Home Search Collections Journals About Contact us My IOPscience

Flexible piezoelectric energy harvesting from jaw movements

This content has been downloaded from IOPscience. Please scroll down to see the full text.

2014 Smart Mater. Struct. 23 105020

(http://iopscience.iop.org/0964-1726/23/10/105020)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 142.51.1.212
This content was downloaded on 07/10/2014 at 07:39

Please note that terms and conditions apply.

 Smart Materials and Structures

Smart Mater. Struct. 23 (2014) 105020 (8pp) doi:10.1088/0964-1726/23/10/105020

Flexible piezoelectric energy harvesting

from jaw movements
Aidin Delnavaz and Jé ré mie Voix
Mechanical Engineering Department, É cole de technologie supérieure (É TS), Montreal, Canada

E-mail: aidin.delnavaz@ens.etsmtl.ca

Received 1 May 2014, revised 15 June 2014

Accepted for publication 18 June 2014
Published 16 September 2014

Abstract
Piezoelectric fiber composites (PFC) represent an interesting subset of smart materials that can
function as sensor, actuator and energy converter. Despite their excellent potential for energy
harvesting, very few PFC mechanisms have been developed to capture the human body power
and convert it into an electric current to power wearable electronic devices. This paper provides a
proof of concept for a head-mounted device with a PFC chin strap capable of harvesting energy
from jaw movements. An electromechanical model based on the bond graph method is
developed to predict the power output of the energy harvesting system. The optimum resistance
value of the load and the best stretch ratio in the strap are also determined. A prototype was
developed and tested and its performances were compared to the analytical model predictions.
The proposed piezoelectric strap mechanism can be added to all types of head-mounted devices
to power small-scale electronic devices such as hearing aids, electronic hearing protectors and
communication earpieces.
Keywords: flexible energy harvesting, jaw movements, piezoelectric fiber composite, bond
graph
(Some figures may appear in colour only in the online journal)

1. Introduction The conversion of energy generated by human body

movements to electricity can be achieved, among other ways,
Continuous progress in reducing the size and power con- by piezoelectric effect. Piezoelectric materials become elec-
sumption of portable electronic devices and recent technolo- trically polarized and produce an electric charge when subject
gical advances in energy harvesting from the human body has to a mechanical stress. Different piezoelectric materials have
increased the odds of achieving battery-less wearable elec- been discovered in nature or synthesized in laboratory.
tronic devices. Human powered electronics would curb the However, those with mechanical flexibility are best suited for
current dependency on batteries, give more autonomy to energy harvesting from the human body, because they can
mobile technologies, bring more convenience to electronic follow the bodyʼs contours and be easily integrated to wear-
consumers and finally reduce the environmental effects able devices [2].
associated with battery disposal. Polymer-based piezoelectric materials exhibit excellent
Kinetic energy (limb motion), strain energy (muscle mechanical flexibility properties. For example, poly-
force), and thermal energy (body heat) are three main forms vinylidene fluoride (PVDF), the most common polymer in
of energy that can be harvested from the human body. The this category, can accommodate a strain of 2% or even higher.
jaw movement that normally occurs when chewing, eating Therefore, it has been widely investigated for various human
and speaking is the most promising muscle activity from body energy harvesting applications ranging from a PVDF
which energy could be harvested in the region of the head. microbelt capable of harnessing power from human breathing
[3] to a PVDF backpack shoulder strap to harvest energy from
For instance, one can obtain approximately 580 J only from
general human body movements [4]. However, PVDF mainly
daily chewing, which is equivalent to an average power of
suffers from its inherent low electromechanical coupling
approximately 7 mW [1].
factor [5].

0964-1726/14/105020+08$33.00 1 © 2014 IOP Publishing Ltd Printed in the UK

Smart Mater. Struct. 23 (2014) 105020 A Delnavaz and J Voix

Figure 1. Structure of the piezoelectric fiber composite (a) microscopic image, horizontal lines: piezoelectric fibers, vertical lines:
interdigitated electrodes. (b) Exploded view of layers, from core to exterior (symmetrical): PZT ceramic fibers in epoxy, copper interdigitated
electrodes and Kapton film.

Unlike PVDF, the performance of inorganic ceramic- PFC which is also called active fiber composite has high
based piezoelectric materials is superior in electromechanical degree of structural flexibility and is used in this research as a
conversion efficiency. For example, lead zirconate titanate piezoelectric converter. The multilayer structure of PFC is
(PZT), the most well-known material in the ceramic piezo- shown in figure 1. Its core layer comprises of uniaxially
electric category, has a piezoelectric coefficient ten times aligned piezoceramic fibers of 250 μm in diameter. The fibers
higher than PVDF [2]. But, ceramics are generally brittle, are surrounded by a polymer matrix of epoxy resin. This layer
which hinders their possible use for flexible energy is encapsulated by a network of copper electrodes called
harvesting. interdigitated electrodes. The interdigital pattern of the elec-
Therefore, ceramic piezoelectric elements must be inte- trodes allows the electric potential to form along the length of
grated in the form of fibers, ribbons or wires onto a flexible the fiber (which is called direction 3), therefore taking
substrate using a microelectromechanical system. This results advantage of the greater d33 piezoelectric coupling coefficient.
in low profile piezoelectric structures having both flexibility On the outside layer there are two thin Kapton ® sheets that
and a high electromechanical conversion capability. PZT provide good mechanical protection and electrical insulation.
ribbons [6], PZT nanofibers [7] and piezoelectric fine wires A relatively small amount of applications has been
[8] are some of the new materials technologies that are based reported as using PFC in a wearable energy harvesting plat-
form. Fitting piezoelectric fibers into a shoe sole was among
on piezoelectric long elements with nanoscale cross-sections.
the very first attempts to harvest parasitic power from heel
Yet, their output power is in nanowatt range and they appear
striking [14]. Different piezoelectric materials including PFC
to be limited to powering nanosystems.
have been investigated for energy recovery in glove structures
Composites with thicker piezoelectric fibers such as
[5]. This investigation proves the superiority of PFC over
macro fiber composites (MFC) [9–12] and piezoelectric fiber
PVDF for the flexible energy harvesting. The performance of
composites (PFC) [5, 13] have been developed over the past
PFC and PVDF has also been compared for energy harvesting
years and are now commercially available. They are similar
from renewable sources [15]. In addition, an experimental
because both consist of the same three main components: comparison between several composite piezoelectric materials
unidirectional piezoceramic fibers, interdigitated electrodes for power generation has been reported [16]. More recently,
and an adhesive polymer matrix. However, the methods used hollow PZT fibers that could be directly woven into fabrics
to form their fibers are different. MFC have rectangular cross- have been developed [17]. The majority of researches in this
section fibers machined from low cost piezoelectric wafers field mainly focus on incorporating PFC in inertia driven
using a computer-controlled dicing saw, whereas the fibers of mechanisms [13, 18]. To the best knowledge of the authors,
PFC are formed through a particular injection molding pro- none of these applications address the challenge of integrating
cess called viscose suspension spinning process and are PFC in a real stretchable and wearable mechanism and none
typically circular in cross-section. The rectangular shape of of the published papers presents a theoretical energy con-
MFC fibers gives the maximum contact area between the PZT version model.
fibers and the interdigitated electrodes due to the larger sur- The main objective of this paper is to demonstrate, in
face area when compared to the circular fibers of PFC, that principle, the potential of using PFC in the chin strap of head-
results in more efficient transfer of electric field to the PZT mounted devices to form a flexible wearable mechanism for
fibers. energy harvesting from jaw movements. Furthermore, this

2
Smart Mater. Struct. 23 (2014) 105020 A Delnavaz and J Voix

Figure 2. Jaw movement energy harvester and its schematic representation.

Figure 3. Energy chain in jaw movement energy harvesting.

paper presents an energy conversion model of the entire tension. Opening the mouth further stretches the side straps
system based on the bond graph method. In this method, the and causes: (a) distributed force stress in the PFC contact
governing differential equations are systematically derived by surface (direction 1) and (b) tensile stress in the PFC cross-
the graphical representation of the energy flow from section (direction 3). Since the PFC cross-section is much
mechanical to electrical domains. The bond graph analytical smaller than the PFC contact area below the chin and as the
model was used to predict the performance of the system for longitudinal piezoelectric coefficient (d33) is usually larger
different stretch levels in the strap. It was also used to find the than the transverse piezoelectric coefficient (d31), one can
optimal stretch level and determine the resistive load at which assume that the most electric charge accumulated in the PFC
maximum power is harnessed. electrodes are from the tensile stress (direction 3). Finally, the
The rest of the paper is organized as follows. Section 2 electric charge flows through the resistive load and generates
presents the jaw movement energy harvester device. The the electric current in the circuit. By closing the mouth, the
system model is developed in section 3. The experimental system returns to its initial position and a reverse current of
setup and procedure is elaborated in section 4 and the results same magnitude is generated.
are discussed in section 5. Finally, conclusions and future
research needs are presented in section 6.
3. Analytical system modeling

2. Jaw movement energy harvester This section gives insight into the energy conversions and
transformations taking place as the proposed jaw movement
The energy harvesting device consists of a flexible piezo- energy harvesting device is in operation. Also, the bond
electric element made of PFC that fits below the chin and is graph model of the system based on these physical energy
attached to a head-mounted device by two elastic rubber exchanges is developed.
straps as shown in figure 2. The head-mounted device is a pair
of earmuffs in our case. Other types of head-mounted devices 3.1. Energy chain
could also have been used, such as tactical helmets, sports
helmets or headphones. The chain of energy conversions and transformations is illu-
Accordingly, the person wearing the device must adjust strated in figure 3. The two moments when energy conver-
the strap assembly to a snug fit to keep the strap under sions occur are in muscles and the piezoelectric chin strap.

3
Smart Mater. Struct. 23 (2014) 105020 A Delnavaz and J Voix

Figure 4. Bond graph representation of the energy harvesting system.

Energy conversion implies that energy is converted from one TF1 transforms the piezoelectric element deformation to
type of energy to another. Thus, chemical energy available in the strain. Therefore, the modulus of the first transformer is
the bonds of molecules in food is the starting point. Motor γ = 1 l and satisfies the following relations
proteins of the muscles are powered by the hydrolysis of
⎧ f = γαlwt
adenosine tri-phosphate which converts chemical energy into TF1 ⎨ , (2)
mechanical work. The work is done when the force exerted by ⎩ δ˙ = γy˙
a muscle results in movement. The movement leads to a
in which l, w and t are respectively length, width and thick-
mechanical strain in the elastic side straps. The piezoelectric
ness of the piezoelectric element. The mechanical part of the
chin strap can convert kinetic energy of the strain into electric
piezoelectric system is modeled by the middle 1-junction,
power. In addition to these energy conversions, figure 3 also
where the input stress (α) is distributed among piezoelectric
illustrates the energy transformations. These are solely che-
transformer (σ) and elements that represent inertial (σm),
mical, mechanical or electrical. For instance, metabolism is
damping (σb) and elastic (σY ) properties of the piezoelectric
the set of chemical transformations within the cells that allows
material. Also, TF2 models the piezoelectric transducer
organisms to live. In the mechanical segment, the elastic side
whose constitutive equations are given by
straps transform the jaw movements into strain. Finally, the
conditioning circuit transforms the electric charge produced in ⎧
⎪δ =
σ
+ dE
the piezoelectric element to the electric current that suits ⎨ Y , (3)
⎪
the load. ⎩ D = εE + dσ
where δ is the strain, d is the piezoelectric strain constant, E is
3.2. Bond graph
the electric field, D is the electrical displacement, and ε is the
The bond graph model of the energy harvesting device is permittivity of the piezoelectric material. In the right-hand
presented in figure 4. It is composed of three junctions, two side of equation (3), the terms containing d indicate the
transformers and several branches. In this figure, k (λ ) is the energy conversion from one energy domain to another.
spring constant of the elastic side straps, m, b, Y, and C are Therefore, the modulus of TF2 is defined as η = Ylwd , and
respectively the mass, damping ratio, Youngʼs modulus and gives the following relationships between mechanical and
capacitance of the piezoelectric strap, and R is the load electrical variables
resistance.
⎧ σlwt = ηv
At the far left branch, the source element (S) is located. It TF2 ⎨ , (4)
represents the jaw movement as a general source of power ⎩ i = ηδ˙
that, depending on the subject, can present any relationship
where v is the voltage and δ̇ is the derivative of strain. Finally,
between the force (f) that it applies and the velocity at which it
the right 0-junction represents the electrical part of the system
excites the energy harvesting device ( ẋ ). The left 0-junction
and divides the produced current (i) between the piezoelectric
models the strap assembly and splits ẋ between the piezo-
capacitance (iC) and the load resistance (iR).
electric chin strap (ẏ ) and the elastic side straps (ż ). The
In each bond graph junction, flow and effort are expres-
elastic material of the side straps becomes progressively
sed in terms of different variables: f and ẋ are the pair of
stiffer as it deflects and hence is modeled by a nonlinear
power variables for the left 0-junction. σ and δ̇ are the state
stiffness element, k (λ ), where λ is the side strap stretch ratio
variables in the piezoelectric 1-junction. However, the mul-
and is defined as the ratio between the strapʼs final length (Zf)
tiplication of these two variables is equal to the power of the
after putting on the device and its initial (free) length (Z0)
unit volume and hence the total power needs to be corrected
before putting on the device:
by multiplying with the volume of the piezoelectric material
Zf [19]. The electric power in the right 0-junction is considered
λ= . (1) as the product of v and i.
Z0
Considering the direction of power arrows and the
property of junctions, the three following equations are

4
Smart Mater. Struct. 23 (2014) 105020 A Delnavaz and J Voix

Figure 5. Experimental setup.

derived as

⎧ x˙ = y˙ + 2z˙ ⎡ ⎛ 1 b⎞ ⎛1 ⎡ k (λ ) l ⎤
⎪ V ⎢ s3 + ⎜ + ⎟ s2 + ⎜ ⎢ Y +
⎨ α = σm + σb + σY + σ . ⎝m⎣ ⎥
(5) ⎣ ⎝ RC m⎠ 2wt ⎦
⎪
⎩ i = iC + i R
b η2 ⎞
+ + ⎟s . (8)
Considering the properties of the bond graph elements, mRC lwtmC ⎠
integrating the first equation and substituting the transformer 1 ⎛ k (λ ) l ⎞ ⎤ ⎛ k (λ ) η ⎞
+ ⎜Y + ⎟⎥ = X ⎜ ⎟s
relations (equations (2) and (4)) into set of equation (5), the mRC ⎝ 2wt ⎠ ⎦ ⎝ 2wtmC ⎠
dynamic model of the system can be expressed as
The output voltage can be calculated for a given input
⎧ δ 2f vibration frequency. The Laplace variable is substituted by jω
⎪x = +
⎪ γ k (λ ) and ωn2 = (Y + k (λ ) l 2wt ) m is the square of the natural
⎪ f ηv . frequency of the converter. The system damping ratio is
⎨ = mδ¨ + bδ˙ + Yδ + (6)
⎪ γlwt lwt defined by ζ = b 2mωn . The square of the piezoelectric
⎪ v coupling coefficient is expressed as μ2 = d 2Y 2lw mC . Mak-
⎪ ηδ˙ = Cv˙ +
⎩ R ing these substitutions and solutions for V yields

Taking the Laplace transform of equation (6) and using jA (ω)

V = X, (9)
X, Δ, F and V as the respective transformed variables of x, δ, f B (ω) + jG (ω)
and v yields
where the frequency dependent terms of A, B and G are
⎧ 1 2 defined by
⎪X = Δ + F
⎪ γ k (λ ) ⎧ k (λ) dYl
⎪ 1 η ⎪ A (ω) = ω
⎨
⎪ γ lwt
(
F = ms2 + bs + Y Δ +)lwt
V, (7) ⎪ 2mCt
⎪ ⎛ 1 ⎞ ω2
⎪ ⎛ 1⎞ ⎨ B (ω) = −⎜ + 2ζωn ⎟ ω2 + n . (10)
⎪ ηsΔ = ⎜ Cs + ⎟ V ⎪ ⎝ RC ⎠ RC
⎩ ⎝ R⎠ ⎪ ⎛ 2ζωn ⎞
⎪ G (ω) = ⎜ − ω2 + ωn2 + + μ2 ⎟ ω
⎩ ⎝ RC ⎠
where s is the Laplace variable. Combining the relations in
equation (7) and rearranging certain terms yields the rela-
tionship between the jaw movement input and output voltage

5
Smart Mater. Struct. 23 (2014) 105020 A Delnavaz and J Voix

4. Experimental study

4.1. Description of the experimental setup used

The experimental test setup is shown in figure 5 and consists

1
of a 6 2 digit multimeter (Agilent Technologies, CA, USA),
and a breadboard with the required electric components for
the test. The energy harvesting device is composed of a pair
of earmuffs (Headguard, France) and a strap assembly. The
core of the strap is a rectangular piezoelectric element, PFC-
W14 (Advanced Cerametrics, NJ, USA) for which the engi-
neering properties are presented in table 1 . Two side straps,
made of a braided polyester elastic, are attached to the flexible
piezoelectric chin strap by two clips.

4.2. Experimental procedure

Given the properties of the piezoelectric material, the tension Figure 6. Voltage generation of the jaw movement energy harvester
in the piezoelectric strap can be controlled by regulating its measured by 10X oscilloscope probe for the normal gum chewing
initial stretch in the side straps by adjusting the length of the when R = 900 kΩ and λ = 1.
side straps before putting on the device (Z0). Since the pre-
scribed strap tension influences the chewing pattern and Table 1. Properties of PFC-W14
rhythm, it is quite reasonable to assume that the jaw move- Property Symbol Unit Value
ment cycle frequency (ω) and the maximum deformation
amplitude of the strap (X) are functions of the stretch in the Dimensions l×w×t mm 132×14×0.3
Charge coefficient d pC/N 550
strap (λ). These two variables as well as k (λ ) are needed to
Youngʼs modulus Y GPa 24.4
solve equation (9) and should be measured before or during Capacitance C nF 10
the test. Mass m kg 0.11
To measure X, a simple mechanism was designed: it is Damping ratio ζ — 0.1
composed of a sufficiently long unstretchable cord that is
attached to the earmuffs just like the main strap is. A plastic
nut allows the cord length to fit around the chin. When the 4.2.2. Assessment of the overall performance. Once the
subject starts moving his jaw, similar deformations are optimum load resistance value was found, the main
applied to both elastic strap and non-elastic cord. The elastic experiment focused on calculating the power delivered to
strap stretches, and the nut allows the cord to lengthen by this load for different stretch levels in the side straps. For each
sliding through the plastic nut. Hence, knowing the final and stretch level the subject chewed gum for 60 s adapting his
initial length of the cord, one can measure the maximum chewing manner with the prescribed tension. The voltage of
stretch in the strap assembly. Also, ω can be simply calcu-
the load corresponding to each stretch level was then
lated by counting the number of chews during the period of measured using the digital multimeter and the harnessed
measurement. k (λ ) can be measured before the test as a effective power was calculated using
function of the stretch in the strap by applying a force through
a dynamometer and measuring the resulting elongation. V2
The following experiments were designed to test the P= . (11)
R
performance of the energy harvesting device and validate the
effectiveness of the developed bond graph model.

4.2.1. Determining the optimal load resistance. The first 5. Results and discussions
experiment was designed to find the optimum load resistance
where the power output is maximized. The side straps were The voltage output of the piezoelectric energy harvester
adjusted to maintain sufficient tension to secure the device in device during 60 s of the normal gum chewing was measured
its normal position below the chin (λ = 1). The test subject using 10X oscilloscope probe and the results are shown in
was asked to chew gum for 60 s and the maximum output figure 6. These results were obtained for R = 900 kΩ and
voltage was measured for a given resistor. This test was λ = 1.
repeated for an assortment of resistors between 1 kΩ and Also, the power output of the energy harvesting device is
20 MΩ in value. The maximum harvested power was then illustrated in figure 7 as a function of the resistance values
calculated for a given λ and the corresponding resistance when λ = 1. To plot the bond graph predictions curve,
value was consequently chosen to represent the load for all λ equation (9) was first solved for the voltage and then the
values, because ω is assumed to slightly vary with λ. power was calculated by using equation (11). The

6
Smart Mater. Struct. 23 (2014) 105020 A Delnavaz and J Voix

Figure 7. Comparison between the bond graph predictions and the Figure 9. Comparison between the bond graph predictions and the
experimentally measured generated power (P) as a function of the experimentally measured power (P) delivered to the load of
load resistance (R) for λ = 1. R = 900 kΩ (optimum resistive load for λ = 1) as a function of the
side strap stretch ratio (λ).

trend for X is mixed. It first increases, then remains nearly

constant and finally decreases. Therefore, three distinct levels
of strap stretch can be identified based on X: low stretch
(λ < 1.2), medium stretch (1.2 < λ < 1.6) and high stretch
(λ > 1.6).
Figure 8 also includes the plot of k (λ ). This curve
exhibits an exponential increase at the higher stretch ratios,
because rubber, as the main component of the elastic side
strap, has an intrinsic nonlinear hardening characteristic. The
increase in k (λ ) at high stretches (when the strap is tight)
results in the remarkable reduction in X because of the extra
effort needed in order to open the jaw against the higher strap
tension. On the other hand, X increases at low stretches (when
the strap is looser) mainly due to the insufficient tension that
causes the strap to be displaced instead of being stretched.
The test subject could confirm this type of strap displacement
Figure 8. Experimental parameters required to solve equation (9): during the test. In addition, no important changes in X can be
non-elastic cord maximum elongation (X), jaw movement frequency observed at medium stretches.
(ω), and elastic side strap spring constant (k (λ )) as a function of the The variation of power generation versus stretch ratio is
side strap stretch ratio (λ).
represented in figure 9 for both bond graph model predictions
and experimental measurements. To plot the bond
experimental measurements show that the maximum power graph curve, the experimental results presented in figure 8
transfer happened at R = 900 kΩ. According to figure 7, were employed to solve equations (9) and (11). As shown in
although the bond graph model predicts a slightly lower this figure, the power curve increased at both low and high
resistance value for the power transfer peak, its curve was stretch levels, while it decreased at medium stretch ratios. For
sufficiently close to the experimentally measured power. It low stretches, the effect of increase in X is larger than the
validates the developed bond graph model for optimization of amount of decrease in ω and hence the generated power
the load resistance in the piezoelectric energy harvesting increases in proportion to the stretch ratio. At medium stret-
applications. ches, power generation decreases mostly because of the
The performance analyses were subsequently carried out decrease in ω while X is mostly constant. Finally, despite the
and ω and X are presented in figure 8. In this figure, the substantial reduction in X at high stretches, power generation
horizontal axis is the side strap stretch ratio (λ) that is a still increased by λ due to the remarkable increase in k (λ ).
measure of the tension in the strap. According to figure 8 and According to figure 9, the experimental results match the
considering the fact that the test subject was free to adapt his curve predicted by the bond graph model. Therefore, it can be
chewing pattern with the level of stretch in the strap, one can successfully used for modeling other flexible piezoelectric
observe that ω slightly decreases as λ increases. However, the energy harvesting mechanisms. It predicts a local power peak

7
Smart Mater. Struct. 23 (2014) 105020 A Delnavaz and J Voix

of 7 μW at λ = 1.2 for the low and medium λ. Although this References

value is lower than the power harnessed for the high λ, it
guarantees the comfort of the wearer at all times. [1] Goll E, Zenner H-P and Dalhoff E 2011 Upper bounds for
energy harvesting in the region of the human head IEEE
Trans. Biomed. Eng. 58 3097–103
[2] Qi Y and McAlpine M C 2010 Nanotechnology-enabled
6. Conclusions and future research needs flexible and biocompatible energy harvesting Energy
Environ. Sci. 3 1275
[3] Sun C, Shi J, Bayerl D J and Wang X 2011 PVDF microbelts
This research opens the possibility of harvesting energy from for harvesting energy from respiration Energy Environ. Sci.
jaw movements to sustainably power small-scale electronic 4 4508–12
devices. An energy harvester device containing the piezo- [4] Granstrom J, Feenstra J, Sodano H A and Farinholt K 2007
electric chin strap was designed, a bond graph model was Energy harvesting from a backpack instrumented with
piezoelectric shoulder straps Smart Mater. Struct. 16
developed and a prototype was fabricated and tested. The 1810–20
results show that the optimum resistance value of the elec- [5] Swallow L M, Luo J K, Siores E, Patel I and Dodds D 2008 A
trical load is about 900 kΩ for λ = 1. The maximum gener- piezoelectric fibre composite based energy harvesting device
ated power for the optimum resistive load was theoretically for potential wearable applications Smart Mater. Struct. 17
and experimentally investigated and the strap stretch is found 025017
[6] Qi Y, Jafferis N T, Lyons K, Lee C M, Ahmad H and
to be the most significant design parameter. According to the McAlpine M C 2010 Piezoelectric ribbons printed onto
experimental findings, although increasing the strap stretch rubber for flexible energy conversion Nano Lett. 10 524–8
can increase power generation, the increased effort required to [7] Chen X, Xu S, Yao N and Shi Y 2010 1.6 V nanogenerator for
chew might negatively affect power generation. Therefore, mechanical energy harvesting using PZT nanofibers Nano
the need for energy recovery from the jaw movements should Lett. 10 2133–7
[8] Yang R, Qin Y, Li C, Zhu G and Wang Z L 2009 Converting
remain within reasonable boundaries so as to give precedence biomechanical energy into electricity by a muscle-
to the comfort of the wearer. It appears that the best results movement-driven nanogenerator Nano Lett. 9 1201–5
can be achieved when the strap stretch ratio is about 1.2. For [9] Yang Y, Tang L and Li H 2009 Vibration energy harvesting
this ratio, the maximum power transfer is about 7 μW. The using macro-fiber composites Smart Mater. Struct. 1 115025
developed bond graph model presents a satisfactory match [10] Cha Y, Kim H and Porfiri M 2013 Energy harvesting from
underwater base excitation of a piezoelectric composite
with the experimental measurements and this emphasizes the beam Smart Mater. Struct. 22 115026
effectiveness of the proposed approach for modeling the [11] Shahab S and Erturk A 2014 Electrohydroelastic dynamics of
human power energy harvesting system. macro-fiber composites for underwater energy harvesting
from base excitation Proc. SPIE 9057 90570C
[12] Erturk A and Delporte G 2011 Underwater thrust and power
generation using flexible piezoelectric composites: an
Future research needs experimental investigation toward self-powered swimmer-
sensor platforms Smart Mater. Struct. 20 125013
The quantity of power generated by the designed jaw
[13] Qi S, Shuttleworth R, Oyadiji S O and Wright J 2010 Design of
movement energy harvester device is still too little to be a multiresonant beam for broadband piezoelectric energy
directly adopted in practical applications. Future designs may harvesting Smart Mater. Struct. 19 094009
focus on increasing the number of piezoelectric elements to [14] Kymissis J, Kendall C, Paradiso J and Gershenfeld N 1998
supply the demanded power of small-scale and wearable Parasitic power harvesting in shoes Digest of Papers. 2nd
Int. Symp. on Wearable Computers (Cat. No.98EX215) (Los
electronic devices. Moreover, an appropriate power manage-
Alamitos, CA: IEEE Computer Society Press) pp 132–9
ment circuit would be required to charge a tiny rechargeable [15] Vatansever D, Hadimani R L, Shah T and Siores E 2011 An
battery integrated into the device. Finally the entire strap investigation of energy harvesting from renewable sources
assembly should be encased in such a way that would make it with PVDF and PZT Smart Mater. Struct. 20 055019
totally biocompatible. [16] Sodano H A, Lloyd J and Inman D J 2006 An experimental
comparison between several active composite actuators for
power generation Smart Mater. Struct. 15 1211–6
[17] Guillot F M, Beckham H W and Leisen J 2013 Hollow
piezoelectric ceramic fibers for energy harvesting fabrics
Acknowledgements J. Eng. Fiber. Fabr. 8 75–81
[18] Van Blarigan L, Danzl P and Moehlis J 2012 A broadband
The authors would like to thank Sonomax Technologies Inc. vibrational energy harvester Appl. Phys. Lett. 100 253904
[19] Gao R X and Cui Y 2005 Vibration-based sensor powering for
and its ‘Industrial Research Chair in In-ear Technologies’ for manufacturing process monitoring NAMRC Thirty-Three
its financial support and for providing specific equipment (New York: Society of Manufacturing Engineers)
required for the experimental setups. p TP05PUB144