2017 IEEE 8th International Conference on CIS & RAM, Ningbo, China

Design of a Two-Stage Force Amplification Frame

for Piezoelectric Energy Harvesting
Shihao Wen and Qingsong Xu∗
Department of Electromechanical Engineering, Faculty of Science and Technology
University of Macau
Avenida da Universidade, Taipa, Macau, China
Email: {mb65441, qsxu}@umac.mo

Abstract—As alternative power sources for replacing batteries, stack and decreases the stiffness of piezoelectric stack, as well
piezoelectric energy harvesters are designed for the applications as enhancing the electric power output.
which demand low-power electronics and sustainable electrical So far, several studies have been conducted on the design of
power. This paper presents the design of a new piezoelectric
energy harvester with two-stage force amplification frame for the force amplification frame. There are two major mechanisms.
purpose of harvesting energy from human footstep. The two-stage Feenstra et al. [7] proposed a convex frame for harvesting
force amplification frame can magnify the walking force applied energy from backpack. Zhou et al. [8] further studied this type
to the piezoelectric stack and thereby enhances the power output. of frame and proposed a numerical modeling. Nevertheless,
Analytical model is established to obtain the relationship between their amplification ratio is relatively small. Chen et al. [9] of-
the input and output based on elastic beam theory. The frame
parameters are designed and optimized to meet the requirements fered a concave type model with a nonlinear theoretic analysis.
on low-frequency condition and to achieve the maximum force Force magnification ratio is effectively improved by parametric
amplification ratio. Finite element analysis is conducted to verify design model. Wang et al. [10] investigated a two-stage
the performance of the proposed frame. force magnification frame based on concave structure. The
maximum amplification ratio is 21 by using three piezoelectric
I. I NTRODUCTION stacks and four force amplification frames. Concerning the
performance of the two-stage magnification frame, there is still
Recently, the advances in electronics, wearable devices and room to enhance the capability of the structure based on the
wireless technologies have drawn increasingly attention from multiple parameters design. In fact, regarding the mechanism
researchers [1]. Conventionally, these devices need batteries of two-stage force amplification structure, it is possible to just
to supply electrical power. However, battery technologies face utilize one piezoelectric stack to obtain a large amplification
critical challenges as they require continual recharge and have ratio and compact size.
a limited lifespan. When reaching a critical point, the charge To this end, a new piezoelectric energy harvester consisting
performance could drop dramatically. Therefore, in view of of one two-stage force amplification frame and one piezoelec-
the required power for operating, electronics power has been tric stack is designed and studied in this paper. In addition,
reduced to micro-watts level. Piezoelectric energy harvesting a novel analytical modeling is proposed to analyze the con-
technique as an alternative power source has been gradually nections between design variables. Multi-parameter analysis is
developed, because it can convert environmental mechanical carried out to optimize the design variables. FEA simulation
energy into electrical energy. study is conducted to predict and verify the performance of
One of the effective approach is to harvest energy from the energy harvester under low-frequency situation.
human locomotion [1], [2], [3], [4]. While human is walking
on the ground, a small deformation is formed to alleviate II. M ECHANICAL D ESIGN
the impact between foot and ground. The work done by the The designed two-stage force amplification frame is shown
walking force can be converted into electric energy by using in Fig. 1. Both force magnification structures are convex,
piezoelectric material. One possibility is to directly use d33 because they hold the potential to improve the amplification
mode piezoelectric stack [5], [6]. However, the stand-alone ratio versus previous work and can obtain a relatively compact
piezoelectric stack could only harvest a small amount of ener- size in comparison with concave type.
gy and it is vulnerable to damage without any protection. To The working principle of the two-stage force amplification
overcome these issues, the application of force amplification frame lies in that the bottom of frame 1 is rigidly fixed. The
frame is suggested, which can convert the applied low force walking force, Fi , is applied on the top of frame 1. Frame 1
in the input end to a larger force at both sides of piezoelectric transfers the input force to the frame 2. It can transfer the
force to push against piezoelectric stack. During the process,
This work was supported in part by the Macao Science and Technology the force is magnified and electric power is produced.
Development Fund under Grants 143/2016/A and 090/2015/A3.
∗ Corresponding author. Q. Xu (Phone: +853 8822 4278; Fax: +853 8822 Fig. 2 shows three major types of beams which can be
2426; E-mail: [email protected]). applied in the design, including beam-type, beam-hinge-type

978-1-5386-3135-5/17/$31.00 ©2017 IEEE 490

 2017 IEEE 8th International Conference on CIS & RAM, Ningbo, China

Fig. 1. The two-stage amplification frame.

Fig. 3. The free-body diagram of flexure beam.

(a) (b) (c)

Fig. 2. Types of beams. (a) Beam-type, (b) Beam-hinge-type 1, (c) Beam-

hinge-type 2.

TABLE I
T HE COMPARISONS BETWEEN THREE TYPES OF BEAM

Variables Beam-type Beam-hinge-type 1 Beam-hinge-type 2

Safety Factor Good Medium Bad
Force Medium Good Bad
Size Good Medium Medium
(a)

1 and beam-hinge-type 2. Each one has its own strengths and

weaknesses. The comparisons between these three types of
beam are shown in Table I. Beam-type frame is the original
design. As the whole beam has the same thickness, it could
share the stress and reduce the rick of damage. On the other
hand, as the maximum stress occurs on the end side of beam,
both beam-hinge-type 1 frame and beam-hinge-type 2 frame
contain hinges, which tend to reduce the stiffness of the beams
and result in more strain energy to be generated. However,
the ability to bear stress is reduced for the structures with
hinges. The increase of the beam thickness alleviates the (b) (c)
damage, while results in the decrease of the amplification
Fig. 4. (a) The deformation of flexure beam, (b) The elongation of flexure
ratio. This is the reason why the beam-hinge-type 2 frame has beam, (c) The deformation of side block.
a bad performance. Hence, the beam-type is selected in this
design by considering the safety factor, force amplification,
and physical size of the device. beams are considered as elastic beams and other components
are taken as rigid body.
III. A NALYTICAL M ODELING Assign Fi , Fo , lx and ly as the input force, output force,
In view of analytical modeling of the force amplifier, related vertical and horizontal dimensions of the flexure beam, re-
work is relatively less than the modeling of displacement spectively. The amplification ratio n can be regarded as the
amplifier. But the basic theory is similar [9], [11], [12]. ratio between the input force Fi from human walking and
Therefore, the energy method is carried out in this design the transverse force Fo in y direction. In view of the moment
owing to its effectiveness [13]. equilibrium at the beam, the relationship is expressed as:
First, one quarter of frame 1 is analyzed as it is a double-
2Mr = Fo ly − Folx = Fi lb cos θ − Folb sin θ (1)
symmetry structure. The simplified model and the free-body
diagram of connecting arms are shown in Fig. 3. Flexure where Mr is the moment applied on the beam, θ is the tilt

491
 2017 IEEE 8th International Conference on CIS & RAM, Ningbo, China

angle, and lb denotes the length of A1 B1 .

The corresponding moment at the end point B1 can be
written as:
M1 (x) = Fn x + Mr , 0 ≤ x ≤ lb (2)
where Fn is the normal force along the beam direction which
is equal to Fi cos θ − Fo sin θ .
The force analysis of the elastic beam is shown in Fig. Fig. 5. The equivalent circuit of piezoelectric material.
4(a). According to Euler-Bernoulli beam theory, the bending
equation of the flexure beam can be generated as:
d2w M (x) According to equations (5), (6), (11), and (12), the total
= (3) deformations in x and y directions become:
dx2 EI (x)
where E denotes the Young’s modulus of the material, w Δx = x1 − x2 + x3 (13)
represents the deformation of the beam to neutral axis x, and
I (x) is the area moment inertia of the structure. Δy = y1 − y2 (14)
Integrating equation (3) by applying the boundary condition
and substituting x = lb into it, the deformation of the beam Based on the law of conservation of energy, the work done
becomes: by the input force is transformed to the strain energy of the
Fn lb3 M1 lb2 frame. Then, the following relationships can be obtained:
w (lb ) = + (4)
6EIbeam 2EIbeam
1 1
where Ibeam denotes the area moment inertia of beam which E1 = − Fo Δx − Fi Δy (15)
is equal to bt 3 /12. 2 2
Then, the deformations in x and y directions can be written  lb  lb  lb
Fl2 (x) M12 (x) M22 (x)
as: E2 = dx + dx + dx (16)
x1 = w (lb ) sin θ (5) 0 2EAbeam 0 2EIbeam 0 2EIblock

y1 = w (lb ) cos θ (6) E1 = E2 (17)

The mechanical analysis of the side block is shown in Fig. Equation (17) is a function of the force variables Fi and Fo .
4(c). In view of the moment equilibrium at the side block, the The amplification ratio can be calculated from equations (15),
following relationship is generated. (16) and (17) by applying the symbolic computation. For the
M0 = −Folblock − Mr (7) two-stage amplification frame, the total magnification ratio is
equal to n1 n2 . The relationships can be expressed as:
Similarly, the corresponding moment at the end point C1
F1o F2o F2o
and the deformation of beam at lbock are obtained as: n = n1 n2 = = (18)
F1i F2i F1i
M2 (x) = −Fo x − M0, 0 ≤ x ≤ lblock (8)
Furthermore, in view of piezoelectric constitutive equation,
Fo l 3 2
M0 lblock the voltage is produced in low frequency and open-circuit
x2 = − block − (9)
6EIblock 2EIblock condition. When a force F2o is applied to the piezoelectric
where lblock denotes the half length of the side block, and Iblock stack, the expression for the electric output can be written as:
represents the area moment inertia of beam which is equal to g33 F2o
3
bblock tblock /12. V= (19)
d
The elongation of the beam and the corresponding defor-
mations in x and y directions are shown in Fig. 4(b). The where V denotes the generated voltage, d is the width of
corresponding equations can be generated as: piezoelectric stack, and g33 is the piezoelectric constant.
Concerning the equivalent circuit of piezoelectric material
Fl lb2
Δlb = (10) shown in Fig. 5, the electrical power can be obtained:
EAbeam
x3 = Δlb cos θ (11) V2
P= (20)
R
y3 = Δlb sin θ (12)
where R is external resistor.
where Fl is the force along the beam direction which is equal It is found that the electrical power has a positive influence
to Fi sin θ + Fo cos θ , and Abeam is the cross-sectional area of on the amplification ratio. That is, a larger magnification ratio
the beam. results in more electrical power.

492
 2017 IEEE 8th International Conference on CIS & RAM, Ningbo, China

2.2
1650
TABLE II
T HE CONSTRAINTS AND OBJECTIVE OF DESIGN VARIABLES 1600 2.1

1550
2

Output variables Symbol Constraint Objective 1500

Safety Factor
1.9

Input Deformation d [0, 3 mm]

Force
1450
1.8
Safety factor k [2, 3] 1400

1.7
Amplification ratio n To maximize 1350

1.6
1300

1250 1.5
10 11 12 13 14 15 16 17 18 19 20 10 11 12 13 14 15 16 17 18 19 20
Frame1 beam width Frame1 beam width
IV. PARAMETER O PTIMIZATION
(a) (b)
Concerning parameter optimization of the device, it is 1320
2.4

important to select some appropriate parameters. In the lit- 1300

2.3

erature, Wang et al. [10] proposed a two-stage amplification 2.2

1280
frame. But the authors only considered the thickness of the 2.1

Safety Factor
Force
2
frame beam and the tilt angle as design variables. The size 1260

1.9
optimization has not been adequately studied. In fact, by the 1240
1.8

help of size optimization, the frame weight can be reduced 1220

1.7

obviously and its dynamic performance can be improved. In 1200 1.6

3 3.2 3.4 3.6 3.8 4 4.2 4.4 4.6 4.8 5 3 3.2 3.4 3.6 3.8 4 4.2 4.4 4.6 4.8 5

addition, although Chen et al. [9] discussed the impact of Frame1 beam height Frame1 beam height

various parameters for a single force amplifier, the authors (c) (d)
just mentioned the results without any inference process. 1300
3

2.9

In view of previous works, it is found that a small value 1200

2.8

of tilt angle and thinner thickness of beam tend to increase 1100

2.7

Safety Factor
the amplification ratio. In fact, as long as the parameters are 1000 2.6
Force

2.5
related to the beam, the output force is definitely influenced. 900

2.4

Therefore, the optimization of the width, height, thickness and 800

2.3

length of the frame beam are indispensable in order to obtain 700

2.2

a compact size and maintain the stability of structure. 600

12 13 14 15 16
Frame1 beam length
17 18 19 20
2.1
12 13 14 15 16
Frame1 beam length
17 18 19 20

For example, stress distribution can be improved by properly (e) (f)

increasing the width and decreasing the length. Finally, the 2000 7

design variables are 1800

6
1600

Design variables(h1b , l1b , w1b ,t2b , h2b , l2b , w2b ,t2b ) (21) 1400 5

Safety Factor
1200
Force

4
where h, l, w and t denote the height, length, width and 1000

thickness of beam, respectively. 800 3

600
Moreover, there are limitations on the design of the two- 400
2

stage force amplifier. The priority in this design is to ensure a 200

0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3
1
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3

normal walking motion. In practice, large deformation may Frame1 beam thickness Frame1 beam thickness

result in height difference, which may pose a risk to the (g) (h)
people walking on the street. Thus, the total deformation of Fig. 6. The force and stress performance of various design variables. (a)
the frame along with the walking direction is limited to less Force vs. width, (b) safety factor vs. width, (c) force vs. height, (d) safety
than 3 mm to avoid the hidden danger. Under this case, the factor vs. height, (e) force vs. length, (f) safety factor vs. length, (g) force vs.
thickness, (h) safety factor vs. thickness.
amplification ratio could be restricted. Additionally, the frame
must guarantee not to be damaged by the applied force. Thus,
the maximum yield stress in the material should be higher
than the total maximum stress in the frame. The safety factor the frame 1 and frame 2 influence each other, which could
is defined as the ratio of the maximum yield stress in the reduce the amplification ratio. Therefore, in the following anal-
material and the total maximum stress in the frame. The whole ysis, parameter optimization is implemented by considering
system is safe when the safety factor exceeds 1. Thus, the the influence of the whole system.
amplification ratio is optimized under the restriction on the The parameter optimization is conducted by utilizing AN-
maximum stress and deformation on the top end of the frame. SYS software. The piezoelectric stack is selected as the type of
The main constraints and objectives of design parameters are P-885.91 from PI company. The mesh method for the frames
shown in Table II. and the piezoelectric stack are tetrahedron with the size of 1
In view of the two-stage amplification frame, it is not mm and hexahedron with the size of 0.5 mm, respectively.
suggested to optimize the force amplifier separately. Single First, the performance of frame 1 with the design variables
force amplifier can only achieve unilateral magnification. It is tested under FEA analysis. The input force with 100 N is
cannot satisfy the optimization of the overall size. Moreover, applied at top of frame 1. The bottom is fixed. The results

493
 2017 IEEE 8th International Conference on CIS & RAM, Ningbo, China

TABLE III
PARAMETERS OPTIMIZATION RESULT

Design Variables Symbols Values

Frame1 Beam Length l1b 20
Frame1 Beam Thickness t1b 0.5
Frame1 Beam Height h1b 4
Frame1 Beam Width w1b 20
Frame2 Beam Length l2b 15.7
Frame2 Beam Thickness t2b 0.8
Frame2 Beam Height h2b 3.1
Frame2 Beam Width w2b 10
(a)

are shown in Fig. 6(a) to 6(h). The force denotes one side of
reaction force of the output.
It is found that the same trend appears in the change of
width, height and thickness. It is noted that a force increase
results in the decrease of safety factor. It is important to
find a trade-off between the two parameters. Concerning the
influence of height, it exhibits a critical value where the force
reaches the maximum value. It is different from the other
parameters causing the maximum value in the both sides of
design variables. As for the impact of length and thickness, (b)
these two variables are significant as they have larger slope.
Fig. 7. The stress distribution of frames. (a) Two-stage energy harvester, (b)
Theoretically, the amplification ratio could become very large Frame 1, (c) Frame 2.
if the thickness becomes thinner and length becomes longer.
However, the thickness is hard to become less than 0.2 mm
due to manufacturing issues. In addition, the thinner beam is
easier to cause bending and lends to deformation before the
external force is applied. Only when the safety factor satisfies
the requirement, the amplification ratio could be considered.
Next, the MOGA (Multi-Objective Genetic Algorithm)
method is applied to optimize the selected parameters. There
are 300 initial samples, 100 samples per iterations and 5
candidate points in order to search for the best solution. The
parameters optimization results are shown in Table III.
V. FEA S IMULATION AND M ODEL V ERIFICATION
The frame was fabricated from Aluminum 7075-T6 due to Fig. 8. The safety factor of two-stage amplification frame.
its better material properties. The detailed material information
is shown in Table IV. TABLE IV
The amplification ratios of frame 1 and frame 2 are tested, M ECHANICAL PROPERTIES OF MATERIAL AL-7075-T6
respectively, under FEA simulation. A force of 100 N is
Properties
  Values
applied on both sides of input end. The output ends are fixed Density kg/m3 2810
to find the reaction force of frames. The stress distributions of Young’s modulus(GPa ) 71.7
frames are shown in Fig. 7(a) and 7(b). The performance of Tensile yield strength(MPa ) 503
each single amplification frame and the force comparison of Poisson’s ratio 0.33
analytical modeling and FEA simulation are showed in Table
V. It is found that the modeling result is consistent with the
FFA simulation result. becomes more complex, and the actual amplification ratio is
In comparison with the stress distribution of each single smaller than the product of each single amplification ratio. It is
frame (see Table V and Table VI), the two-stage frame bears further confirmed that the optimization of the two-stage frame
more stress and creates more energy because the force is is better than the optimization of frames, separately.
magnified step-by-step. The pushing force needs more energy
and is relatively easier to create deformation. On the other VI. R ESULTS AND D ISCUSSIONS
hand, in view of the performance of force amplification ratio, It can be found that the displacement is greatly decreased at
when the two frames are combined together, the mechanism each output end. The force is magnified with the decrease of

494
 2017 IEEE 8th International Conference on CIS & RAM, Ningbo, China

TABLE V store the harvested energy for subsequent use.

T HE PERFORMANCES AND COMPARISONS OF EACH SINGLE FRAME
VII. C ONCLUSION
Performances Frame 1 Frame 2
Von-Mises Stress(MPa ) 49.65 72.68 The proposed two-stage amplification frame can satisfy
Force amplification ratio-FEA 9.54 8.88 the design requirements for practical conditions in harvesting
Force amplification ratio-Modeling 9.60 8.57 energy of human walking on ground. The analytical model is
Size: L × W × H (mm) 50 × 20 × 24 40 × 10 × 21.20 verified by FEA simulation study, which is suitable for single
beam-type force amplification frame. For the two-stage frame,
TABLE VI multi-parameter optimization is important to fully optimize
T HE PERFORMANCES OF TWO - STAGE ENERGY HARVESTER
the structure in order to obtain large amplification ratio, large
Performances Two-stage force amplification frame safety factor and compact physical size. Through the opti-
Force amplification ratio-FEA 26 mization, the maximum force amplification ratio is obtained
Input deforamtion (mm) 1.77 as 26. Besides, the designed two-stage energy harvester could
Frame 1’s safety factor 2.15 be applied not only in harvesting energy from footstep, but
Frame 2’s safety factor 2.22
also in other portable applications like shoes and backpack.
Frame 1’s stress(MPa ) 234.06
Frame 2’s Stress(MPa ) 226.81
Size: L × W × H (mm) 50 × 20 × 40 R EFERENCES

[1] G. Bassani, A. Filippeschi, and E. Ruffaldi, “Human motion energy

displacement. Based on the principle of energy conservation, harvesting using a piezoelectric mfc patch,” in Proc. of 2015 37th Annual
International Conference of the IEEE Engineering in Medicine and
the input energy is equal to the sum of output energy and Biology Society (EMBC), 2015, pp. 5070–5073.
stored strain energy in the two-stage frame, i.e., Ei = Eo + Es . [2] H.-S. Cho, J.-H. Yang, S.-H. Park, K.-S. Yun, Y.-J. Kim, and J.-H.
However, not all of input energy is converted into output ener- Lee, “An exploratory study on the feasibility of a foot gear type energy
harvester using a textile coil inductor,” Journal of Electrical Engineering
gy. The remaining strain energy is stored in the amplification & Technology, vol. 11, no. 5, pp. 1210–1215, 2016.
frame, which could protect the structures from damage. Hence, [3] L. Xie and M. Cai, “An in-shoe harvester with motion magnification for
a compromise between these two kinds of energy is the key to scavenging energy from human foot strike,” IEEE/ASME Transactions
on Mechatronics, vol. 20, no. 6, pp. 3264–3268, 2015.
maintain an optimal force amplification ratio and safety factor. [4] N. Sharpes, D. Vučković, and S. Priya, “Floor tile energy harvester
It is should be noted that the two-stage amplification frame for self-powered wireless occupancy sensing,” Energy Harvesting and
is sensitive to external load, and the external load cannot Systems, vol. 3, no. 1, pp. 43–60, 2016.
[5] T.-B. Xu, E. J. Siochi, J. H. Kang, L. Zuo, W. Zhou, X. Tang, and
exceed the design requirement. Concerning the design of X. Jiang, “Energy harvesting using a pzt ceramic multilayer stack,”
structure, a larger external load leads to thicker and longer Smart Materials and Structures, vol. 22, no. 6, p. 065015, 2013.
frame. [6] S. Zhao and A. Erturk, “Deterministic and band-limited stochastic
By the FEA simulations, it is observed that four design energy harvesting from uniaxial excitation of a multilayer piezoelectric
stack,” Sensors and Actuators A: Physical, vol. 214, pp. 58–65, 2014.
variables influence the performance of beam structure. For [7] J. Feenstra, J. Granstrom, and H. Sodano, “Energy harvesting through
example, increasing the width of frame beam leads to an a backpack employing a mechanically amplified piezoelectric stack,”
increase of the safety factor. Multi-parameter optimization is Mechanical Systems and Signal Processing, vol. 22, no. 3, pp. 721–
734, 2008.
well-suited in this situation as parameters are related with each [8] W. Zhou and L. Zuo, “A novel piezoelectric multilayer stack energy
other. It creates more room to enhance the force amplification harvester with force amplification,” in Proc. of ASME 2013 Interna-
ratio as well as reducing the physical size. tional Design Engineering Technical Conferences and Computers and
Information in Engineering Conference, 2013, paper no. DETC2013-
Concerning the application of the designed frame, in com- 13611.
parison with the single force amplifier, the two-stage force [9] W. Chen, Y. Wang, and W. Deng, “Deformable force amplification frame
amplifier saves more space and has a larger force amplification promoting piezoelectric stack energy harvesting: Parametric model, ex-
periments and energy analysis,” Journal of Intelligent Material Systems
ratio, which is more suitable for practical application of and Structures, vol. 28, no. 7, pp. 827–836, 2017.
energy harvesting from footstep. It takes full advantage of the [10] L. Wang, S. Chen, W. Zhou, T.-B. Xu, and L. Zuo, “Piezoelectric
characteristics of material and structure to achieve 26 times vibration energy harvester with two-stage force amplification,” Journal
of Intelligent Material Systems and Structures, vol. 28, no. 9, pp. 1175–
force output. Inside a room with finite area, it reduces the 1187, 2017.
number of piezoelectric stack and thereby reduces the cost [11] Q. Xu and Y. Li, “Analytical modeling, optimization and testing of a
and improves the stability. Assume that nine two-stage energy compound bridge-type compliant displacement amplifier,” Mechanism
and Machine Theory, vol. 46, no. 2, pp. 183–200, 2011.
harvesters are arranged together to form a square area of 150
[12] Q. Xu, “Design and development of a flexure-based dual-stage nanoposi-
× 150 mm2 . The safety factor of each harvester is over 2. tioning system with minimum interference behavior,” IEEE Transactions
The array of harvester could bear over 1800 N force which on Automation Science and Engineering, vol. 9, no. 3, pp. 554–563,
is about 180 kg. The structure would not be damaged when 2012.
[13] M. Ling, J. Cao, Z. Jiang, and J. Lin, “Theoretical modeling of attenuated
three people of 60 kg weight stand on it. displacement amplification for multistage compliant mechanism and its
In the future work, a prototype will be fabricated for exper- application,” Sensors and Actuators A: Physical, vol. 249, pp. 15–22,
imental testing. Suitable electrical circuit will be designed to 2016.

495