a level of specialized modeling sophistication or even in some cases complete physical testing in order to capture the behaviour of the integrated system under multiple hazards such as earthquakes, tsunamis, tornadoes, blasts, fire, floods and so on. The complexities associated with the performance assessment of structural systems under extreme loads along with specific applicability of most structural analysis programs to specific material/structural components, has raised an increasing interest in multi-platform and experimental hybrid simulation methods, among the earthquake engineering research community. In this document the procedure for conducting multi-platform hybrid simulations, using a variety of structural analysis programs, and some of the recent developments at the University of Toronto, is presented through various examples.

Mechanical Response of an Industrial Piping System Under Strong Cyclic Loading

Conference Paper

Jul 2021

The paper presents a combined experimental and numerical investigation of cyclic loading response of an internally pressurized steel piping system. The piping system comprises three elbows and is subjected to quasi-static end-displacement excitation. Global deformation and local strain measurements are obtained, indicating significant strain ratcheting at the critical locations of the elbows. The piping system failed under low-cycle fatigue undergoing through-thickness cracking at the flank of the most strained elbow. Post-fatigue metallographic examination of the elbows indicated that fatigue cracking initiates from the inner surface of the pipe elbow. In all elbows, several micro-cracks develop along the inner surface of elbow flanks, whereas the outer surface remained practically intact before through-thickness cracking. Finite element simulations, with a properly calibrated cyclic-plasticity model calibrated properly in terms of small-scale material tests, provide very good predictions in terms of local strain evolution at critical locations.

Can a buried gas pipeline experience local buckling during earthquake ground shaking?

Article

Nov 2018
SOIL DYN EARTHQ ENG

The damage potential of spatially variable seismic ground motion on buried pipelines has long been confirmed by field evidence, but it is still debatable whether transient seismic loads can be truly detrimental to the pipeline integrity. In the absence of systematic scrutiny of the effects of local site conditions on the seismic behaviour of such structures, this study presents a staged approach to numerically investigate the elastic-plastic buckling response of buried steel natural gas pipelines subject to transient differential ground motions arising from strong lateral site inhomogeneities. The first stage involves the study of 2D linear viscoelastic and equivalent-linear site response for the case of two sites and the resulting seismic demand in terms of longitudinal strains for input motions of various intensities and frequency content. The influence of key problem parameters is examined, and the most unfavourable relative ground deformation cases are identified. In the second stage of analysis, the critical in-plane ground displacement field is imposed monotonically on a near-field trench-like 3D continuum soil model encasing a long cylindrical shell model of the pipeline. Next, the performance of the buried pipeline is assessed under axial compression. The impedance contrast between the laterally inhomogeneous soil profiles is shown to govern the amplitude of induced elastic strains, which are maximized for low-frequency excitations. It is also demonstrated that peak axial strains along the pipeline considering equivalent-linear soil behaviour under strong earthquake motion can be as much as two orders of magnitude larger than their linear counterparts, as a result of the severe, spatially variable moduli degradation. It is finally shown that the seismic vibrations of certain inhomogeneous sites can produce appreciable axial stress concentration in the critically affected pipeline segment near the material discontinuity, enough to trigger coupled buckling modes in the plastic range. This behaviour is found to be controlled by pronounced axial force-bending moment interaction and is not accounted for in code-prescribed limit states.

Novel Metamaterial-Based Foundation Concept Applied to a Coupled Tank-Pipeline System

Conference Paper

Jul 2018

The recent advance of seismic metamaterials has led to various concepts for the attenuation of seismic waves. A particularly promising type of material is the locally resonant metamaterial, which is capable of attenuating seismic waves at wave lengths much greater than the dimensions of its unit cells. Based on this concept, the so called Metafoundation has been design. It is able to protect a fuel storage tank from ground motions at various fluid levels. In order to show the effectiveness of the proposed design, the response of the Metafoundation is compared to the response of a tank on a traditional concrete foundation. The design process of conceiving the Metafoundation, optimizing it for a specific tank, and its seismic response are described herein. Furthermore, the response of a tank during a seismic event can cause severe damages to pipelines connected to the tank. This phenomenon can be of critical importance for the design of a seismic tank protection and must be treated with care. Since the coupled structure (tank+foundation+pipeline) exerts highly non-linear behavior, due to the complexity of the piping system, a laboratory experiment has been conducted. More precisely, a hybrid simulation that uses the Metafoundation and a tank as a numerical substructure (NS) and a piping system as a physical substructure (PS) was employed. In order to make the results relatable to the current state of the art, additional experiments were performed with concave spherical bearing devices (CSBs) as an isolation system in the NS. While the Metafoundation performed better than the CSB isolated structure for some records, a clear advantage could not be concluded. However, the Metafoundation offers a clear attenuation of tank stresses and may potentially mitigate also piping stresses when tuned to the right frequency in the future.

A Generalized Numerical/Experimental Distributed Simulation Framework

Article

Jan 2018
J EARTHQ ENG

This paper describes a generalized distributed simulation framework that has been developed at the University of Toronto. The proposed framework is characterized by the ability to integrate diverse numerical integration and substructure modules and experimental specimens. This feature is enabled by a standardized data-exchange format and a communication protocol through which any potential integration or substructure modules can communicate with each other. The data-exchange format is designed with the information required for various simulation purposes such as real-time simulation, hybrid simulation with thermal load, etc. Current development of the simulation framework is presented through several application examples, which are rarely seen from previous simulations.

Seismic Qualification of Piping Systems by Detailed Inelastic Response Analysis: Part 4 — Second Round Benchmark Analyses With Stainless Steel Piping Component Test

Conference Paper

Jul 2017

Some studies concerning ultimate strength of piping under seismic loads concluded that there is a large design margin until failure, even if the stress calculated based on the current design method does not satisfy design criteria. To provide a more rational seismic design, a new Code Case for seismic design of piping is now under development in the framework of JSME Nuclear Codes and Standards. The Code Case incorporates a dynamic elastic-plastic analysis procedure by employing finite element analysis as an alternative to the current design analysis method of elastic assumption. To confirm the applicability of inelastic response analysis, benchmark analyses have been conducted. In the first round benchmark, a carbon steel elbow analysis was performed. In this report, a second round benchmark with a stainless steel elbow and tee is introduced. The second benchmark aims to establish an analysis procedure for stainless steel piping and tee piping of complicated shapes. The second benchmark results provided a practical analysis method for stainless steel piping, and the Code Case was expanded so that it could be applied not only to carbon steel piping but also to stainless steel piping. The second benchmark also challenged analyses of a tee having complicated geometry. These results provide some important knowledge, and they will be included in the Code Case.

Probabilistic seismic analysis of an LNG subplant

Article

Oct 2017
J LOSS PREVENT PROC

Refrigerated liquefied gas (RLG) terminals that are part of lifeline facilities must be able to withstand extreme earthquakes. A liquefied natural gas (LNG, ethylene) terminal consists of a series of process facilities connected by pipelines of various sizes. Although tanks, pipes, elbows and bolted flanges have been a major concern in terms of seismic design, generally, they have not been analysed with modern performance-based procedures. In this study, the seismic performance of pipes, elbows and bolted flanges is analysed and seismic fragility functions are presented within the performance-based earthquake engineering framework.Particular attention was paid to component resistance to leakage and loss of containment even though several different limit states were investigated. The LNG tank, support structures and pipework, including elbows and flanges, were analysed with a detailed 3D finite element model. For this purpose, we developed a mechanical model of bolted flange joints, able to predict the leakage limit state, based on experimental data. A significant effort was also devoted to identification of a leakage limit state for piping elbows, and we found the level of hoop plastic strain to be an indicator. Given the complexity of the FE model of the LNG plant, we selected the Cloud method for probabilistic seismic demand analysis, due to its advantages in terms of consistency in the seismic input and of computational savings. Then, using a series of nonlinear time history analyses, we studied the behaviour of critical components such as elbows and bolted flange joints. In order to develop fragility curves, we selected a set of 36 ground motions from a database of historic earthquake accelerations. The results of seismic analysis show that bolted flange joints remain significantly below their leakage threshold whilst elbows at the top of the LNG tank are likely to show leakage. Moreover, fragility functions were computed, based on a linear regression approach, and we deduce that elbows located on the tank platform are relatively unsafe against earthquakes. Finally, the estimated probability of loss of containment was above the probability associated with ultimate limit states involved in structural Eurocodes.

Hybrid Simulation of Structure-Pipe-Structure Interaction within a Gas Processing Plant

Abstract and Figures

Recommended publications

A Failure Estimation Method of Steel Pipe Elbows under In-plane Cyclic Loading

Effects of the Loading History on the Local Buckling Behavior and Failure Mode of Welded Beam-to-Col...

Seismic Response Case Study of Isolated Floor System Having Special Biaxial Spring Units

Seismic Simulation Test of a Fire Protection Riser Pipe with Fixed Groove Joints