See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/254112642

Viscoelastic and viscoplastic behavior of a fully recycled carbon fiber-

reinforced maleic anhydride grafted polypropylene modified polypropylene
composite

Article  in  Journal of Composite Materials · June 2012

DOI: 10.1177/0021998311423858

CITATIONS READS

3 462

3 authors:

M. Szpieg Konstantinos Giannadakis

Swerea - Swedish Research Luleå University of Technology
7 PUBLICATIONS   67 CITATIONS    12 PUBLICATIONS   85 CITATIONS   

SEE PROFILE SEE PROFILE

Leif E. Asp
Chalmers University of Technology
102 PUBLICATIONS   1,649 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Kombatt View project

Contimuum modelling of composites consolidation View project

All content following this page was uploaded by Leif E. Asp on 07 November 2014.

The user has requested enhancement of the downloaded file.

 Journal of Composite Materials http://jcm.sagepub.com/

Viscoelastic and viscoplastic behavior of a fully recycled carbon fibre-reinforced maleic anhydride
grafted polypropylene modified polypropylene composite
M. Szpieg, K. Giannadakis and L. E. Asp
Journal of Composite Materials published online 21 October 2011
DOI: 10.1177/0021998311423858

The online version of this article can be found at:

http://jcm.sagepub.com/content/early/2011/10/20/0021998311423858

Published by:

http://www.sagepublications.com

On behalf of:

American Society for Composites

Additional services and information for Journal of Composite Materials can be found at:

Email Alerts: http://jcm.sagepub.com/cgi/alerts

Subscriptions: http://jcm.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

>> Proof - Oct 21, 2011

What is This?

Downloaded from jcm.sagepub.com at Lulea Univ. of Technology on October 25, 2011

XML Template (2011) [19.9.2011–5:03pm] [1–14]
K:/JCM/JCM 423858.3d (JCM) [PREPRINTER stage]

JOURNAL OF
COMPOSITE
Article M AT E R I A L S
Journal of Composite Materials
0(0) 1–15
! The Author(s) 2011
Viscoelastic and viscoplastic behavior Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
of a fully recycled carbon fiber-reinforced DOI: 10.1177/0021998311423858
jcm.sagepub.com
maleic anhydride grafted polypropylene
modified polypropylene composite

M. Szpieg1,2, K. Giannadakis1 and L. E. Asp1,2

Abstract
The effect of maleic anhydride grafted polypropylene (MAPP) coupling agents on properties of a new composite made of
recycled carbon fibers and recycled polypropylene (rCF/[rPP þ MAPP]) was studied experimentally. This new material
presented significantly improved properties, compared to the previous generation, without the addition of MAPP
(Giannadakis K, Szpieg M and Varna J. Mechanical performance of recycled carbon fibre/PP. Exp Mech 2010; published
online.). This was mostly attributed to improvement of the fiber/matrix interface. The inelastic and time-dependent
behavior of the MAPP modified composite material in tension was analyzed. A series of quasi-static tensile and creep
tests were performed to identify the material model, which accounts for: (a) damage-related stiffness reduction, (b) devel-
opment of stress and time-dependent irreversible strains described as viscoplasticity, (c) nonlinear viscoelastic behavior.
The damage-related stiffness reduction was found to be less than 10%. Although damage-dependent stiffness was not the
main source of nonlinearity, it was included in the inelastic material model. In creep tests, it was found that the time and
stress dependence of viscoplastic strains follows a power law, which makes the determination of the parameters in the
viscoplasticity model relatively simple. The viscoelastic response of the composite was found to be linear in the inves-
tigated stress domain. The material model was validated in constant stress rate tensile tests.

Keywords
polymer–matrix composites, mechanical properties, recycling

Introduction
Recycling is becoming more and more important
in composites production. New legislation polices,
increasing environmental awareness, and threat of
shortage of raw materials, in combination with the
ever increasing use of carbon fiber composites, force
society to recycle these materials. Carbon fibers, even
when trapped in composites, are valuable products
worth to be recovered. The energy needed to produce
pristine carbon fibers is very high, 286 MJ/kg.1 The high
stiffness and strength and low density of recovered
carbon fibers allow them to be reused in new engineer-
ing materials for use in lightweight components.2–4
With carbon fiber composites, recycling technologies
involving burning or chemical decomposition of the
polymer matrix have been developed2,5–7 in order
to process carbon fiber-reinforced polymer (CFRP)
composite waste.
There are several methods to produce composites
using recovered carbon fibers. Injection moulding
(IM) and bulk moulding compound (BMC)8 are two
direct methods of remoulding recovered carbon fibers
into recycled composite materials. In the IM process,
material is fed into a heated barrel, mixed, and forced
into a mould cavity, where it cools and hardens to the
configuration of the mould cavity. The BMC is a com-
bination of chopped short fibers and resin in the form of
a bulk pre-preg. The BMC is suitable for either com-
pression or IM. Wong et al.9 manufactured a
1
Division of Polymer Engineering, Luleå University of Technology, SE-971
87 Luleå, Sweden.
2
Swerea SICOMP AB, P.O. Box 104, SE-431 22 Mölndal, Sweden.
Corresponding author:
L.E. Asp, Swerea SICOMP AB, P.O. Box 104, SE-431 22 Mölndal, Sweden.
Email: leif.asp@swerea.se

Downloaded from jcm.sagepub.com at Lulea Univ. of Technology on October 25, 2011

XML Template (2011) [19.9.2011–5:03pm]
K:/JCM/JCM 423858.3d (JCM)
2 Journal of Composite Materials 0(0)
polypropylene composite reinforced with recycled
carbon fibers (rCFs) using IM. By this process fibers
aligned in one direction, which resulted in high stiffness
of the composite material. Turner et al.10 developed a
moulding compound process, involving both sheet
moulding compound (SMC) and BMC. The formula-
tion of the BMC was tuned to overcome the poor flow
properties of the resin and the entangled form of the
fibers.
Another method to manufacture composite materials
with longer fibers, using recovered carbon fibers is to
produce the intermediate dry two-dimensional (2D) or
three-dimensional (3D) nonwoven preforms.2 The 2D
or 3D nonwoven dry products are then either compres-
sion moulded with resin layers11,12 or reimpregnated
through a liquid resin process.13 In Pickering et al.,11 a
rCF SMC was manufactured using an intermediate mat
material. Mats made of carbon fibers recovered using a
fluidized bed process were manufactured employing a
paper-making process and interleaved with epoxy
resin film. In Wong et al.,12 fibers recovered via a pyrol-
ysis process and converted into a randomly oriented mat
were employed to manufacture a composite material
through a resin film infusion process in a matched die
tool. This composite was characterized in terms of its
microstructure and mechanical performance in Pimenta
et al.3 Janney et al.13 introduced the three dimensional
engineered preform (3-DEP) process for making highly
uniform, complex geometry, net shape, chopped fiber
preforms for polymer matrix composites. This process
was developed for virgin fibers by Materials Innovation
Technologies LLC14 and employed by Janney et al.13
to make preforms for high-quality composite parts
using rCF.
Fiber alignment is a crucial part to improve the
mechanical performance of composite materials manu-
factured using rCF. A study by Jiang et al.15 demon-
strated a method to convert fluffy rCFs into a controlled
form. A hydrodynamic centrifugal method was
employed to produce mats with high degree of fiber
alignment.15 In the 3-DEP process developed by
Janney et al.16 rCFs are aligned circumferentially in a
cone by centrifugal forces.
In the present work, a fully rCF/recycled polypropyl-
ene (rPP) composite material was manufactured using
press forming. In this process, the fiber mats and poly-
mer films are stacked, preheated, and press formed in a
warm tool, see for example, Åström.8 It is well known
that the fiber/matrix interface properties are often the
weakest point controlling performance of a composite
material.17 In order to ensure good interfacial adhesion
and load transfer across the interface in the rCF/rPP
composite a maleic anhydride grafted polypropylene
(MAPP) coupling agent has been added to the rPP
matrix. The MAPP improves the interface bonding
Downloaded from jcm.sagepub.com at Lulea Univ. of Technology on October 25, 2011
[1–14]
[PREPRINTER stage]
between the fiber and polymer matrix by two mecha-
nisms. First, the long molecular chain is responsible
for chain entanglements and co-crystallization with
the nonpolar PP matrix. These entanglements provide
mechanical integrity to the host material. Second, the
anhydride groups chemically interact with the func-
tional groups on the fiber surface.9 The presence of
MAPP has been found to improve the interface and
increase the stiffness and strength of injection moulded
rCF/rPP composites.9 In this work, the approach taken
in Wong et al.9 is extended to press formed rCF/rPP
composites.
When a polymer composite material is mechanically
loaded, its deformation may increase with time even if
the load is constant due to their inherent viscoelastic
nature. There are several methods available in the liter-
ature to characterize materials viscoelastic behavior.18
The most common approach is to perform a creep test
using dead load application. The stress applied to the
specimen is constant and strain is recorded as a function
of time. The viscoelastic response may be either linear or
nonlinear depending on stress level. For low stresses, the
response is usually linear, while for higher stresses it is
often found to be nonlinear. The development of visco-
elastic strains has been expressed in a general nonlinear
constitutive equation for uniaxial loading by Lou and
Schapery,19 which is employed later in this article.
In addition to viscoelastic strains developing in a
material due to mechanical loading also viscoplastic
strains (VP-strains) may occur. At unloading, after
creep loading at high stress levels, often the strain
does not fully recover. The development of VP-strain
is a function of both stress and time. If the specimen is
loaded in creep many times (or for very long time) then
the rate of VP-strain development will eventually slow
down and become negligible for that particular level of
stress and lower stresses. The specimen would then be
‘conditioned’ and thus only pure viscoelastic creep
strain would develop upon loading. The development
of VP-strains has been expressed via a nonlinear func-
tion presented by Zapas and Crissman20 employed later
in this article.
Research on time-dependent properties is needed by
the increased use of polymeric composites for structural
components, which require long-term durability. Most
of the papers in the literature report properties of
recycled carbon fiber-reinforced polymer (rCFRP) con-
cerning only the main mechanical properties of the com-
posite materials such as stiffness and strength.9–11,15,21
However, loading curves of composites with short, ran-
domly oriented fibers, which are considered in this arti-
cle, are usually nonlinear.22 In the previous study,22 one
reason for the nonlinearity was a very weak fiber/matrix
interface, which led to poor mechanical performance of
the composite material. As a result, these thermoplastic
XML Template (2011) [19.9.2011–5:03pm]
K:/JCM/JCM 423858.3d (JCM)
Szpieg et al.
matrix/short fiber composites experience a complex,
time-dependent stress–strain response with loading
rate effects, hysteresis loops, and development of per-
manent strain. The most common reasons for this
response are microdamage development, resulting in
stiffness degradation, viscoplasticity, which may be
related to submicrodamage, and nonlinear viscoelastic-
ity of both constituents in the composite.23–25
The objective of this article is to perform a thorough
investigation of the inelastic behavior of a recycled
carbon fiber-reinforced MAPP modified recycled poly-
propylene (rCF/[rPP þ MAPP]) composite material.
This new rCF/(rPP þ MAPP) composite is introduced
here as an improved version of the rCF/rPP material
developed in Giannadakis et al.22 The sources of the
inelastic behavior will be analyzed by studying the
output from the strain response for different load
cases. The significance of the following phenomena
will be ranked and included in an inelastic material
model; (a) reduced elastic modulus due to microdamage
development at high stresses, (b) viscoplastic (irrevers-
ible) strains development with time at high loads, and
(c) viscoelastic composite material behavior, which may
be nonlinear with respect to stress.
The inelastic material model
In the development of the inelastic material model, it is
assumed that the damage development, viscoelastic, and
viscoplastic responses may be decoupled. It should be
pointed out that this assumption may be inaccurate,
especially at high stresses. Separation of the microdam-
age influenced by the viscoelastic and VP-strain
responses (which are also affected by damage) gives:
eðs,tÞ ¼ dðsmax ÞeVE ðs, tÞ þ dðsmax ÞeVP ðs, tÞ:
where, d(smax) is the function describing the effect of
microdamage, eVE and eVP are the viscoelastic and
VP-strains, respectively, s is the applied stress and t is
the time. To describe the viscoelasticity, the general
thermodynamically consistent theory of nonlinear vis-
coelastic materials developed by Lou and Schapery19
and presented in more general form by Schapery26 is
used in this article. In uniaxial loading considered in
this article, the viscoelastic model contains three stress
invariant dependent functions, namely g1, g2, and as,
which characterize the nonlinearity. If a region with
linear viscoelastic behavior exists, g1 ¼ g2 ¼ as ¼ 1. In
addition, the function d(smax) with values dependent on
the maximum (most damaging) stress state experienced
in service is incorporated to account for microdamage.25
In monotonously increasing uniaxial loading, the ‘most
damaging’ is the current stress state. As demonstrated
below (see Equation (8)), the damage function
Downloaded from jcm.sagepub.com at Lulea Univ. of Technology on October 25, 2011
[1–14]
[PREPRINTER stage]
3
represents the normalized elastic compliance of the
composite, which is degrading due to experienced high
stress/strain. Certainly, the form of interaction between
d and VP-strain eVP used in Equation (1) can be
debated. In Marklund et al.25 it was concluded from a
limited number of empirical observations that the inter-
action as proposed in Equation (1) fits the test results
better than does a standalone term of eVP. Actually, it
states that larger irreversible strains develop in a dam-
aged composite than in an undamaged composite. It
has to be noted that microdamage itself without any
viscoplasticity can introduce irreversible strains. The
final form of the material model written for the one-
dimensional case is:
0 1
ðt
  d ðg 2 sÞ
e ¼ dðsmax Þ  @e0 þ g1 S c  c0 d þ eVP ðs, tÞA
d
0
ð2Þ
In Equation (2) integration is over ‘reduced time’
introduced as:
ðt ð 0
dt0 0 dt
c¼ and c ¼ : ð3Þ
as as
0 0
In Equations (2) and (3) e0 represents the elastic
strain, which may be nonlinear with respect to stress.
The S(c) is the transient component of the linear
viscoelastic creep compliance, g1 and g2 are stress-
dependent material properties, as is the shift factor,
which under fixed conditions is a function of stress
only, and  is representing a normalized integration
time variable. These functions also depend on tempera-
ð1Þ ture and humidity, however, under fixed environmental
conditions they are functions of stress only. It was
shown by Schapery26 that the viscoelastic creep compli-
ance can be written in the form of Prony series:
  
X c
SðcÞ ¼ Ci 1  exp  ð4Þ
i
i
where Ci are stress-independent constants and  i are
constants, so-called, retardation times. Below descrip-
tions of the decoupled damage development, viscoplas-
tic and viscoelastic responses are presented. Strategies to
obtain these are outlined as well.
Damage development and stiffness degradation
The physical meaning of the damage-related function
d(smax) can be revealed by comparing the elastic strains
(eel) in damaged and undamaged composite at given low
stress s (i.e., linear regime).
XML Template (2011) [19.9.2011–5:03pm] [1–14]
K:/JCM/JCM 423858.3d (JCM) [PREPRINTER stage]

4 Journal of Composite Materials 0(0)


eel ¼ s Eðsmax Þ e0 ¼ s=E0
In Equation (5), E0 is the elastic modulus of the unda-
maged composite and E(smax) is the elastic modulus
after applying a load corresponding to a maximum
stress smax. Excluding stress from Equation (5), the elas-
tic strain is obtained as:
ð5Þ Assuming that Equation (9) is valid for the analyzed
material, it is possible to design test procedures for
determination of the parameters CVP, M, and m. First,
a sequence of creep tests at fixed stress, s(t) ¼ s0 is to be
conducted. In this case, the integration in Equation (9)
is analytical and the VP-strain accumulated during the
time interval t 2 [0;t1] is:

E0 t m
1
eel ¼ e0 ð6Þ eVP ðs0 ,t1 Þ ¼ CVP sMm
0 ð10Þ
Eðsmax Þ t

From Equation (1), the elastic response of the dam- which can also be presented in the form:
aged material may be written as:
 t m
eVP ðs0 ,tÞ ¼ Aðs0 Þ ð11Þ
eel ¼ dðsmax Þe0 ð7Þ t

Comparing with Equation (6), d(smax) is identified as: A ¼ CVP s0 ,  ¼ M  m ð12Þ

E0 If the creep test is performed for a time interval t1 þ

dðsmax Þ ¼ ð8Þ
Eðsmax Þ t2, the accumulated VP-strain will be:
t þ t m
Obviously d(smax) has the physical meaning of nor- 1 2
eVP ðs0 ,t1 þ t2 Þ ¼ CVP sMm
0 ð13Þ
malized elastic compliance, which is inverse to the nor- t
malized elastic modulus dependence on the highest
experienced stress level. According to Equation (9), interruption of the con-
In this study, the elastic modulus dependence on the stant stress test for an arbitrary time interval tr (for
previously applied maximum strain/stress level is mea- example, strain recovery) has no effect on VP-strain
sured to evaluate the significance of the modulus degra- development. During strain recovery at any arbitrary
dation and the amount of microdamage accumulation. point in time, ta between t1/t* and tr/t* in Equation
(14), the stress is zero and only the total loading time
(t1 þ t2) is of importance.
Viscoplastic strains
The development of VP-strains in composites has eVP ðs,t1 þ ta þ t2 Þ
been described by a functional developed by Zapas 8 9m
> t1 =  tr =  tr þ t2 =  >
and Crissman.20 This approach has been successfully >
< ð t ð t ð t >
=
M M
used for modeling of natural fiber composites by Varna ¼ CVP s0 dþ 0d þ s0 d
>
> >
>
and co-workers23–25 and rPP þ MAPP.27 It has also been : 0 t1 =  tr =  ;
applied with limited success to a previous version of the t t
t þ t m
rCF/rPP composite without any modification by MAPP ¼ CVP sMm
1 2
ð14Þ
0
to the rPP.22 Below, the model is presented in a slightly t
different form introducing normalized time  ¼ t/t*.
As a consequence, instead of testing at a stress s0 for
8 9m
> t=  > the time t1 þ t2 continuously, one could perform the
>
< ðt >
= testing in two steps: (1) creep at stress s0 for time t1
eVP ðs,tÞ ¼ CVP sð ÞM d ð9Þ unloading the specimen and measuring the permanent
>
> >
>
: 0 ; strain e1VP developed during this step. During the creep
test, viscoelastic strains are also developing simulta-
where CVP, M, and m are stress-independent constants neously with VP-strains. Therefore, the VP-strains
to be determined, and t* is a characteristic time con- cannot be measured immediately after load removal.
stant, for example, 3600 s if the time interval of inter- It has to be done after the recovery of viscoelastic
est is 1 h. This normalized time is used in order to strains; (2) the stress s0 is applied again for time t2
have a CVP expression independent on time in and after strain recovery, the new VP-strain e2VP devel-
Equation (9). A generalization of Equation (9) is pos- oped during the second test of length t2 is measured. The
sible, if necessary, by replacing s()M with a more total VP-strain after these two creep tests is found by
general function f(s()). adding the two VP-strains, e1VP and e2VP :

Downloaded from jcm.sagepub.com at Lulea Univ. of Technology on October 25, 2011

XML Template (2011) [19.9.2011–5:03pm]
K:/JCM/JCM 423858.3d (JCM)
Szpieg et al.
etVP
1 þt2
¼ etVP
1
þ etVP
2
From the above discussion, it follows that the time
dependence of VP-strain at a fixed stress level can be
determined by performing a sequence of creep and
strain recovery tests on the same specimen.
The VP-strain time dependence in creep loading at
the same stress level s0 is given by Equation (11). Hence,
the test data should be fitted with a power function with
respect to time. According to Equation (9), the constant
m determined as the best fit by a power function has to
be independent on the stress level used in the creep test
(it has to be proven experimentally). A(s0) is also
obtained in the fitting procedure. Following Equations
(10) and (12), the stress dependence of VP-strains in
creep tests (and consequently A(s0)) also has to obey
a power law. Performing the above described sequences
of creep tests at several stress levels, it is possible to plot
A(s0) vs. stress and fit it using Equation (12). In result,
constants M and CVP are determined. After that the VP-
strain model, Equation (9) can be applied to any arbi-
trary stress controlled loading ramp.
The assumption that the creep test interruption does
not affect the VP-strain development has to be checked
experimentally for the analyzed material. In other
words, the assumption and validation that VP-strains
are accumulative is an important step in validation of
the VP-strain model (Equation (9)).
Viscoelastic strains
In a creep test, the stress is applied at t ¼ 0 and kept
constant for a certain time t1. Then, as the load is
removed, the strain recovery period begins according
to s ¼ s[H(t) – H(t – t1)], where H(t) is the Heaviside
step function.
The material model (see Equations (2)–(4)) may
therefore be applied separately to the creep interval t
2 [0, t1] and the strain recovery following the creep
test at t > t1. The creep and recovery strain development
(ecreep and erec)25 is described in Equations (16) and (17):
  
X t
ecreep ¼ e0 þ g1 g2 s Ci 1  exp  þ eVP ðs, tÞ
i
as  i
X     
t1 t  t1
erec ¼ g2 s Ci 1  exp  exp  þ eVP ðs, t1 Þ
i
as  i i
The time-dependent VP-strain has to be subtracted
from the measured strains to have a pure viscoelastic
response to analyze. The VP-strain after creep test of
length t1 is given by Equation (10). As it is valid for
Downloaded from jcm.sagepub.com at Lulea Univ. of Technology on October 25, 2011
[1–14]
[PREPRINTER stage]
5
ð15Þ any arbitrary instant in time, a convenient form for sub-
traction can be obtained by using:
 m
t
eVP ðs0 , tÞ ¼ eVP ðs0 , t1 Þ ð18Þ
t1
Equations (16) and (17) are used to fit the experimen-
tal creep and strain recovery data. Stress-independent
constants Ci, i ¼ 1, . . ., I and stress–dependent e0, as, g1,
and g2 are found by data fitting, using the method of
least squares (LSQ). The retardation times  i in the
Prony series are chosen arbitrary, but the largest  i
should be at least a decade larger than the length of
the conducted creep test. In the current investigation,
the retardation times were spread more or less uni-
formly over the logarithmic time scale, with a factor
of about 10 between them. Even if it is not a necessary
condition, the additional requirement Ci > 0 is used as it
has been found to improve the fit to test data.25
Experimental
Materials and composite preparation
Recovered carbon fibers. Recovered carbon fibers
CFM-PYR-H-128 were supplied by Hadeg Recycling
GmbH, Germany. The fibers were produced by milling
of carbon fiber composites, which were obtained from
production waste generated by the aircraft industry.
The fibers had been recovered using a pyrolysis process
at approximately 1200 C. Due to the cotton-like struc-
ture of the recovered carbon fibers (see Figure 1(a)) they
had to be reprocessed into fiber mats with a controlled
fiber distribution.29 The fiber length varied between 1
and 20 mm. A scanning electron microscope (SEM)
was used to analyze the morphology of the recovered
carbon fibers. As depicted in Figure 1(b), there is some
contamination, stemming from the pyrolysis process,
present on the fibers. However, the degree of contami-
nation is low and the fiber quality is comparable to the
fibers recovered via some other recovery method avail-
able, for example, the fluidized bed process.21
Recovered carbon fiber preform manufacture and
ð16Þ
compaction response. The procedure to manufacture
a rCF preform was first presented in Szpieg et al.29 By
this approach, principles of papermaking were
employed to obtain preforms with a uniform fiber dis-
tribution for subsequent composite manufacturing. A
ð17Þ total of 40 g of the recovered carbon fibers were dis-
persed in distilled water using a mixer as shown in
Figure 2. The dispersion duration of the fibers was
approximately 5 min. The cylindrical plastic barrel had
a bottom area of 0.071 m2, and was 0.50 m high. The
rotational speed of the stirrer was 100 rpm. After
XML Template (2011) [19.9.2011–5:03pm]
K:/JCM/JCM 423858.3d (JCM)
6 Journal of Composite Materials 0(0)
Figure 1. Recycled carbon fibers: (a) cotton-like structure, view with the naked eye, (b) individual fibres (7.5 k).
Figure 2. Dispersion of the recycled carbon fibers.
dispersion, a filter was used to drain off the water and
protect the fiber mat from damage. The fiber mat was
dried on the filter for approximately 12 h at room tem-
perature, after which it was removed from the filter and
cut into the preform shape. The carbon fiber preform
with the area of 0.022 m2, still containing some
moisture, was then placed in a Vacuum Oven
GALLENKAMP for approximately 5 h at 100 C to
dry. The nominal areal weight of the dry rCF preform
was 0.30 kg/m2. Compaction experiments have previ-
ously been performed to characterize the out of plane
mechanical behavior of the preform.29 The mechanical
response of the rCF preform indicated that the fibers
were randomly distributed in the plane.29
Recycled polypropylene compounding and film
processing. PUREÕ process scrap polypropylene was
supplied by Lankhorst Indutech bv, The Netherlands,30
in the form of tape wound onto bobbins. The mechan-
ical and thermal properties of the reprocessed PUREÕ
scrap were earlier characterized by Szpieg et al.30 In the
current study, the polypropylene material was com-
pounded with MAPP coupling agents in order to
improve the interfacial adhesion with the rCFs22 in
the composite material to be manufactured. The
Downloaded from jcm.sagepub.com at Lulea Univ. of Technology on October 25, 2011
[1–14]
[PREPRINTER stage]
Figure 3. Press forming used to manufacture the recycled
polypropylene film and the composite material.
compounding process was performed using a laboratory
scale extruder, BrabenderÕ AEV 651. A total of 35 g of
PP scrap (the maximum possible amount of material in
the Brabender) was blended with 10% by weight of
MAPP. The blending was performed during 7 min at a
temperature of 200 C, and the rotating speed of the
screws was 80 rpm. 7 min for the extrusion time was
chosen, as it mimics a typical extrusion time for poly-
propylene using an industrial-scale extruder.30
The compounded rPP with MAPP (rPP þ MAPP)
was reprocessed into a film with a nominal thickness of
0.30 mm by press forming using a HASCOÕ Z121/196
196/5 press (see Figure 3). The dimensions of the man-
ufactured film were approximately 150  150 mm. The
film was manufactured by pressing with a load of
150 kN at a temperature of 200 C for 5 min, and
during the subsequent 5 min cooling ramp. The nomi-
nal areal weight of the rPP films was approximately
0.66 kg/m2.29
As reported in Szpieg et al.,31 the properties of
the rPP (elasticity, strength, viscoelasticity, and
XML Template (2011) [19.9.2011–5:03pm]
K:/JCM/JCM 423858.3d (JCM)
Szpieg et al.
viscoplasticity) are practically unaffected by the
presence of MAPP. Therefore, any composite strength
increase is attributed to the improved fiber/matrix inter-
face due to the addition of MAPP. This assumption is
reasonable, as both the elastic properties of the compos-
ites with and without MAPP were almost the same.
Composite manufacturing. Two polypropylene films
were introduced between three carbon fiber preform
layers in a stack (e.g., rCF/rPP/rCF/rPP/rCF). The
stack was heated and pressed in the press tool used for
rPP film manufacturing, resulting in a composite plate
thickness of approximately 1.60 mm. Using a compac-
tion law,29 the required amount of the constituents (rPP
film and rCFs preforms) was calculated to obtain a
nominal fiber volume fraction of 30%. The following
processing parameters were used: press time of 7 min
at 200 C, an applied pressure of 13 MPa, and subse-
quent cooling to room temperature for 5 min.
Mechanical characterization of the
recycled composite
All tests were performed at room temperature (23 C)
and relative humidity of 20%. Rectangular-shaped sam-
ples with dimensions 150  20 mm were cut and pol-
ished for the mechanical investigation. For all tests, an
INSTRON 3366 testing machine was used, equipped
with a 10 kN load cell, measuring strains in the axial
direction using an extensometer with 50 mm gage
length. The clamping area was 30 mm on both sides of
the specimens, and sandpaper was used to avoid sample
sliding from the clamps during the tensile loading. The
gage length between the grips in the testing machine was
90 mm.
Strain rate effect. The stress–strain responses were
obtained in monotonic tensile loading until failure.
Two different loading rates were used, 0.05 mm/min
and 5 mm/min.
Elastic modulus. The elastic modulus of the rCF/
(rPP þ MAPP) composite was determined in a standard
quasi-static tensile test. The crosshead speed was 5 mm/
min, corresponding to a strain rate of approximately
5% strain per minute. The specimen was loaded up to
0.2% strain and unloaded. The elastic modulus was cal-
culated using the secant modulus in the stress–strain
curve in the strain region between 0.05 and 0.15%.
This strain level was used expecting that damage and
irreversible phenomena would not develop at these rel-
atively low strains. Due to the possibility of poorly man-
ufactured samples (e.g., at positions close to an edge), an
elastic modulus measurement was performed on all of the
considered samples prior to any further testing. This was
Downloaded from jcm.sagepub.com at Lulea Univ. of Technology on October 25, 2011
[1–14]
[PREPRINTER stage]
7
done in order to exclude samples exhibiting extreme
values of the elastic modulus. Samples exhibiting an elas-
tic modulus diverging more than 10% from the mean
elastic modulus were excluded from further investigation.
In total, 47 samples were tested.
Stiffness degradation. The elastic properties may
degrade due to microdamage development as high stres-
ses. Thus, the elastic modulus dependence on the previ-
ously applied maximum stress level was measured to
evaluate the significance of the modulus degradation
during the loading history.
The loading ramp used in the tests is illustrated in
Figure 4. It consists of a sequence of blocks, each con-
taining the following steps:
. loading to 0.15% strain followed by unloading for
measuring of the elastic modulus,
. waiting at ‘almost zero’ stress for decay of all visco-
elastic effects (constant load of about 5 N, corre-
sponding to approximately 0.14 MPa is defined as
‘almost zero’),
. loading up to a certain stress level and unloading to
‘almost zero’ stress for a time of ti,
. waiting at ‘almost zero’ stress for decay of all visco-
elastic effects during a time period of 10ti,
. elastic modulus determination by applying 0.15% of
strain in addition to the residual value of the previous
loading step. The same sequence was repeated for a
higher level of applied strain. The stress levels applied
were: 14, 20, 30, 40, 45, 47, and 50 MPa.
In each of the multiple high load steps (see Figure 4),
irreversible strains were developed. Hence, after unload-
ing to ‘almost zero’ stress and a relatively large recov-
ery period, rather large strains were still present.
Consequently, the modulus determination in the strain
region between 0.05 and 0.15% as described in section
Elastic modulus was not possible. Instead of using the
Figure 4. Example of strain dependence on time during one
cycle of the test.
XML Template (2011) [19.9.2011–5:03pm] [1–14]
K:/JCM/JCM 423858.3d (JCM) [PREPRINTER stage]

8 Journal of Composite Materials 0(0)

strain region between 0.05 and 0.15%, the correspond-

ing stress region was found in the first loading curve and
this stress region was then used to calculate the elastic
modulus in the following loading steps.

Viscoplastic–viscoelastic testing and characterization

Creep. A series of creep tests were performed to obtain
the parameters in the inelastic material model, which
accounts for development of stress- and time-dependent
irreversible strains described as viscoplasticity, and non-
linear viscoelastic behavior. For the viscoelastic charac-
terization, single creep tests with the duration of 40 min
and the recovery of eight times the loading period were
performed. The tests were performed using a creep test- Figure 5. Sequence of creep and strain recovery steps to
ing rig, utilizing dead weights and a standard extensom- identify the time dependence of viscoplastic strains (VP-strains) at
eter. The extensometer recorded the development of a fixed stress level.
strains during the creep and the following period of
strain recovery. The tests were performed at room tem-
perature (23 C) and relative humidity of 20%.

Viscoplastic accumulation. The test program to iden-

tify the time dependence and stress dependence of VP-
strains in creep tests (constants: CVP, M and m) is based
on the discussion in section Viscoelastic strains. To iden-
tify the time dependence of VP-strains, a sequence of
steps was selected at a fixed level of stress, each consist-
ing of creep and strain recovery. An INSTRON 3366
testing machine with a 10 kN load cell in load controlled
mode was used, measuring strains in the loading direc-
tion (axial direction) using an extensometer, and mea-
suring transverse strains using a strain gauge. The test
sequence performed on the same specimen is schemati-
cally illustrated in Figure 5.
Often, the VP-strain rate decreases with time.
Therefore, the length of the creep loading time tk in
each of the following steps was increased proportionally
increasing also the strain recovery time. The time scale
depends on the time region to be covered in expected
application of the model. Creep loading steps of dura-
tions t1 ¼ 3, t2 ¼ 10, t3 ¼ 20, and t4 ¼ 30 min were used.
The strain recovery time after the load application step
was eight times the length of the loading step. The whole
creep–recovery curve was recorded (see Figure 6) for
axial strain and transverse strain development at 37
and 40 MPa. The VP-strains are given by the final recov-
ery strain value. According to the model (see Equation
(14)), the VP-strains are not developing during the strain
recovery part and when the same load is applied again
in the next step, the viscoplastic process continues with-
out any effect of the test interruption time interval.
In the example shown in Figure 6, four data points
for the specimen were obtained to establish the VP-
strain dependence on time for the stress level of
37 MPa. This test was performed at several stress
Figure 6. Axial and transverse viscoplastic strain (VP-strain)
development in a creep test; s0 ¼ 37 MPa.
levels (using other specimens) to make sure that the
shape of the time dependence is not changing with
stress. Often, the VP-strain developed during the first
loading step is larger than the VP-strain created
during the following step. It can be seen in Figure 6
that the increase of VP-strains is slowing down with
time.
Validation test at constant stress rate. The achieved
inelastic material model was used to simulate the
response of the rCF/(rPP þ MAPP) composite in a con-
stant stress rate test. A validation test was conducted,
using an INSTRON 3366 testing machine equipped
with a 10 kN load cell. A stress rate of 1.0 MPa/min
was applied until specimen failure. The duration of
the test was approximately 40 min. Simulation was
performed using a numerical integration to obtain VP-
strain according to Equation (9). Since in this case ana-
lytical integration is also possible, numerical results
were compared with the analytical ones, and the time

Downloaded from jcm.sagepub.com at Lulea Univ. of Technology on October 25, 2011

XML Template (2011) [19.9.2011–5:04pm]
K:/JCM/JCM 423858.3d (JCM)
Szpieg et al.
Figure 7. Stress–strain curves for the recycled carbon fiber
(rCF)/(recycled polypropylene [rPP] þ maleic anhydride grafted
polypropylene [MAPP]) composite at two crosshead speed rates,
5 mm/min and 0.05 mm/min.
Table 1. Mechanical properties of the rCF/(rPP þ MAPP)
composite
Crosshead Tensile
speed (mm/min) strength (MPa)
0.05 47.8 (5.1)
5 50.1 (4.7)
rCF: recycled carbon fiber, rPP: recycled polypropylene, MAPP: maleic
anhydride grafted polypropylene.
step in numerical integration was selected to obtain
coinciding results. Incremental procedure presented in
Megnis et al.,32 and used also in Szpieg et al.,29 was
applied for the viscoelastic part of the model.
Results and discussion
Strain rate effect
The loading curves for the MAPP modified recycled
composite (rCF/[rPP þ MAPP]) are shown in Figure 7.
No effect of strain rate on stiffness and strength can
be observed in the test data (only three specimens were
tested for each case not allowing for any statistical anal-
ysis). In contrast, a strong effect of strain rate on strain
to failure is evident, see Figure 7. The effect of the strain
rate on the mechanical properties, that is, the strain to
failure and tensile strength are presented in Table 1.
Three samples of the rCF/(rPP þ MAPP) were used
for each of the mean values presented.
Elastic modulus
To determine E0 in Equations (5)–(8), a standard tensile
test at 5 mm/min was performed. In the elastic modulus
measurement, the average value of the initial modulus
Downloaded from jcm.sagepub.com at Lulea Univ. of Technology on October 25, 2011
[1–14]
[PREPRINTER stage]
9
was 8.7 GPa with a standard deviation of 2.1 GPa. The
large scatter is attributed to still existing structural inho-
mogeneity such as resin-rich layers.22 The elastic mod-
ulus for the MAPP modified composite is consistent
with that of the unmodified material, for which values
between 6.5 and 9.5 GPa were reported in Giannadakis
et al.22 The stiffness of the rCF/(rPP þ MAPP) compos-
ite is significantly lower than that of another rCF/rPP
composite manufactured by IM reported by Wong
et al.9 The stiffness of the composite reported in Wong
et al.9 was approximately 14 GPa, compared to the
8.7 GPa for the current material (although the volume
fraction of fibers was significantly lower in the material
by Wong et al.9). This is unexpected as one would think
that reinforcing fibers become shorter during an IM
process, causing a stiffness reduction. Even though the
fibers become shorter, they also become aligned during
this process, which explains the higher stiffness of the
injection moulded composite.
Strain to Effect of MAPP on the tensile strength
failure (%) of the recycled composite
1.26 (0.19) Micrographs of fracture surfaces of the current compos-
0.84 (0.07) ites with MAPP and these previously investigated with-
out MAPP22 are depicted in Figure 8. The SEM
micrograph of the fracture surface of a tensile specimen
without the addition of MAPP is depicted in Figure 8(a).
This figure shows considerable amount of fiber pullouts
and clean matrix-free fiber surfaces indicating a very
weak fiber/matrix interface. Figure 8(b) shows the frac-
ture surface of the MAPP modified composite. The fibers
are fully covered by the matrix and there is no pullouts
indicating a good adhesion between the constituents.
The improved fiber/matrix interface results in much
improved tensile strength of the composite, see Table 2.
In fact, addition of MAPP has almost doubled the tensile
strength, while the elastic modulus remained practically
unchanged. The results presented in Table 2 come from
tensile tests of three samples of the rCF/(rPP þ MAPP)
composite and two samples of the rCF/rPP composite.
Stiffness reduction
Figure 9 presents the elastic modulus dependence on the
applied stress for the rCF/(rPP þ MAPP) composite.
The experimental data were fit by functions as:
EðsÞ 1  4 s 5 20MPa
¼ ð19Þ
E0 1  0:1 s20
32 s  20MPa
A partial conclusion, derived from Figure 9, is
that the observed stiffness reduction during these
tests is small, in fact less than 10%. Although damage-
dependent stiffness as expressed in Equation (19) is not
XML Template (2011) [19.9.2011–5:04pm] [1–14]
K:/JCM/JCM 423858.3d (JCM) [PREPRINTER stage]

10 Journal of Composite Materials 0(0)

Figure 8. Scanning electron microscope (SEM) micrographs showing the positive effect of maleic anhydride grafted polypropylene
(MAPP) on fiber/matrix interface strength: (a) recycled carbon fiber/recycled polypropylene (rCF/rPP), (b) rCF/(rPP þ MAPP).

Table 2. Effect of MAPP on the mechanical properties of the rCF/(rPP þ MAPP) and rCF/rPP composites

rCF/(rPP þ MAPP) rCF/rPP1

Material property Sample 1 Sample 2 Sample 3 Mean value Sample 1 Sample 2 Mean value

Tensile strength (MPa) 54.1 51.3 44.8 50.1 27.8 28.0 27.9
Elastic modulus (GPa) 8.7 9.7 8.6 9.0 7.9 11.4 9.6
MAPP): maleic anhydride grafted polypropylene, rCF: recycled carbon fiber, rPP: recycled polypropylene.

Figure 9. Composite elastic modulus reduction due to micro-

damage accumulation in high stress region.
the main source of nonlinearity, it is included in the
inelastic material model.
Viscoplasticity
The VP-strain evolution in specimens at 37 MPa and
40 MPa (one specimen for each case) is shown in log–
log axes in Figure 10. According to Equation (11), the
dependence in log–log axes should be linear, which is
confirmed by Figure 10. In this figure, m is the slope of
Figure 10. Viscoplastic strain (VP-strain) development as a
function of time in log–log axes. Recycled carbon fiber (rCF)/
(recycled polypropylene [rPP] þ maleic anhydride grafted
polypropylene [MAPP]) composite at 37 and 40 MPa. Solid line
represents fitting using m ¼ 0.37 (t* ¼ 3600 s).
the linear dependence and A is the vertical shift of
the line.
Since m has to be stress independent, the value of m ¼
0.37 was found as the best fit data in Figure 10. The
exponent m is constant for all stress levels. The value

Downloaded from jcm.sagepub.com at Lulea Univ. of Technology on October 25, 2011

XML Template (2011) [19.9.2011–5:04pm]
K:/JCM/JCM 423858.3d (JCM)
Szpieg et al.
Figure 11. Stress dependence of viscoplastic strain (VP-strain)
presented in terms of A in order to determine parameters CVP
and M.
Figure 12. Predicted evolution of viscoplastic strains (VP-
strains) with time in creep tests.
of A obtained by fitting is different for each of the stress
levels. The solid lines in Figure 10 represent the result of
the ‘visual fitting’. Additional data points for A were
obtained in a similar way from one step of the VP
creep tests (40 min) at different stress levels using the
established time dependence m ¼ 0.37, the experimental
value of eVP and Equation (11). The obtained A depen-
dence on stress level is shown in Figure 11. The scatter is
large but still, the trend of nonlinear increase is clear.
The fit is shown in Figure 11 as a solid line. From the
fitting equation in Figure 11, the parameters in the VP-
law (Equation (13)) can be determined as:
m ¼ 0:37, CVP ¼ 1:95  1015 , M ¼ 23:81
It has to be noted that these values are sensitive with
respect to the VP-strain data at low stress levels (see s ¼
20 MPa in Figure 11), as the accuracy of strain measure-
ments is approximately 0.005%. Consequently, if one
was to use VP-strain data from loading at low stress
levels, that is, approximately 20 MPa in this particular
Downloaded from jcm.sagepub.com at Lulea Univ. of Technology on October 25, 2011
[1–14]
[PREPRINTER stage]
11
Figure 13. Relationship between transverse and axial visco-
plastic strains (VP-strains) in creep tests showing linearity
(Specimen 4 at 37 MPa and Specimen 5 at 40 MPa).
case, a different and less accurate power law would
result.
Use of the determined parameters (see Equation (20))
to predict VP-strain development in creep test is dem-
onstrated in Figure 12. These parameters are used in
Equation (10) to predict VP-strains at 37 and 40 MPa.
One can see that in this stress region VP-strains are sig-
nificantly affected by a small increase in stress level.
As shown in Figure 6, complementary to the VP-
strain development in the loading (axial) direction, neg-
ative VP-strains increase in the transverse direction. For
practical reasons, the transverse strains shown in Figure
13 are presented as positive, where each point represents
a certain instant of time during a creep test at a fixed
stress. Two specimens are presented: Specimen 4 is sub-
jected to creep at 37 MPa and Specimen 5 to creep at
40 MPa. For each of the specimens, both VP-strain
components are proportional at all times, which
allows us to introduce a ‘viscoplastic Poisson’s ratio’.
According to Figure 13, the ratio is practically constant
at the two stress levels. Due to limited data, it is not
possible to investigate if the small difference is due to
variations between the specimens or due to the effect of
stress level. Linear fit to all the data and the correspond-
ing equation is also shown in Figure 13.
The value of the viscoplastic Poisson’s ratio is equal
to 0.275. If the observed proportionality holds not only
for a creep test but also for more general loading ramps,
the introduction of constant VP-Poisson’s ratio could be
very useful for implementation in numerical codes.
ð20Þ
Viscoelasticity
In order to obtain the pure viscoelastic response, VP-
strains were subtracted from the creep data using
Equation (9). Viscoelastic compliances were obtained
by dividing the creep strain data by the stress level.
The results are presented in Figure 14. A very large
XML Template (2011) [19.9.2011–5:04pm]
K:/JCM/JCM 423858.3d (JCM)
12
Figure 14. Creep compliance at different stress levels.
Table 3. Linear viscoelastic parameters
 (s) 10 200 700
C (% / MPa) 0.000774 0.001126 0.000809
Figure 15. Experimental and predicted constant stress rate
loading curves at a stress rate of 1.0 MPa/min. Predictions based
on the average E ¼ 8.68 GPa.
scatter at each stress level is discovered, which is typical
for PP composites.33 However, no systematic increase
with increasing stress levels can be observed. Therefore,
it is concluded that in the considered stress region, the
composite can be considered as linear viscoelastic
assuming g1 ¼ g2 ¼ as ¼ 1 and e0 proportional to stress.
Creep and strain recovery data were used to deter-
mine Ci by simultaneously fitting the creep and strain
recovery data using LSQ method. The same set of  i was
used for all specimens. This procedure was performed
separately for each specimen and then the Ci values were
averaged to obtain an average linear viscoelastic
response.
The  and obtained average C values are presented in
Table 3. This completes the model identification given
by Equations (2)–(4).
Downloaded from jcm.sagepub.com at Lulea Univ. of Technology on October 25, 2011
[1–14]
[PREPRINTER stage]
Journal of Composite Materials 0(0)
Figure 16. Decomposition of the simulated strain in elastic,
linear viscoelastic and total strain (viscoelastic þ viscoplastic)
components.
15,000
0.006947
Figure 17. Experimental loading curves in constant stress rate
test for Specimens 8 and 9 and simulations adjusting the elastic
modulus to each individual specimen. The E-modulus value for
each specimen used in simulations is shown in the legend.
Validation of the material model
The predicted stress–strain curve is presented in Figure
15 together with experimental curves from the constant
stress rate tests for two specimens. The agreement is
good even close to the rupture stress. Figure 16 shows
the significance of different components in the material
model. At 40 MPa, approximately 89% of the strain is
linear viscoelastic, of which 71% is elastic strain includ-
ing stiffness reduction effect. The remaining 11% of the
strain is due to viscoplasticity. At the end of the test,
about 58% of the strain is linear viscoelastic, of which
44% is elastic strain including stiffness-reduction effect
and the rest (42%) is viscoplasticity. For the predictions
in Figure 15, the average properties were used. Due to
the large scatter in data, these predictions are not very
accurate, as the validation was performed using two
specimens only (see Figure 15). A different simulation
was performed (see Figure 17), using individual elastic
modulus of each specimen and average viscoelastic and
XML Template (2011) [19.9.2011–5:04pm]
K:/JCM/JCM 423858.3d (JCM)
Szpieg et al.
viscoplastic behavior. The simulation is in very good
agreement with test results indicating that the large var-
iability of results is more related to elastic modulus than
to viscoelasticity and viscoplasticity. It is clear that at
stress levels close to rupture, the inelastic model does
not provide an accurate prediction.
Conclusions
This article presents a combined experimental and the-
oretical study of a new, fully recycled short fiber, carbon
fiber thermoplastic composite material. The effect of
MAPP coupling agents on properties of a composite
made of rCFs and rPP (rCF/[rPP þ MAPP]) was stud-
ied experimentally. It was shown that adding 10 wt% of
MAPP to the rPP matrix leads to an almost 100%
increase of the composite strength. This was attributed
to improvement of the fiber/matrix interface. The elastic
modulus remains unchanged. The inelastic stress–strain
behavior of the composite was demonstrated to be
related to elastic modulus reduction at high stresses,
viscoelasticity, and viscoplasticity. It was demonstrated
that the viscoplastic response can be described by
Zapas’ model, which in creep predicts a power law
dependence of the VP-strains with respect to time and
stress level. Parameters in the model were determined by
creep and strain recovery tests. In addition, the material
was found to exhibit viscoplastic Poisson’s ratio of
0.275. This finding allows introduction a constant VP-
Poisson’s ratio, which may be very useful in implemen-
tation in numerical codes.
The viscoelasticity of the rCF/(rPP þ MAPP) com-
posite is linear in the whole considered stress region. The
parameters in the viscoelastic model were also
identified.
As a result of these analyses, a complete inelastic
material law was developed for the rCF/(rPP þ
MAPP) composite. The developed material model was
validated by simulation of the strain response in con-
stant stress rate tests. Simulation results employing the
inelastic material model are in general in a good agree-
ment with the experimental data.
Acknowledgment
Thanks are due to Prof. Janis Varna for his helpful suggestions
during the course of this work.
References
1. Suzuki T and Takahashi J. Prediction of energy intensity
of carbon fiber reinforced plastics for mass-produced
passenger cars. The Ninth Japan International SAMPE
symposium, November 29–December 2; Tokyo, Japan,
2005: 14–19.
2. Pimenta S and Pinho ST. Recycling carbon fibre rein-
forced polymers for structural applications: Technology
Downloaded from jcm.sagepub.com at Lulea Univ. of Technology on October 25, 2011
[1–14]
[PREPRINTER stage]
13
review and market outlook. Waste Management 2011;
31(2): 378–392.
3. Pimenta S, Pinho ST and Robinson GS. Experimental
study on the mechanical performance of recycled CFRP.
ICCM 17, July 27–31; Edinburgh; 2009.
4. Yip HLH, Pickering SJ and Rudd CD. Characterisation of
carbon fibres recycled from scrap composites using fluidized
bed process. Plast Rubber Compos 2002; 31(6): 278–282.
5. Pickering SJ. Recycling technologies for thermoset com-
posite materials-current status. Compos Part A 2006; 37:
1206–1215.
6. Bassam JJ, Pomykala JA Jr, Daniels EJ and Greminger
JL. A process to recover carbon fibres from polymer-
matrix composites in end-of-life vehicles. JOM, J Miner
Met Mater Soc 2007; 56: 43–47.
7. Meyer LO, Schulte K and Grove-Nielsen E. CRFP-recy-
cling following a pyrolysis route: Process optimization
and potentials. J Compos Mater 2009; 43: 1121–1132.
8. Åström BT. Manufacturing of polymer composites.
London, UK, Chapman & Hall.
9. Wong KH, Pickering SJ and Brooks R. Recycled carbon
fibre reinforced polypropylene composites: Effect of cou-
pling agents on mechanical properties. Composites
Innovation 2007 - Improved Sustainability and Environmental
Performance; Organised by NetComposites; 2007 October
4–5; Barcelona, Spain; 2007.
10. Turner TA, Pickering SJ and Warrior NA. Development
of high value composite materials using recycled carbon
fibre, SAMPE’09 Conference, May 18–21, Baltimore,
MD: SAMPE, 2009.
11. Pickering SJ, Turner TA and Warrior NA. Moulding
compound development using recycled carbon fibres.
SAMPE Fall Technical Conference, November 6–9,
Dallas, TX: SAMPE; 2006.
12. Wong KH, Pickering SJ, Turner TA and Warrior NA.
Compression moulding of a recycled carbon fibre
reinforced epoxy composite. SAMPE Conference; May
18–21; Baltimore, MD; 2009.
13. Janney MA, Newell WL, Geiger E, Baitcher N and
Gunder T. Manufacturing complex geometry composites
with recycled carbon fiber. SAMPE’09 Conference, May
18–21, Baltimore, MD: SAMPE; 2009.
14. MIT. ‘Materials Innovation Technologies, LLC’.
www.nemergingmit.com (2010, accessed July 2010).
15. Jiang G, Wong KH, Pickering SJ, Walker GS and Rudd
CD. Alignment of recycled carbon fibre and its application
as a reinforcement. SAMPE Fall Technical Conference and
Exhibition; November 6–9; Dallas, TX; 2006.
16. Janney M and Baitcher N. Fabrication of chopped fiber
preforms by the 3-DEP process. Composites and polycon
2007. Tampa, FL: American Composites Manufactureres
Association, 2007.
17. Thompson MR, Tzoganakis C and Rempel GL. Terminal
functionalisation of polypropylene via the alderene reac-
tion. Polymer 1998; 39: 327–334.
18. Lakes RS. Viscoelastic measurement techniques. Rev
Scient Inst 2004; 74(4): 797–810.
19. Lou YC and Schapery RA. Viscoelastic characterization
of a nonlinear fiber-reinforced plastic. J Compos Mater
1978; 5: 208–234.
XML Template (2011) [19.9.2011–5:04pm]
K:/JCM/JCM 423858.3d (JCM)
14
20. Zapas LJ and Crissman JM. Creep and recovery behavior
of ultra-high molecular weight polyethylene in the region
of small uniaxial deformations. Polymer 1984; 25(1):
57–62.
21. Wong KH, Jiang G, Pickering SJ, Rudd CD and Walker
GS. Characterisation of recycled carbon fibre:
Mechanical properties and surface chemistry. School of
Mechanical, Materials and Manufacturing Engineering.
38th International SAMPE Technical Conference;
November 6-9; Dallas, TX; 2006.
22. Giannadakis K, Szpieg M and Varna J. Mechanical per-
formance of recycled carbon fibre/PP. Exp Mech 2010;
published online.
23. Nordin LO and Varna J. Nonlinear viscoplastic and non-
linear viscoplastic material model for paper fiber compos-
ites in compression. Compos Part A 2006; 37: 344–355.
24. Marklund E, Varna J and Wallström L. Nonlinear visco-
elasticity and viscoplasticity of flax/polypropylene com-
posites. J Eng Mater Technol 2006; 128(4): 527–536.
25. Marklund E, Eitzenberger J and Varna J. Nonlinear vis-
coelastic viscoplastic material model including stiffness
degradation for hemp/lignin composites. Compos Sci
Technol 2008; 68: 2156–2162.
Downloaded from jcm.sagepub.com at Lulea Univ. of Technology on October 25, 2011
View publication stats
[1–14]
[PREPRINTER stage]
Journal of Composite Materials 0(0)
26. Schapery RA. Nonlinear viscoelastic and viscoplastic con-
stitutive equations based on thermodynamics. Mech
Time-Depend Mat 1997; 1: 209–240.
27. Data sheet CFM-PYR-H-1, Hadeg Recycling GmbH,
Germany, 2007.
28. Szpieg M, Wysocki M and Asp LE. Reuse of polymer
materials and carbon fibres in novel engineering compos-
ite materials. Plast Rubber Compos Macromol Eng 2009;
38: 419–425.
29. Szpieg M, Wysocki M and Asp LE. Recycled PP aimed as
a composites precursor material. Plast Rubber Compos
Macromol Eng 2009; 38: 412–418.
30. Szpieg M, Giannadakis K and Varna J. Time dependent
nonlinear behaviour of recycled polypropylene (rPP) in
high tensile stress loading. J Thermoplast Compos Mater
2011; 24: 625–652.
31. Megnis M, Varna J, Allen DH and Holmberg A.
Micromechanical modeling of visco-elastic response
of GMT composite. J Compos Mater 2001; 35(10):
849–882.
32. Nordin LO and Varna J. Nonlinear viscoelastic behavior
of paper fiber composites. Compos Sci Technol 2005; 65:
1609–1625.