Engineering Structures 177 (2018) 566–578

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Seismic behavior of recycled plastic lumber walls: An experimental and T

analytical research
⁎
Juan P. Herrera , Daniel Bedoya-Ruiz, Jorge E. Hurtado
Universidad Nacional de Colombia, Código Postal 170004, Manizales, Colombia

A R T I C LE I N FO A B S T R A C T

Keywords: Recycled Plastic Lumber (RPL) is a wood-like material made from recycled plastics that aims to diminish the
Recycled Plastic Lumber environmental pollution resulting from plastic wastes. This material is used in different kinds of nonstructural
Structural walls and structural applications. Recently, RPL has been proposed as a suitable material to develop structural walls
Cyclic loading test that comprise the seismic resistant system of housings, in order to lessen the housing deficiency. This article
Seismic behavior
presents the results drawn from an experimental campaign carried out over three full-scale RPL walls, which
Ductility
Multilinear Hysteretic Model
were tested under cycling loading conditions to determine structural parameters such as strength, hysteretic
Earthquake resistant design behavior, ductility, energy dissipation, equivalent damping and characteristic failure modes of the RPL walls,
which are necessary to design and to assess seismically the housings. Finally, a multilinear hysteretic model
capable of simulating the nonlinear dynamic behavior exhibited by RPL walls was implemented, in order to
simulate and to assess the seismic behavior of them under strong and destructive real earthquakes.

1. Introduction
The use of residues resulting from the human activity is currently a
potential and sustainable source of materials for the building sector,
particularly for the housing development. Plastic is a material com-
monly used in the human activities, that is almost impossible to break it
down by natural processes on account of its strength and durability. In
the last 65 years, 8.3 billion tonnes of virgin plastic have been produced
of which 6.3 billion tonnes of plastic wastes have been generated; Of
this value, roughly 9% is recycled, 12% is incinerated, and 79% is in a
landfill or in the natural environment [1]. It is beyond dispute that from
the civil engineering and others branches of knowledge, it is required to
implement a sustainable use of the plastic wastes in order to protect the
natural environment and to obtain useful materials. Different re-
searchers, standards societies, universities, governs and building com-
panies, have joined forces both to reduce the pollution of plastic on the
planet and to make the plastic a suitable material for the building
sector.
Recycled Plastic Lumber (RPL) is a wood-like material made from
recycled plastic, used as a substitute for the raw materials in structures
made from concrete, metals, and wood. RPL needs a few chemical and
industrial processes, and the current technology allows the correct se-
paration and cleaning of its raw material. The manufacture of RPL al-
lows making the most of large quantities of plastic wastes and
converting them into useful and durable products [2]. This material has
shown be rot resistant and not be susceptible to the corrosion or insect
attacks, assuring the durability [3]. Structural elements made of RPL
have a nonhomogeneous cross-section due to the cooling process during
extrusion; this feature together with the nonlinear nature of the mate-
rial makes different its tension and compression behavior, and some
mechanical properties are difficult to determine [4]. As a consequence,
in order to characterize this material and to enable the market accep-
tance of RPL in structural and building applications, ASTM Interna-
tional has developed specifications and test methods standards [5–9]. In
addition, from materials engineering, RPL has been studied to know its
mechanical properties, such as density, elasticity modulus, compres-
sive, flexural, shear and tensile strength [3,10,4,11], creep behavior
[4,12], among others, that are important for the structural design.
The first applications of RPL were in urban furniture and in low-
stress or non-critical load outdoor structures, such as picnic tables, park
benches, trash receptacle covers, among others. With the satisfactory
performance of these applications, RPL gained popularity and it was
seen as a novel material for others applications, such as docks, board-
walks, and decks [2]. However, some of the first RPL decking structures
sagged with time due to their low stiffness and their tendency to creep
under its own weight [2]. As a consequence, manufacturers and
structural designers learned to design these structures with lower stress
on the elements, for example, decreasing the joist spacings and/or using

⁎
Corresponding author.
E-mail addresses: jpherrerac@unal.edu.co (J.P. Herrera), dabedoyar@unal.edu.co (D. Bedoya-Ruiz), jehurtadog@unal.edu.co (J.E. Hurtado).

https://doi.org/10.1016/j.engstruct.2018.10.006
Received 15 June 2018; Received in revised form 28 August 2018; Accepted 2 October 2018
0141-0296/ © 2018 Elsevier Ltd. All rights reserved.
J.P. Herrera et al.
Fig. 1. Test setup and geometry of the RPL walls specimens (mm).
thicker deck boards, in order to reduce the time-dependent properties
that change the overall shape of the elements [13]. Gradually, RPL
started to be used in several structural applications where concrete,
steel or wood, were the traditional material, such as joists, marine
pilings, bridges (for military tanks, railroads and vehicular traffic), one
and two-story housings, among others. In 2000, Mclaren et al. [14]
described and implemented laboratory tests to the connections and to
the structural elements of the first recycled plastic bridge in the world
made from a glass fiber reinforced recycled plastic lumber composite
material, in order to evaluate and to verify the mechanical properties of
RPL and the structural behavior of the bridge. In 2009, Chandra and
Kim [15] described the first bridges of the world made from RPL; they
showed that the use of the RPL in bridges had increased.
Recently, RPL has been investigated and implemented for the con-
struction of housings of one and two-story [11,16], looking to diminish
the housing deficiency and the accumulations of plastic wastes in the
natural environment; in fact, some building companies and researchers
have developed housing systems based RPL structural walls that define
the structural system of seismic resistance [16,17]. Gulhane and Gul-
hane [11] developed analytical studies of housings built with precast
RPL elements that allowed identifying some structural features, such as
large horizontal displacements in the walls that formed the housings
when they were subjected to seismic or wind loads. These displace-
ments suggested doubling the thickness of the walls and using a hollow
section in order to improve the structural behavior. Compression and
flexural strength were not exceeded by the stress induced by seismic,
wind or typical loads of use or occupancy of the housings, but care must
be taken with the excessive lateral displacements respect to the height
of the building. Moreover, they found that the RPL elements had a
specific density of 2.2–2.8 times less than the masonry walls or concrete
567
Engineering Structures 177 (2018) 566–578
elements, and a compressive strength of 25–26 MPa, which is more than
the compressive strength of the masonry wall (1.8–2.9 MPa) and similar
to concrete compressive strength (17–28 MPa). From the seismic-re-
sistant engineering, the use of light and resistant materials constitutes a
natural seismic-resistant system for any kind of building [18]; housings
built of RPL could have a better seismic performance given its low
density and high strength than those built of usual materials, such as
those present in the earthquakes of Haiti 2010, Chile 2010, Nepal 2015,
Ecuador 2016 and Mexico 2017. Hence, it is necessary to assess the
behavior of the structural systems based RPL walls under cyclic loads
that simulate the strong phase of an earthquake, in order to determine
structural parameters, such as stiffness, lateral resistance, ductility,
energy dissipation capacity and damping, which allow a reliable design
of the housings [19].
The plan of this paper is as follows: first, laboratory tests on three
full-scale walls were carried out in order to evaluate the hysteretic
behavior, the strength, the energy dissipation capacity, damping, and
failure modes of the RPL walls when subjected to cyclic loading con-
ditions. Then, it is proposed a nonlinear dynamical model that is able to
simulate the behavior exhibited by RPL walls under cyclic loading
conditions. A multilinear hysteretic model [20] is used here. Later, an
assessment of the seismic performance of the RPL walls when subjected
to recent earthquakes using the proposed nonlinear dynamical model is
performed. The paper ends with some conclusions.
2. Test specimens
2.1. Description of constituent elements
To manufacture RPL, the raw materials are sorted, cleaned and
J.P. Herrera et al.
ground into small flakes. Then, by an extrusion process, the flakes are
homogenized and rapidly melt; finally, the molten mixture is dis-
charged into a mold and cooled, shaping the finished product. The type
of RPL manufactured for the walls is a commingled RPL. The density of
the RPL for the walls was 821 kg/m3 [7], its compressive strength was
18.06 MPa [5], its flexural strength was 19.32 MPa [6] and its Young
modulus was 1102.55 MPa [6].
The walls are formed of: (1) RPL battens that have a rectangular
cross-section 50 mm wide and 90 mm deep, and they are vertical,
horizontally and diagonally arranged; and (2) rectangular RPL boards
with 1100 mm length, 100 mm wide and 20 mm of thickness, hor-
izontally arranged. These structural elements are assembled by drywall
screws #8 × 2 in., located at the places where both structural elements
converge to guarantee the structural integrity of the system. This type
of screw has shown to be suitable for the thickness of the structural
elements that form the RPL walls, and besides, it is typically used in the
assembly of RPL elements that form others structural and nonstructural
applications. Nevertheless, it is worth to emphasize that the type of
connection, the type, the localization, and the number of screws change
the structural response; therefore, these features and the configuration
of the connections could be modified according to a design or capacity
requirements in particular.
In this research, it was chosen the structural configuration and the
connection type previously presented, which are typical in the building
of structural systems based RPL walls for one and two-story housings.
The geometry of the RPL walls tested is sketched in Fig. 1.
2.2. Test setup and load history
In order to provide lateral support, the RPL walls were anchored to
rigid concrete foundation beam by longitudinal and transverse steel
bars No.4; these bars went through battens and the boards as can see in
Fig. 1. Then, the “wall-foundation beam” system was anchored to a
strong reaction floor with steel stud bolts as illustrated in Fig. 1. A
lateral bracing was employed to constrain the RPL walls to in-plane-
displacements. The lateral displacement induced by the hydraulic ac-
tuator was measured using a LVDT placed at the top of the RPL walls.
Also, the RPL walls were instrumented with LVDTs vertical and diag-
onally arranged. Each RPL wall was subjected to the cyclic loading
pattern shown in Fig. 2, according to [21], where Δm is the ultimate
displacement corresponding to the failure limite state.
3. Test results
3.1. Hysteretic behaviour and strength
Hysteresis response curves were drawn from the cyclic load tests
carried out on the three RPL walls, in order to assess the performance,
strength, and ductility of the system under seismic actions. The
Fig. 2. Cyclic loading pattern.
568
Engineering Structures 177 (2018) 566–578
hysteretic behavior of each one of the three RPL walls can be observed
in Figs. 3–5. In general, the RPL walls exhibited: stiffness degradation,
scant strength deterioration in the last cyclic loading, and pinching
effect due to the type of connections and to the relative displacement
between the boards. In addition, Figs. 3–5 show that the maximum top
displacement reached in the RPL walls was in the order of 160 mm (7%
of drift ratio level).
The ductility of the system is computed for each peak values from
hysteresis loops of each RPL wall, as shown Figs. 3–5. Every ductility
level allows describing the behavior of the system and the ability of the
system to maintain high levels of deformation without damage. Ac-
cording to the above, the RPL walls experimented: a completely elastic
behavior ( μ = 1) until a drift ratio level ⩽ 2%; scant ductility
(1 ⩽ μ ⩽ 3) between 2.5% and 6% of drift ratio level; and a completely
ductile behavior ( μ ⩾ 3) to a drift ratio level ⩾ 6%.
The positive and negative envelope curves corresponding to the
peak values from hysteresis loops of each RPL wall were computed and
are depicted in Figs. 6–8. It can be seen that both positive envelopes and
negative envelopes of each RPL wall show a similar behavior; therefore,
the average of these envelopes is used to estimate the structural para-
meters of the system.
Table 1 shows the experimental values of the structural parameters
of each RPL wall computed according to the ASTM E2126-11 [21] and
using the average envelope. As can be seen, the Elastic shear strength
(νpeak ) and Maximum load (Ppeak ) of the RPL walls are within 10% of
each other, which means that both the structural response and struc-
tural parameters of the RPL walls are consistent [21].
3.2. Energy dissipation and damping
Structural designs based on fragility analysis aim at increasing the
safety of structural elements and avoiding the sudden failure of the
material, controlling of damage of the structure when it is subjected to
severe loads due to earthquakes. The damage is related to the ability of
the structural system to dissipate the energy transmitted by external
cyclic loads by means of the hysteretic behavior; this dissipated energy
can be computed by the area enclosed by the hysteresis loops (see
Figs. 3–5).
The ability of the structural system in terms of energy dissipation
can be quantified by means so-called equivalent viscous damping ratio
(ξeq ) [22,23] which is defined as:
Ei
ξeq = ·100%,
4π (Ee )i (1)
where Ei is the energy dissipated by the structural element in the i-th
loading cycle computed from the measured hysteresis loops and (Ee )i is
the elastic strain energy stored by an equivalent linear elastic system
when the maximum displacement in the i-th cycle is reached under
static conditions; this energy is computed by the area under the load vs.
displacement curve obtained in the cyclic tests, which is a right triangle
whose base equals the maximum positive displacement reached in the
model (Δmax )i and whose height (Pm )i is the average peak load of the
analyzed cycle, i.e.:
|(Pmax )i| + |(Pmin )i |
(Pm )i = ;
2 (2)
where (Pmax )i and (Pmin )i represent the maximum and minimum values
that the load reaches in the i-th cycle. With these data, the energy (Ee )i
can be computed as:
(Pm )i (Δmax )i
(Ee )i = .
2 (3)
The accumulated dissipated energy and the equivalent viscous
damping of the RPL walls as a function of the drift ratio are depicted in
Figs. 9 and 10 respectively. As can be seen, the structural system de-
velops higher energy dissipation with an increase in the drift ratio level.
J.P. Herrera et al.
Fig. 3. Hysteresis cycles Wall 1.
All walls exhibited loss of energy in the last cycle. However, the
rounded shape of the hysteresis curves and the area enclosed by those
loops suggest that the system has good energy dissipation capacity, as
this increases with the drift. The shape of the hysteresis loops effec-
tively determines the ability of these walls to dissipate energy when
they are subjected to loads beyond its linear range. On the other hand,
it can be observed a degree of pinching in the center of the hysteresis
loops due to the type of connections and to the relative displacement
between the boards, which affects, to some extent, the ability of the
system to dissipate energy. These effects will be considered in the
nonlinear dynamical model that will be proposed below.
3.3. Coefficient of energy dissipation capacity
The coefficient of energy dissipation capacity (Rμ ) is a designed-
oriented concept which aims to approximate the nonlinear behaviour of
structural systems when subjected to seismic loads. It is used in most of
the earthquake resistant design codes, in order to reduce the magnitude
of the loads assessed from an elastic viewpoint.
We used the model proposed by Takada et al. [24] to determine Rμ ,
which is independent of the period of the structure:
Rμ = ε 2μ−1 , (4)
where ε is an adjustment factor which represents the degree of devia-
tion of the Rμ vs. μ relation from the equal energy expression 2μ−1 .
Results drawn form Monte-Carlo simulations indicate that the average
value of ε is between 1.05 and 1.34 [24,25]. Using an average of the
ductilities shown in Table 1 and taking ε = 1.05 and ε = 1.34 , we have
569
Engineering Structures 177 (2018) 566–578
that Rμ = 2.543 and Rμ = 3.246 , respectively.
3.4. Observed failure modes of the RPL walls subjected to cyclic loading
Fig. 11 shows the characteristic failure modes of the RPL walls after
the test cyclic loading. For drift ratio levels between 0% and 3%, the
RPL walls did not present any kind of failure; for a drift ratio level more
than 3%, the RPL walls only experimented minor relative displacements
of the boards; the maximum drift ratio level reached by the RPL walls
was 7%, that is, 160 mm of lateral top displacement. At this drift ratio
level, the structural elements of the walls did not present any pattern of
formation of cracks; nevertheless, the boards of the RPL walls experi-
mented excessive relative displacement each other, in the order of
15 mm (see Fig. 11a). Due to that and to the fatigue produced by
loading cyclic conditions, some boards-battens connections experi-
mented bearing failure, tear-out failure, and single shear failure in the
drywall screws (see Fig. 11b).
4. Nonlinear dynamical model
The Multilinear Hysteretic Model proposed by Mostaghel [20] is a
physically-based analytical model capable of describing the behavior of
general hysteretic systems; this model includes effects of pinching,
stiffness degradation, and load deterioration. One of the principal ad-
vantages of this formulation is that each one of the system parameters
represents a definite physical quantity that can be measured through
experiments.
A single-degree-of-freedom (SDOF) Multilinear Hysteretic System
J.P. Herrera et al. Engineering Structures 177 (2018) 566–578

Fig. 4. Hysteresis cycles Wall 2.

with n linear segments would need a total of (2n−1) equations to cor- deformation of the ith spring of this set are αi k and ui , respectively. The
rectly model the system. The necessity for the additional equations parameter αi is the stiffness ratio, defined as:
arises from the fact that the stiffness degradation and load deterioration
are caused by the deterioration of the properties of the system, which ki
αi =
starts at the inception of plastic deformation in the system. The effects kinit (9)
of stiffness degradation and load deterioration are defined in terms of
the total absorbed hysteretic energy. where ki is the stiffness for the i-th spring and kinit is the elastic initial
To develop the equations that govern the behavior of a Multilinear stiffness. It is important to emphasize that k is the sum of the stiffnesses
Hysteretic System, it is necessary to consider the SDOF shown in Fig. 12 αi k of all springs that are directly connected to the mass; therefore, the
and the following complementary functions: sum of all αi is equal to unity.
In addition to the above, the system includes: a spring of stiffness
N (w ) = 0.5 ⎡1 + sgn(w ) ⎤ ⎧1 + ⎡1−sgn(w ) ⎤ ⎫ =
⎢
⎣
⎥⎨
⎦⎩
⎢
⎣
⎥⎬
⎦⎭
{
1, w ⩾ 0
0, w < 0 (5)
αn k which is directly connected to the mass and its deformation is re-
presented by x; and m other springs located on the right side of the
mass, each one with a stiffness ksi and an initial gap δsi . The equilibrium
equation for the system is defined as [20]:
M (w ) = 0.5 ⎡1−sgn(w ) ⎤ ⎧1−⎡1 + sgn(w ) ⎤ ⎫ =
⎢
⎣
⎥⎨ ⎢
⎦⎩ ⎣
⎥⎬
⎦⎭
{
0, w ⩾ 0
1, w < 0 (6) n−1
mx¨ (t ) + cx ̇ (t ) + αn kx (t ) + k ∑ αi·ui (t )+
N (w ) = 0.5 ⎡1 + sgn(w ) ⎤ ⎧1−⎡1−sgn(w ) ⎤ ⎫ =
⎢
⎣
⎥⎨ ⎢
⎦⎩ ⎣
⎥⎬
⎦⎭
{
1, w > 0
0, w ≤ 0 (7) m
i=1

∑ αsi k (|x (t )|−γsi δ ) sgn (x (t )) N (|x (t )|−γsi δ ) = Po (t )

i=1 (10)
M (w ) = 0.5 ⎡1−sgn(w ) ⎤ ⎧1 + ⎡1 + sgn(w ) ⎤ ⎫ =
⎢
⎣
⎥⎨
⎦⎩
⎢
⎣
⎥⎬
⎦⎭
{
0, w > 0
1, w ≤ 0 (8) where αsi ≔ ksi/ kinit is the stiffness-hardening ratio for the spring of
where sgn (w ) is the signum function and N (w ) is the unit step function. stiffness ksi and γsi ≔ δsi/ δ1 is the stiffness-hardening ratio of the gap for
As can be seen in Fig. 12, the SDOF Multilinear Hysteretic System is the i-th spring (δsi ) to the limiting deformation of the first spring (δ1) .
composed of a mass, (n + m) number of springs, and a viscous damping The response force of the system F (t ) is given by:
coefficient (c ) , which is excited by an external force Po (t ) . The system
has a set of (n−1) springs in parallel connection that are directly con-
nected both to the mass and to a set of sliders; the stiffness and the

570
J.P. Herrera et al. Engineering Structures 177 (2018) 566–578

Fig. 5. Hysteresis cycles Wall 3.

n−1 z i̇ (τ )= y ̇ (τ ) ϕki (τ ){N (y ̇ (τ ))[M (z i (τ )−λpi γi ϕli (τ )) M (y (τ ))+

F (t )=αn kx (t ) + k ∑ αi·ui (t )+
M (z i (τ )−γi ϕli (τ )) N (y (τ ))] + M (y ̇ (τ ))[N (z i (τ )+
i=1
m λpi γi ϕli (τ )) N (y (τ )) + N (z i (τ ) + γi ϕli (τ )) M (y (τ ))]} (15)
∑ αsi k (|x (t )|−γsi δ ) sgn (x (t )) N (|x (t )|−γsi δ )
i=1 (11) for i = 1, 2, 3, …, n−1, with the initial condition z (0) = 0 . The total
nondimensional hysteretic energy hi (τ ) absorbed by a slider attached to
In order to nondimensionalize the equations that govern the beha-
the i-th spring is governed by the following differential equation:
vior of the system, the following expressions are used:
hi̇ (τ )=ϕli (τ ) αi γi y ̇ (τ ) [N (y ̇ (τ )) N (y (τ )−γpi )+
τ = tω; ω= kinit / mass ; x (t ) = δ1 y (τ );
ui (t ) = δ1 z i (τ ); Po (t ) = kinit δ1 p (τ ) F (t ) = kinit δ1 f (τ ) (12a-f) ‾ (y ̇ (τ )) M (y (τ ) + γpi ) + λpi N
M ‾ (y ̇ (τ )) M (y (τ ))+
λpi M (y ̇ (τ )) N (y (τ ))]· 1−{N
‾ (y ̇ (τ ))[M
‾ (z (τ ) −
As a result, the nondimensionalized differential equation of the
system corresponding to Eq. (10) is given by: λpi γi ϕli (τ )) M
‾ (y (τ )) + M
‾ (z (τ )−γi ϕli (τ )) N (y (τ ))]+
M (y ̇ (τ ))[N (z (τ ) + λpi γi ϕli (τ )) N (y (τ ))+
n−1
y¨ (τ ) + 2ξy ̇ (τ ) + αn y (τ ) + ∑ αi·z i (τ )+ N (z (τ ) + γi ϕli (τ )) M (y (τ ))]} (16)
i=1
m for i = 1, 2, 3, …, n−1, where 0 ⩽ λpi ≤ 1 represents the pinching effects
∑ αsi (|y (τ )|−γsi ) sgn (y (τ )) N
‾ (|y (τ )|−γsi ) = p (τ ) in function of the strength ratio for the i-th spring; γpi ≔ (1−λpi ) is the
i=1 (13) pinched complementary strength ratio; γi ≔ δi/ δ1 is the ratio of the
limiting deformation for the i-th spring (δi ) to limiting deformation of
and the nondimensional force of the system corresponding to Eq. (11) is
the first spring (δ1) with the condition δ1 ⩽ δ2 ⩽ δ3…⩽δn − 1. The stiffness-
represented by:
degradation function ϕki (τ ) and the load-deterioration function ϕli (τ )
n−1
for the i-th spring can be represented by the following expressions,
f (τ )=αn y (τ ) + ∑ αi·z i (τ )+
respectively:
i=1
m 1
∑ αsi (|y (τ )|−γsi ) sgn (y (τ )) N (|y (τ )|−γsi ) ϕki (τ ) =
1 + λki hi (τ ) (17)
i=1 (14)
where z i (τ ) is the nondimensional hysteretic displacement of the i-th 1
ϕli (τ ) =
spring, defined by the differential equation: 1 + λli hi (τ ) (18)

571
J.P. Herrera et al. Engineering Structures 177 (2018) 566–578

Fig. 6. Positive, negative and average envelopes Wall 1.

Fig. 7. Positive, negative and average envelopes Wall 2.

572
J.P. Herrera et al. Engineering Structures 177 (2018) 566–578

Fig. 8. Positive, negative and average envelopes Wall 3.

where λki ⩾ 0 and λli ⩾ 0 are associated stiffness-degradation and load-
deterioration factors, respectively.
Eqs. (13), (15) and (16) provide the (2n−1) equations needed to
completely define the behavior of the multilinear system. The di-
mensionalized response of the system is then obtained applying the Eqs.
(12a-f). In general, a Multilinear Hysteretic Model with n lines segments
involves (7n−4) parameters. The set of all subset of parameters that
must be identified is represented by:
θ = {ξ , αs, γs, λp , α, λk , λl , δ }
4.1. Parameters identification of the multilinear hysteretic model for the
RPL walls
The identification of model parameters is a process in which the
parameters must be tuned using a real data from experimental tests so
that the response of the model matches as much as possible the real one.
The set of optimal values of the parameters is obtained by minimizing
the difference between analytical and experimental response. Several
methods have been proposed for the identification of hysteretic sys-
tems, such as Gauss-Newton [26], Least squares [27,28], extended
Kalman filters [29,30], genetic algorithm (GA) [31,32], multi-objective
optimization algorithms[33], bayesian model updating techniques
[34], Particle Swarm Optimization [35,36], among others.
Particle Swarm Optimization (PSO) is a population-based stochastic
optimization algorithm, introduced by Kennedy and Eberhart [37]. The
algorithm mimics the social behavior of swarm of animals, such as flock
of birds or fish schooling, in which an individual (particle) adapts its
behavior according to its own experience and to the behavior of its
neighbors. PSO starts with a set of random solutions, called the initial
population, where an optimum solution is reached updating genera-
tions [38]. Particles are attracted towards better positions in a design
space according to the local best-known position and the best position
(19) of the swarm, by simple expressions related to the position and velocity
of each particle. The model parameters are related to the position of
each particle.
In order to simulate the experimental behavior of the RPL walls, a
six-line hysteretic system and the experimental data of the RPL wall 2
are used. The parameters of the model are identified using a classical
PSO algorithm [38]. Table 2 shows the results of the identification
process; the parameters mass, kinit and δ1 were obtained from the ex-
perimental tests.
It can be seen from Figs. 13 and 14 that the estimated response of
the system using a six-line hysteretic model is in good agreement with
the experimental ones; the approximation with the identified para-
meters of the hysteretic model is good enough for simulation purposes.
Also, the proposed nonlinear dynamical model simulated successfully
the degradation and pinching effects experimented by the RPL walls.

Table 1
Parameters computed from the cyclic loading tests according to the ASTM E2126-11 norm.
RPL Wall Elastic stiffness Elastic shear strength Yield load Maximum load Ultimate displacement Yield displacement Ductility ( μmax ) = Δu /Δyield )
(K e ) (νpeak ) (Pyield ) (Ppeak ) (Δu ) (Δyield )
kN/mm kN/m kN kN mm mm

Wall 1 0.140 0.0070455 6.8672 7.750 157.70 48.984 3.2194

Wall 2 0.149 0.0069318 6.8835 7.625 159.45 46.201 3.4512
Wall 3 0.168 0.0072273 7.4065 7.950 159.70 43.979 3.6313

573
J.P. Herrera et al. Engineering Structures 177 (2018) 566–578

Fig. 9. Cumulative dissipated energy for the RPL walls.

5. Seismic assessment of the RPL walls
The equivalent SODF system for the RPL wall 2 was subjected to
several strong and destructive real earthquakes, such as El Centro 1940
[39], Northridge 1994 [39], Chile 2010 [40], Japan 2011 [39], Nepal
2015 [41] and Ecuador 2016 [42], in order to compute the seismic
response of the structural system when a seismic excitation is applied;
likewise, the ability of the six-line hysteretic model was tested. The
parameters of the Table 2 were used within the model for the simula-
tion. The estimated seismic response of the RPL wall was assessed in

Fig. 10. Equivalent viscous damping ratio for the RPL walls.

574
J.P. Herrera et al. Engineering Structures 177 (2018) 566–578

(a) Relative displacement

of the boards

(b) Type of failure in some boards

and battens connections
Fig. 11. Typical failure mode of the RPL walls subjected to cyclic loading conditions.

Table 2
Identified parameters of the multilinear hysteretic model for plastic lumber
walls.
Parameter Values Unit

n 6
mass 290 kg
kinit 0.2053 kN/mm
ξ 12.544 %
αs [0.04743, 0.02886, 0.05920, 0.09235, 0.04176 ]
γs [3.58690, 3.71320, 4.95360, 5.73630, 9.38250 ]
λp [0.49149, 0.42504, 0.03247, 0.52171, 0.00014, 0.19623 ]
α [0.51546, 0.15331, 0.12846, 0.10459, 0.09781, 0.00037 ]
λk [0.00141, 0.17559, 0.08801, 0.30828, 0.01603]
λl [0.00960, 0.03828, 0.09992, 0.17463, 0.14005]
δ [8.95000, 32.9880, 33.6890, 68.9910, 72.6490 ]

displacement and the damage of them under control. For example, in

Fig. 12. Representation of a SDOF Multilinear Hysteretic System [20].
terms of the displacement and it can be seen in Fig. 15.
In many countries, the seismic design codes classify the different
seismic intensity levels of the country by seismic hazard zones that
allow designing the structures depending on the requirements of the
ductility, strength, and dissipation capacity energy, in order to the
structures have a suitable seismic performance and to keep the
Colombia the seismic hazard zones comprise intensity level of PGA
from 0.05 g to 0.50 g; additionally, the maximum interstory drift ratio
must be less than or equal to 1%, and it is applicable in any seismic
hazard zone of the country. In this way, for the structural system based
RPL walls, the maximum displacement allowed by the Colombian
seismic design code is 24 mm. According to the simulation results
shown in Fig. 15, and having the failure modes observed in the ex-
perimental campaign (see Section 3.4) in mind, for earthquakes with
intensities between 0.25 g and 0.35 g, such as Nepal 2015 and El Centro
1940, the maximum displacement computed was 9 mm (0.375% of drift

575
J.P. Herrera et al. Engineering Structures 177 (2018) 566–578

Fig. 13. Experimental and estimated response of the system.

ratio level); at this level of displacement, the RPL walls do not present
any mode of failure. For earthquakes with intensities between 0.9 g and
1.4 g, such as Northridge 1944, Chile 2010 and Ecuador 2016, the
maximum displacement computed was 24 mm (1% of drift ratio level),
it means that the maximum displacement allowed by the Colombian
seismic design code is reached; nevertheless, at this level of displace-
ment, the RPL walls do not present any mode of failure. For untypical
earthquakes with an intensity in order of 2.8 g, such as Japan 2011
earthquake, the maximum displacement computed was 75 mm (3.1% of
drift ratio level); this displacement is more than three times the max-
imum allowed by the Colombian seismic design code, nevertheless at
this level of displacement, the RPL walls just present minor relative
displacements of the boards. These facts demonstrate that the RPL walls
have a satisfactory performance when are subjected to strong seismic
excitations.

Fig. 14. Experimental and estimated hysteresis cycles of the system.

576
J.P. Herrera et al.
Fig. 15. Estimated displacement response of the RPL wall when is subjected to strong real earthquakes.
6. Conclusions
The experimental campaign reported herein allowed determining
the hysteretic behavior of the RPL walls under cyclic loading condi-
tions, which is characterized by a stiffness degradation, scant strength
deterioration in the last cyclic loading, and some pinching effect. The
shape of the hysteresis loops effectively determines the ability of these
walls to dissipate energy when they are subjected to loads beyond its
linear range. Additionally, the structural parameters of the RPL walls
such as elastic stiffness, elastic shear strength, ultimate strength, duc-
tility, energy dissipation, equivalent damping, and energy dissipation
coefficient were determined from the experimental tests; these para-
meters will be necessary to design the housing systems according to the
seismic design codes.
The experimental campaign showed that for drift ratio levels until
1.5%, the RPL walls recovered its initial condition without maintaining
permanent deformations when the load was removed. Nevertheless, the
RPL walls experimented large displacements reaching a maximum drift
ratio level in the order of 7%, where the main modes of failure happen.
The nonlinear dynamical model proposed simulates successfully the
degradation and pinching effects experimented by the RPL walls under
cyclic loading conditions; the model fits well with the experimental
response and it can be used for the nonlinear seismic analysis and de-
sign of the RPL walls.
The seismic simulation included a set of strong and destructive
earthquakes that have occurred in recent years, which allowed asses-
sing the seismic behavior of the RPL walls in different seismic hazard
zones. For a range of seismic hazard levels between 0.05 g and 1 g, the
RPL walls experiment large drift ratio levels, which are less or equal
than the maximum interstory drift ratio allowed by seismic design
codes; in spite of that, the RPL walls do not present any mode of failure.
577
Engineering Structures 177 (2018) 566–578
For untypical seismic hazard levels, in the order of 2.8 g, the drift ratio
levels of the RPL walls are more than three times the maximum in-
terstory drift ratio allowed by seismic design codes; nevertheless, the
RPL walls experiment a slight mode of failure characterized by minor
relative displacements of the boards. This seismic assessment showed a
satisfactory seismic behavior the RPL walls.
The expectation is that the use of the RPL as a structural material in
the building sector increases together with the effective efforts in in-
vestigation and development to improve its structural performance.
Acknowledgments
This research has been supported by Universidad Nacional de
Colombia and the Sustainable Building Company Econciencia S.A.S.;
This support is gratefully acknowledged.
Some data used in this work were obtained by the Red Nacional de
Acelerógrafos (RENAC) of the Instituto Geofísico de la Escuela
Politécnica Nacional (IG-EPN), Ecuador. The IG-EPN has used for the
installation and extension of the RENAC, resources of the Escuela
Politécnica Nacional, of the SENESCYT Project PIN-08-EPNGEO-0001
”Fortalecimiento del Instituto Geofísico: Ampliación y Modernización
del Servicio Nacional de Sismología y Vulcanología”, and of the Project
of investment ”Generación de Capacidades para la Difusión de Alertas
Tempranas y para el Desarrollo de Instrumentos de Decisión ante las
Amenazas Sísmicas y Volcánicas dirigidos al Sistema Nacional de
Gestión de Riesgos”.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in the
online version, at https://doi.org/10.1016/j.engstruct.2018.10.006.
J.P. Herrera et al.
References
[1] Geyer R, Jambeck JR, Law KL. Production, use, and fate of all plastics ever made.
Sci Adv 2017;3(7):e1700782. https://doi.org/10.1126/sciadv.1700782 http://
advances.sciencemag.org/lookup/doi/10.1126/sciadv.1700782 .
[2] Krishnaswamy BP, Lampo R. Recycled-plastic lumber standards: from waste plastics
to markets for plastic lumber bridges. World Standards Day: 2001 Paper
Competition; 2001. p. 1–12. <http://citeseerx.ist.psu.edu/viewdoc/download?
doi=10.1.1.551.129&rep=rep1&type=pdf>.
[3] Carroll DR, Stone RB, Sirignano AM, Saindon RM, Gose SC, Friedman MA.
Structural properties of recycled plastic/sawdust lumber decking planks. Resour,
Conserv Recycl 2001;31(3):241–51. https://doi.org/10.1016/S0921-3449(00)
00081-1.
[4] MacBain K, Saadeghvaziri MA. Analytical modeling of the mechanical properties of
recycled plastics. J Mater Eng Perform 1999;8(3):339–46. https://doi.org/10.1361/
105994999770346891 http://link.springer.com/10.1361/105994999770346891 .
[5] ASTM D6108-13. Standard test method for compressive properties of plastic lumber
and shapes. Tech rep; ASTM International; West Conshohocken, PA; 2013. https://
doi.org/10.1520/D6108 <www.astm.org > .
[6] ASTM D6109-13. Standard test methods for flexural properties of unreinforced and
reinforced plastic lumber and related products. Tech Rep; ASTM International; West
Conshohocken, PA; 2013. https://doi.org/10.1520/D6109 <www.astm.org>.
[7] ASTM D6111-13a. Standard test method for bulk density and specific gravity of
plastic lumber and shapes by displacement. Tech Rep; ASTM International; West
Conshohocken, PA; 2013. https://doi.org/10.1520/D6111 <www.astm.org>.
[8] ASTM D6112-13. Standard test methods for compressive and flexural creep and
creep-rupture of plastic lumber and shapes. Tech rep; ASTM International; West
Conshohocken, PA; 2013. https://doi.org/10.1520/D6112 <www.astm.org>.
[9] ASTM D6117-16. Standard test methods for mechanical fasteners in plastic lumber
and shapes, ASTM International, West Conshohocken, PA, 2016. Tech rep; ASTM
International; West Conshohocken, PA; 2016. https://doi.org/10.1520/D6117-16
<www.astm.org>.
[10] Li T, Silverstein MS, Hiltner A, Baer E. Mechanical properties of commingled plastic
from recycled polyethylene and polystyrene. J Appl Polym Sci 1994;52(2):315–27.
https://doi.org/10.1002/app.1994.070520216 http://doi.wiley.com/10.1002/
app.1994.070520216 .
[11] Gulhane S, Gulhane S. Analysis of housing structures made from recycled plastic. In:
Proceedings of the international conference on science & engineering for sustain-
able development; 2017, vol. 7. pp. 45–55. https://doi.org/10.21013/jte.
ICSESD201705 <https://research-advances.org/index.php/IRAJTE/article/view/
905>.
[12] Chen CW, Salim H, Bowders JJ, Loehr JE, Owen J. Creep behavior of recycled
plastic lumber in slope stabilization applications. J Mater Civil Eng 2007;February
(February):130. https://doi.org/10.1061/(ASCE)0899-1561(2007)19:2(130).
[13] Nosker TJ, Renfree RW. Recycled plastic lumber: from park benches to bridges. In:
Proceedings of R’2000 5th world congress; 2000. <http://www.environmental-
expert.com/Files%5C0%5Carticles%5C2181%5C2181.pdf>.
[14] Mclaren MG, Assis G, Ph D, Mclaren MG, Engineers PCC, Nyack W, et al.
Introducing the first recycled plastic bridge in the world. Composites Institute’s,
international conference proceedings. CRC press; 2002. p. 1–7.
[15] Chandra V, Kim JS. World’s first recycled plastic bridges. In: ICSDC 2011; 2012. p.
585–93. https://doi.org/10.1061/41204(426)72 <http://ascelibrary.org/doi/10.
1061/41204%28426%2972> [ISBN 9780784412046].
[16] Econciencia S.A.S.. Econciencia – Construcción sostenible; 2018. <http://www.
econciencia.com.co/>.
[17] Ficidet creamos materiales de construcción sostenible; 2018. <http://ficidet.
com/>.
[18] Bedoya-Ruiz D, Hurtado JE, Pujades L. Experimental and analytical research on
seismic vulnerability of low-cost ferrocement dwelling houses. Struct Infrastruct
Eng 2010;6(1–2):55–62. https://doi.org/10.1080/15732470802663789. URL
http://www.tandfonline.com/doi/abs/10.1080/15732470802663789 .
[19] Priestley MJN. Society of earthquake and civil engineering dynamics. Myths and
fallacies in earthquake engineering, revisited. IUSS Press; 2003 [ISBN
8873580092].
[20] Mostaghel N. Analytical description of pinching, degrading hysteretic systems. J
Eng Mech 1999;125(2):216–24. https://doi.org/10.1061/(ASCE)0733-9399(1999)
125:2(216) http://ascelibrary.org/doi/10.1061/%28ASCE%290733-9399%
281999%29125%3A2%28216%29 .
[21] ASTM Standard E2126-11. Standard test methods for cyclic (reversed) load test for
shear resistance of vertical elements of the lateral force resisting systems for
buildings. Tech rep; ASTM International; West Conshohocken, PA; 2011. https://
578
Engineering Structures 177 (2018) 566–578
doi.org/10.1520/E2126-11 <www.astm.org>.
[22] Priestley MJN, Seible FF, Calvi GMGM. Seismic design and retrofit of bridges.
Wiley; 1996. <https://www.wiley.com/en-us/Seismic+Design+and+Retrofit
+of+Bridges-p-9780471579984> [ISBN 9780471579984].
[23] Chopra AK. Dynamics of structures: theory and applications to earthquake en-
gineering. Prentice Hall; 2012 [ISBN 0132858037].
[24] Takada T, Hwang HHM, Shinozuka M. Response modification factor for multiple-
degree-of-freedom systems. Proc 9th World Conf Earthq Eng 1988;5:129–34 http://
www.iitk.ac.in/nicee/wcee/article/9_vol5_129.pdf .
[25] Miranda E, Bertero VV. Evaluation of strength reduction factors for earthquake-
resistant design. Earthq Spectra 1994;10(2):357–79. https://doi.org/10.1193/1.
1585778 http://earthquakespectra.org/doi/10.1193/1.1585778 .
[26] Yar M, Hammond J. Parameter estimation for hysteretic systems. J Sound Vib
1987;117(1):161–72. https://doi.org/10.1016/0022-460X(87)90442-1 https://
www.sciencedirect.com/science/article/pii/0022460X87904421 .
[27] Sues RH, Mau ST, Wen Y. Systems identification of degrading hysteretic restoring
forces. J Eng Mech 1988;114(5):833–46. https://doi.org/10.1061/(ASCE)0733-
9399(1988)114:5(833) http://ascelibrary.org/doi/abs/10.1061/(ASCE)0733-
9399(1988)114:5(833) .
[28] Garrido R, Rivero-Angeles FJ. Hysteresis and parameter estimation of Mdof systems
by a continuous-time least squares method. J Earthq Eng 2006;10(2):237–64.
https://doi.org/10.1080/13632460609350595 http://www.tandfonline.com/doi/
full/10.1080/13632460609350595 .
[29] Zhang H, Foliente GC, Yang Y, Ma F. Parameter identification of inelastic structures
under dynamic loads. Earthq Eng Struct Dynam 2002;31(5):1113–30. https://doi.
org/10.1002/eqe.151.
[30] Yang Y, Ma F. Constrained Kalman filter for nonlinear structural identification.
Modal Anal 2003;9(12):1343–57. https://doi.org/10.1177/1077546304031166
http://journals.sagepub.com/doi/10.1177/1077546304031166 .
[31] Zheng Xueliang, Huang Yumei, Liu Yongchao, Wang Xiaoyue, Gao Feng. A method
identifying the parameters of Bounc-Wen hysteretic nonlinear model based on ge-
netic algorithm. In: 1997 IEEE international conference on intelligent processing
systems (Cat. No.97TH8335); vol. 1. IEEE; 1997. p. 602–5. https://doi.org/10.
1109/ICIPS.1997.672855 <http://ieeexplore.ieee.org/document/672855/>
[ISBN 0-7803-4253-4].
[32] Giuclea M, Sireteanu T, Mitu AM. Use of genetic algorithms for fitting the Bouc-Wen
model to experimental hysteretic curves; 2009 (January 2015), pp. 3–10.
[33] Ortiz GA, Alvarez DA, Bedoya-Ruíz D. Identification of Bouc-Wen type models using
multi-objective optimization algorithms. Comput Struct 2013;114–115:121–32.
https://doi.org/10.1016/J.COMPSTRUC.2012.10.016 https://www.sciencedirect.
com/science/article/pii/S0045794912002441 .
[34] Ortiz GA, Alvarez DA, Bedoya-Ruíz D. Identification of Bouc-Wen type models using
the Transitional Markov Chain Monte Carlo method. Comput Struct
2015;146:252–69. https://doi.org/10.1016/J.COMPSTRUC.2014.10.012 https://
www.sciencedirect.com/science/article/pii/S0045794914002417 .
[35] Ye M, Wang X. Parameter estimation of the Bouc-Wen hysteresis model using
particle swarm optimization. Smart Mater Struct 2007;16(6):2341–9. https://doi.
org/10.1088/0964-1726/16/6/038.
[36] Charalampakis AE, Dimou CK. Identification of Bouc-Wen hysteretic systems using
particle swarm optimization. Comput Struct 2010;88(21–22):1197–205. https://
doi.org/10.1016/j.compstruc.2010.06.009.
[37] Kennedy J, Eberhart R. Particle swarm optimization. In: Neural Networks, 1995
Proceedings, IEEE international conference on 1995;4:1942–1948 vol.4. https://
doi.org/10.1109/ICNN.1995.488968.
[38] Arora JS. Introduction to optimum design. Academic Press; 2011. <https://www.
sciencedirect.com/science/book/9780123813756> [ISBN 9780123813756].
[39] CESMD. Center for engineering strong motion data; 2018a. <https://www.
strongmotioncenter.org/>.
[40] CESMD. Center for Engineering Strong Motion Data. Records courtesy of Professor
Ruben Boroschek, Engr. Pedro Soto, and Mr. Ricardo Leon of Red de Cobertura
Nacional de Acelerografos (RENADIC) of the Department of Civil Engineering,
University of Chile; 2018b. <http://www.strongmotioncenter.org/>.
[41] Takai N, Shigefuji M, Rajaure S, Bijukchhen S, Ichiyanagi M, Dhital MR, et al.
Strong ground motion in the Kathmandu Valley during the 2015 Gorkha, Nepal,
earthquake. Earth, Planets Space 2016;68(1):10. https://doi.org/10.1186/s40623-
016-0383-7 http://earth-planets-space.springeropen.com/articles/10.1186/
s40623-016-0383-7 .
[42] Singaucho JC, Laurendeau A, Viracucha C, Ruiz M. Observaciones del sismo del 16
de Abril de 2016 de magnitud Mw 7.8, Intensidades y Aceleraciones. Revista
Politécnica; 2016.