Вы находитесь на странице: 1из 7

Original Article

Journal of Intelligent Material Systems


and Structures
1–7
Smart sensing layer for the detection Ó The Author(s) 2014
Reprints and permissions:
of damage due to defects in a sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1045389X14554138

laminated composite structure jim.sagepub.com

Muhammad A Nasir1, Hina Akram1, Zaffar M Khan2, Masood Shah1, Saad


Anas3, Zeeshan Asfar3 and Saad Nauman4,5

Abstract
A smart sensing layer based on polystyrene and carbon nanoparticles has been developed. It has been deposited on the
composite specimens for real-time, in situ monitoring of structural health. The strain response of the smart sensing layer
has been recorded for composite laminates using different defect configurations (notch spacing). Numerical simulations
of the stress–strain concentration have been carried out in order to determine the state of strain at the smart sensing
layer, in the presence of different notch configurations. It has been observed that the sensing layer detects well the pres-
ence of large deformations and damage due to defects in the structure, with clearly defined peaks at the points of struc-
tural damage.

Keywords
Sensor, structural health monitoring, composites, smart layer

Introduction system would be to integrate sensors in the structure at


critical positions much like human nervous system
Composite materials have found wide applications as (Chang et al., 2011). Traditional monitoring techniques
structural materials in various applications ranging involve acoustic emission monitoring (Finlayson et al.,
from fairly mundane such as household products and 2001), ultrasonic sensors (Chang, 1998; Lynch and Loh,
as construction material to the high end such as com- 2006), and vibration monitoring techniques (Doebling
plex surgical implants, high-performance sports equip- et al., 1996; Zou et al., 2000). These techniques are diffi-
ment, and components in aerospace vehicles and in cult to implement in in situ manner and costly and cum-
airplanes. The wide spectrum of applications manifests bersome to be integrated inside a sophisticated part
the inherent advantage of composites to tailor the prop-
especially when weight savings are a critical feature of
erties to the intended application area according to the
modern structures especially in aerospace and automo-
performance, weight, and cost constraints. This advan-
tive sectors.
tage has a downside as well, that is, the lack of consis-
A solution to this difficulty is readily offered by the
tent material models and performance criteria that can
smart composite materials that can be described as
be used for most if not all the composites. This is due
to the heterogeneous nature of composite materials,
strong dependence of properties on the manufacturing 1
Department of Mechanical Engineering, University of Engineering and
process, and post-processing/curing conditions and Technology, Taxila, Taxila, Pakistan
2
relative novelty of composites as structural materials in Department of Aeronautics and Astronautics, Institute of Space
high-performance applications. Technology, Islamabad, Pakistan
3
School of Mechanical & Manufacturing Engineering, National University
Owing to the above-mentioned constraints, it is of of Sciences and Technology, Islamabad, Pakistan
utmost importance in composite material parts to be 4
Department of Materials Science and Engineering, Institute of Space
able to monitor their performance in situ and in real Technology, Islamabad, Pakistan
5
time. This paradigm of monitoring the state of a struc- University of Lille Nord of France, ENSAIT, GEMTEX, Roubaix, France
tural integrity, onset of flaws/damage, and strain his-
Corresponding author:
tory mapping is often referred to as ‘‘structural health Saad Nauman, University of Lille Nord of France, ENSAIT, GEMTEX,
monitoring’’ (SHM) (Balageas et al., 2006; Sohn et al., Roubaix 59100, France.
2004). An integral part of any online in situ monitoring Email: saadnauman@hotmail.com

Downloaded from jim.sagepub.com at FRESNO PACIFIC UNIV on January 2, 2015


2 Journal of Intelligent Material Systems and Structures

materials with integrated self-sensing property. There unable to reach a value equal to 0 by the time it reaches
are several approaches of achieving that; one way is to the other end of the barrier, then there is a finite prob-
utilize self-sensing in carbon composite materials where ability of an electron finding itself on the other side of
conductive carbon fibers within pockets of insulating the barrier. When this happens, the electron is said to
resin act as percolation networks, allowing electrical have tunneled through the non-conductive barrier. This
current to flow through them (Abry et al., 1999, 2001). phenomenon is known as the tunnel effect.
In case of strains and/or onset of damage, these perco- In order to invoke tunnel effect, composites have to
lation networks are broken resulting in an increase in be fabricated having the volume concentration of nano-
resistance of the percolation path between the electro- fillers lying in the critical range known as ‘‘percolation
des. This change in electrical resistance can be regis- threshold.’’ At volume concentrations of nanofillers
tered for SHM. One obvious disadvantage is that this greater than the percolation threshold, the composite
type of self-sensing approach cannot be used in the case becomes entirely conductive and is redundant for use in
of non-conducting fibrous systems such as glass and sensing applications where high sensitivity is desired.
aramid fibers. Another disadvantage is difficulty in At concentrations that are lower than the percolation
localizing damage as any defect will change global threshold concentrations, the composite is completely
resistance of the structure and no localized response insulating, as it does not allow electrons to tunnel from
can be expected. Moreover, in multilayered composite one site to another due to their mutual distance.
structures such as laminates or three-dimensional (3D) Whereas for volume concentrations of conductive fillers
woven composites, any damage having layered config- lying within the percolation threshold, a sudden and
uration at its origin such as interlaminar shear cannot rapid drop in resistivity is observed. This is because of
be explicitly detected. the formation of conductive paths that allow electron
Another approach of engineering smart composite hopping or tunneling from one filler site to another
materials is to incorporate conductive fillers inside the known as percolation networks.
composites during manufacturing. These fillers can be The present research work has been carried out with
in the form of micro- or nanoparticles, fibers, or nano- the aim of optimizing tunnel effect sensors for defect
tubes (Johnson et al., 2012; Li et al., 2008). These fillers detection property in laminated composite structures.
can be either incorporated inside the matrix by disper- For this purpose, the response of the sensor has been
sing them using standard dispersion methods (Li et al., recorded for smooth specimens and in the presence of a
2014; Thostenson and Chou, 2006) or can be used to notch. A low-cost coating material made from thermo-
make a smart filament (Nauman et al., 2011a, 2011b, plastic polymer dissolved in a suitable solvent and dis-
2012; Risicato et al., 2014; Sebastian et al., 2014) or persed with carbon nanoparticles (CNPs) has been
reinforcement (Hecht et al., 2007) which can be then used to fabricate sensors. These sensors are in fact a
impregnated with resin using traditional resin impreg- smart layer which can be deposited on a composite
nation techniques. Both these approaches rely on the structure at any area of interest. The main advantages
monitoring of electrical resistance of percolation net- of the proposed sensing layer for SHM are its low cost
works which are formed between the conductive fillers. and simple manufacturing. These smart sensing layers
As expected, these percolation networks are altered can be easily deposited at any area of interest. Another
and broken upon application of deforming forces and advantage is fairly high sensitivity of the sensing layer
due to the onset of damage. Therefore, electrical resis- to applied strains. The goal in this article is to deter-
tance mapping of these percolation networks can serve mine whether the sensor reacts to the stress–strain con-
as an effective method of in situ SHM of composite centration due to the notch or gives an averaged strain
structures. response for the whole sensor. The stress–strain con-
The principle of charge transfer through these perco- centration due to the notch is determined through
lation networks is based on the ‘‘tunnel effect’’ which numerical simulation.
states that an electron when faced with a potential bar-
rier whose potential energy is greater than its kinetic
energy can still ‘‘hop’’ across the potential barrier with- Materials and methods
out the input of energy (Gurney and Condon, 1929).
This is because of the dual nature of sub-atomic parti-
Smart sensing layer
cles predicted by quantum theory of mechanics. Polystyrene (PS) beads were weighed in a beaker on a
According to Heisenberg’s uncertainty principle, an laboratory-grade analytical balance having precision of
electron’s probability of moving from one side of a 0.02 mg. Prior to that, the PS was dried in a vacuum
physical barrier to another one is greater than 0. When oven at 70°C to remove any moisture. Benzene was
an electron wave encounters a non-conductive polymer added in the beaker (5 mL/g Benzene, of total weight of
film, which in a composite acts as a potential barrier, PS and CNPs). Beaker with PS and benzene was placed
instead of instantly going to 0, the wave starts to decay on a hot plate and heated at 50°C along with continu-
exponentially within the potential barrier. If the wave is ous stirring for 7 h. CNPs (Printex L6 supplied by

Downloaded from jim.sagepub.com at FRESNO PACIFIC UNIV on January 2, 2015


Nasir et al. 3

Figure 2. Specimen dimensions with configuration of smooth


and notched specimens with smart sensing layer in the center.
Extensometer attached with specimen for tensile test as well.

Figure 1. GFRP-laminated composite specimens with smart the stress concentration. Specimen details are shown in
sensing layer in the center. Figure 2.
GFRP: glass fiber–reinforced polymer. The strain was also measured using an extensometer
inside the gauge length to compare with the PS/CNP
coating response. All the tensile tests were carried out
Degussa) were weighed (35% of PS weight) and added at a rate of 1 3 1023 s21. The smart sensing layer in
to the same beaker. This weight percentage is based on the smooth specimen is expected to experience uniform
previous research on similar particles and ensures opti- deformation throughout the gauge section. However,
mal quantum tunneling (Nauman et al., 2011a, 2011b, for the notched specimens, a strain concentration is
2012). It was then followed by 24 h of continuous stir- deemed to be established at the smart sensing layer. It
ring at nominal speed. This ensured complete dispersion will be shown, later on, that the smart coating on 7-mm
of CNPs in PS. All the steps were performed at ambient notched specimen can sense only one strain peak while
conditions of moisture and pressure. the same coating on 12-mm notched spacing will see
two strain peaks along the axial length of the specimen.
Composites’ fabrication
For the purpose of making the composite specimens, Data acquisition
eight plies of plain woven glass fabric (Interglass 92110) Electrical connections were made at the end of sensing
were used as the reinforcement. The epoxy resin used layer on the composite specimen surface by connecting
was Araldite LY5052 in conjunction with the hardener fine copper wire using silver-loaded conducting paint
Aradur 5052. This is a low-viscosity resin transfer supplied by RS Components Ltd to reduce contact
molding (RTM)-grade resin system supplied by the resistance. The length of the wires was kept same for all
resin manufacturer HuntsmanÒ. Vacuum-assisted resin the specimens to reduce connection resistivity error.
transfer molding (VARTM) was used for resin infu- Data acquisition module Keithley KUSB-3100 was
sion. The post-cured multi-ply composite sheets were used in conjunction with the Wheatstone bridge, for
cut to form 10 specimens of 300 mm 3 25 mm each as data acquisition from the smart sensing layer. The sen-
shown in Figure 1. All these specimens were coated sing layer served as the unknown resistance in the
with the PS/CNP sensing layer in the center (Figures 1 Wheatstone bridge. The data acquisition module
and 2). Keithley KUSB 3100 allows data conversion from ana-
log to digital format and data logging directly onto the
computer where it can be manipulated and processed
Specimens’ configuration for further information extraction and graphical dis-
The 10 specimens were divided into three sets of speci- play. The electrical resistance measuring method is
mens with different notch configurations. The first set essentially the two-probe method. An instrumentation
consists of smooth specimens with a layer of PS/CNP amplifier along with low-pass filters was employed for
coating. The second set had two holes of 1.5-mm dia- data linearization, amplification, and noise suppression.
meter drilled along a horizontal axis, through the speci-
men at 7-mm separation. These acted as stress raisers
by directly affecting the PS/CNP sensing layer strain. Numerical simulation
The third set had the same configuration as above with The loading conditions and the resultant stresses and
12-mm separation between the holes, in order to reduce strains were first determined using numerical simulation

Downloaded from jim.sagepub.com at FRESNO PACIFIC UNIV on January 2, 2015


4 Journal of Intelligent Material Systems and Structures

Table 1. Elastic constants of the specimen.

E1 E2 E3 n12

18 GPa 18 GPa 4 GPa 0.24

Figure 3. Meshing of composite specimen in the presence of


7-mm apart holes.

in the ABAQUS/Standardä software. The specimen


was modeled as a two-dimensional (2D) composite shell
of 2-mm thickness with dimensions shown in Figure 2.
The partitions have been created in such a fashion that
the area representing the smart sensing layer can be
meshed separately. The simulation is carried out with-
out taking into account the effect of the smart sensing
layer. Considering the mechanical properties and thick-
ness of PS, the coating would have a maximum of 0.5%
contribution to the stresses achieved in the specimens.
This is within the dispersion of the data received by the
sensor and thus is not considered for the simulation.
The meshing was performed in such a way that the
area of interest (smart sensing layer) had a fine mesh
(Figure 3). The mesh area was made up of 6656 A
four-node bilinear plane stress quadrilateral, reduced
integration, hourglass control (SPS4R) elements. The Figure 4. (a) Contour path to calculate peak strain on edge of
sensing layer zone had 1200 elements with an average smart sensing layer in 7-mm notch spacing specimen and (b)
size of 0.1 mm. variation in strain along contour.
The values of the elastic constants are given in
Table 1.
The analysis was carried out for elastic conditions The strain distribution along an axial line is plotted
only. The tensile test was simulated in 100 steps up to for the 7-mm notch spacing specimen in Figure 4(a)
the fracture strain (determined by experimentation). and (b) and that with 12-mm notch spacing is given in
Strain data were obtained for a path passing through Figure 5(a) and (b).
the outermost periphery of the ‘‘smart sensing layer’’ The contour plots in Figure 4(a) and 5(a) show the
mesh. The strain distribution along this path gives an strain concentration produced as a result of the holes.
indication of the strain concentration on the smart sen- The thick red line shows the left edge of the sensor.
sing layer. (It will be discussed later with reference to This edge is the part that undergoes the highest value
Figures 4 and 5 in the later section.) The results of the of strain. In the 7-mm notch spacing (Figure 4(a)), the
smooth specimen show no concentration and thus have edge of the sensor experiences a single peak in strain
not been presented here. concentration. The same contour is shown for 12-mm
notch spacing in Figure 5(a). In this case, the left edge
of the sensor experiences two peaks in strain
Experimental results and discussion concentrations.
The strain response of the smart sensing layer in the It can be deduced from Figures 4(b) and 5(b) that
smooth specimen and in the presence of notches has for a specific value of applied stress, peak of strain var-
some unique features. To explain them better, we will iation in the smart sensing layer zone is almost 50%
first take a look at the results of the numerical simula- higher in the 7-mm notch spacing specimen, while it is
tion to be able to understand the strain response. only 10% higher for the one with 12-mm notch spacing.

Downloaded from jim.sagepub.com at FRESNO PACIFIC UNIV on January 2, 2015


Nasir et al. 5

Figure 6. Tensile stress–strain diagram compared with dR/R–


strain diagram for an un-notched specimen.

slope of the stress–strain curve is seen at a nominal


strain of 0.005 ’ 1/3eu. Before this knee, the resistivity
variation is nonlinear with respect to the applied strain.
With the onset of damage at e = 0.005 (at the edges of
the notches), a sharp increase in resistivity is seen.
After this point, the resistivity response is linear with
Figure 5. (a) Contour path to calculate peak strain on edge of the applied strain. Another jump in resistivity can be
smart sensing layer in 12-mm notch spacing specimen and (b) seen, probably due to damage initiation at the other
variation in strain along contour. notch; however, the stress–strain diagram does not
detect this. After the second jump, the sensor responds
very accurately to the damaged composite structure,
Considering the normalized resistance response of the following the stress–strain curve accurately right up to
smart sensing layer, the absolute values of dR/R cannot the final fracture.
be regarded as an absolute measure of strain. However, In order to further study the effect of damage, on
from a qualitative stand point, the smart sensing layer the sensor response, the notches were made closer to
seems to be able to detect damage in the material in an the sensor with 7-mm separation between them. The
efficient manner. test results are presented in stress–strain-normalized
Figure 6 depicts the stress–strain-normalized resis- resistance plot for the specimen in Figure 8. Notch
tance plot for smooth (un-notched) specimen with spacing of 7 mm represents the presence of notches
smart sensing layer. It is evident that the su is around 1 mm beyond the sensor boundary. As can be seen
460 MPa, while nominal strain at fracture ‘‘eu’’ value from Figure 4(b), the strain concentration increases
stands at 0.0147. The smart sensing layer seems to fol- with the applied strain by 50%. The tensile tests on the
low the composite deformation in a nonlinear fashion specimens show a similar trend as for the 12-mm notch
for almost 50% of the loading range. However, at the spacing. Initially, the smart sensing layer has a non-
start of damage, as shown by the change in slope of the linear response to strain. At the first instance of damage
stress–strain curve, the resistivity rises quickly and the at e = 0.005, the characteristic jump in resistivity can
sensor seems to follow the deformation in a more linear be seen. Then, the zone of linear response can be seen.
fashion. This trend goes right up to the final rupture. After that, as we know the strain in this case is 50%
Figure 7 shows the stress–strain-normalized resis- higher at the sensor edge, one can see the effect of this
tance plot for the specimen with 12-mm notch spacing. in the response of the smart sensing layer in the form of
We know that the nominal strain at fracture eu for the a much higher slope as compared to the specimen with
composite material is around 0.0147. Due to the 12-mm notch spacing. The slopes can be compared
notches, the strain concentration would be three times because we have fixed the starting and end points of the
as that of the smooth specimen. Thus, any damage stress–strain and resistivity–strain curves. In the zone of
around the notch should start at one-third of the higher damage progression, the correlation between the
applied strain on the smooth specimen. This is indeed sensor and stress–strain curve is very good up to frac-
the case as seen in Figure 7. A sharp change in the ture of the smart sensing layer.

Downloaded from jim.sagepub.com at FRESNO PACIFIC UNIV on January 2, 2015


6 Journal of Intelligent Material Systems and Structures

Figure 7. Tensile stress–strain diagram compared with dR/R–


Figure 8. Tensile stress–strain diagram compared with dR/R–
strain diagram for a specimen with 12-mm notch spacing: (a)
strain diagram for the specimen with 7-mm notch spacing: (a)
spike in sensor response undetected by the stress–strain curve,
nonlinear response for initial loading with no damage, (b) mostly
(b) nonlinear response for initial loading with no damage, (c)
linear response to partially damaged specimen, (c) damage
linear response to partially damaged specimen, (d) knee in
progression in zone of high strain concentration, (d) good
stress–strain curve indicating damage initiation and
correlation of sensor response in zone of damage progression,
corresponding spike in sensor response, (e) good correlation of
(e) knee in stress–strain curve indicating damage initiation and
sensor response in zone of damage progression, and (f) sensor
corresponding spike in sensor response, and (f) sensor fracture.
fracture.

Conclusion Declaration of conflicting interests


The strain response of a smart sensing layer, developed The authors declared no potential conflicts of interest with
for SHM, based on PS/CNPs has been observed in the respect to the research, authorship, and/or publication of this
article.
presence of defects. Numerical simulation was carried
out in order to estimate the effect of strain concentra-
tion on the smart sensing layer. It was found out that Funding
the sensing layer reacts well to the presence of defects This research received no specific grant from any funding
or stress raisers in the structure. The principal response agency in the public, commercial, or not-for-profit sectors.
characteristics are listed below.
The tensile test of a smooth specimen is character- References
ized by the initial nonlinear response of sensing layer,
followed by a linear response. It is believed that the lin- Abry JC, Bochard S, Chateauminois A, et al. (1999) In situ
detection of damage in CFRP laminates by electrical resis-
ear response follows a certain amount of damage in the
tance measurements. Composites Science and Technology
matrix of the composite. Even if the damage is pro-
59(6): 925–935.
duced at a distance from the sensor, the resistivity Abry JC, Choi YK, Chateauminois A, et al. (2001) In-situ
response shows it by a jump or peak in the curve. If, monitoring of damage in CFRP laminates by means of
however, this damage occurs closer to the sensor, the AC and DC measurements. Composites Science and Tech-
peak in the smart sensing layer response is amplified. nology 61(6): 855–864.
The sensing layer shows a linear response to the applied Balageas D, Fritzen C-P and Guemes A (2006) Structural
strain in the specimen after a certain amount of damage Health Monitoring. Wiley: USA.
has occurred. We believe that this damage is mostly Chang F-K (1998) Structural Health Monitoring: Current Sta-
limited to the matrix cracking. In the linear region of tus and Perspectives. CRC Press: USA.
Chang F-K, Markmiller JFC, Yang J, et al. (2011) Structural
the sensor response, the load is principally borne by the
health monitoring. In: SB Johnson, T Gormley, S Kessler, et
fibers. When the damage progresses, that is, the fibers
al. (eds) System Health Management: With Aerospace Appli-
break, the sensor records this as well, perfectly follow- cations. Chichester: John Wiley & Sons, Ltd, pp. 419–428.
ing the stress–strain curve right up to the complete frac- Doebling SW, Farrar CR, Prime MB, et al. (1996) Damage
ture of the specimen. It would be interesting to Identification and Health Monitoring of Structural and
normalize the absolute value of the change in resistivity Mechanical Systems from Changes in their Vibration Char-
so that this sensor could be used as a strain gauge as acteristics: A Literature Review. Los Alamos, NM: Los
well as a damage detection tool. Alamos National Laboratory.

Downloaded from jim.sagepub.com at FRESNO PACIFIC UNIV on January 2, 2015


Nasir et al. 7

Finlayson RD, Friesel M, Carlos M, et al. (2001) Health mon- compression and traction detection in glass laminate com-
itoring of aerospace structures with acoustic emission and posites. Sensors 11(10): 9478–9498.
acousto-ultrasonics. Insight 43(3): 155–158. Nauman S, Cristian I and Koncar V (2012) Intelligent carbon
Gurney RW and Condon EU (1929) Quantum mechanics and fibre composite based on 3D-interlock woven reinforce-
radioactive disintegration. Physical Review 33(2): 127. ment. Textile Research Journal 82(9): 931–944.
Hecht DS, Hu L and Gruner G (2007) Electronic properties Nauman S, Lapeyronnie P, Cristian I, et al. (2011b) Online
of carbon nanotube/fabric composites. Current Applied measurement of structural deformations in composites.
Physics 7(1): 60–63. IEEE Sensors Journal 11(6): 1329–1336.
Johnson TM, Fullwood DT and Hansen G (2012) Strain Risicato J-V, Kelly F, Soulat D, et al. (2014) A complex
monitoring of carbon fiber composite via embedded nickel shaped reinforced thermoplastic composite part made of
nano-particles. Composites Part B: Engineering 43(3): commingled yarns with integrated sensor. Applied Compo-
1155–1163. site Materials 1–18.
Li C, Thostenson ET and Chou T-W (2008) Sensors and Sebastian J, Schehl N, Bouchard M, et al. (2014) Health mon-
actuators based on carbon nanotubes and their compo- itoring of structural composites with embedded carbon
sites: a review. Composites Science and Technology 68(6): nanotube coated glass fiber sensors. Carbon 66: 191–200.
1227–1249. Sohn H, Farrar CR, Hemez FM, et al. (2004) A Review of
Li W, He D, Dang Z, et al. (2014) In situ damage sensing in Structural Health Monitoring Literature: 1996–2001. Los
the glass fabric reinforced epoxy composites containing Alamos, NM: Los Alamos National Laboratory.
CNT-Al2O3 hybrids. Composites Science and Technology Thostenson ET and Chou TW (2006) Carbon nanotube net-
99(0): 8–14. works: sensing of distributed strain and damage for life
Lynch JP and Loh KJ (2006) A summary review of wireless prediction and self healing. Advanced Materials 18(21):
sensors and sensor networks for structural health monitor- 2837–2841.
ing. Shock and Vibration Digest 38(2): 91–130. Zou Y, Tong L and Steven GP (2000) Vibration-based
Nauman S, Cristian I and Koncar V (2011a) Simultaneous model-dependent damage (delamination) identification
application of fibrous piezoresistive sensors for and health monitoring for composite structures—a review.
Journal of Sound and vibration 230(2): 357–378.

Downloaded from jim.sagepub.com at FRESNO PACIFIC UNIV on January 2, 2015

Вам также может понравиться