theoretical and experimental investigation of magnetostrictive tagged composite beams
TRANSCRIPT
Theoretical and Experimental Investigation of Magnetostrictive Tagged Composite Beams
by
Florin G. Jichi
Bachelor of Science Technical University of Timisoara, Romania, 1995
__________________________________________________
Submitted in Partial Fulfillment of the
Requirements for the Degree of Master of Science in the
Department of Mechanical Engineering
College of Engineering & Information Technology
University of South Carolina
2000
_____________________________________ _____________________________________ Department of Mechanical Engineering Department of Mechanical Engineering Director of Thesis Second Reader
________________________ Dean of the Graduate School
ACKNOWLEDGEMENTS
As any research endeavor cannot be done alone, but more likely is a team effort,
there are many members involved. Therefore, I would like to express my gratitude and
deeply thanks to Dr. Victor Giurgiutiu for all his guidance and help throughout my entire
time spent here at USC. Without his support, none of what I accomplished would not
have been possible.
In addition, I would like to sincerely thank Dr Abdel-Moez E. Bayoumi for his time
from his busy schedule. My gratitude also goes to, Adrian Cuc, Paulette Goodman,
Jingjing “Jack” Bao, Greg Nall, Radu Pomirleanu, and Andrei Zagrai, my colleagues,
who made my time spent here so pleasant and fun.
I would like to express my forever gratitude and thanks to my mother, father, sister
and brother-in-law for their trust, incommensurable support, love and encouragement.
ABSTRACT
Among novel non-destructive evaluation techniques for structural health
monitoring, the magnetostrictive-tagged fiber-reinforced composites stand out as
especially suitable due to: (a) distributed sensory properties; (b) non-contact damage
detection; and (c) straight forward manufacturing implementation. Experimental data and
mathematical modeling of a magnetostrictive-tagged fiber reinforced composite specimen
under bending (flexural) loading are presented. A brief review of the state of the art
identifies previous work on axially loaded magnetostrictive composites, but finds no
previous work on bending. Description of bending specimen design and fabrication is
followed by the theoretical analysis and by the description of the experimental set-up and
equipment used. Several analysis models were used. Test data, with and without
magnetic annealing between loading cycles, is presented and results are discussed.
Numerical values for the stress and strain versus magnetic flux density coefficients are
given for both annealed and non-annealed cases. Piezomagnetic coefficients for the
magnetostrictive composite are calculated. The correlation between the results developed
in the presented paper for bending and previously published results for axial loading is
found to be within 10% after correction factors depending on the quantity of the
magnetostrictive material are applied. In conclusion, the usefulness of this method for
structural health monitoring and further work are discussed.
TABLE OF CONTENTS
ACKNOWLEDGEMENTS .........................................................................................................................II
ABSTRACT ................................................................................................................................................ III
TABLE OF CONTENTS ........................................................................................................................... IV
LIST OF FIGURES.................................................................................................................................... VI
LIST OF TABLES.....................................................................................................................................XII
1 INTRODUCTION................................................................................................................................1
1.1 PREVIOUS WORK ON MAGNETOSTRICTIVE TAGGED COMPOSITES FOR STRUCTURAL HEALTH
MONITORING (TENSILE EXPERIMENTS) .............................................................................................................1
1.2 CHARACTERISTICS OF ETREMA’S TERFENOL-D MAGNETOSTRICTIVE MATERIAL.................................5
1.3 PRESENT INVESTIGATION ...................................................................................................................7
2 ANALYSIS OF A MAGNETOSTRICTIVE COMPOSITE BEAM ...............................................8
2.1 STATIC ANALYSIS OF SIMPLY SUPPORTED COMPOSITE BEAM UNDER CENTRAL LOAD........................9
2.2 ANALYSIS OF A MAGNETOSTRICTIVE LAMINATED COMPOSITE ........................................................10
2.2.1 Micromechanics analysis of the MS tagged composite specimen ...........................10
2.2.2 Lamination Analysis ................................................................................................13
2.3 7-PLY BALANCED-ORTHOTROPIC MODEL..........................................................................................17
2.4 14-PLY CROSS-PLY ANGLE MODEL ...................................................................................................20
2.5 28-PLY CROSS-PLY ANGLE MODEL ...................................................................................................23
2.6 56-PLY CROSS-PLY ANGLE MODEL ...................................................................................................25
2.7 CONVERGENCE ANALYSIS ...............................................................................................................25
2.7.1 Convergence of the strain........................................................................................26
2.7.2 Convergence of stress..............................................................................................30
3 MAGNETOSTRICTIVE COMPOSITE BEAM EXPERIMENT.................................................39
3.1 SPECIMEN DIMENSIONAL DESIGN .....................................................................................................39
3.2 DESCRIPTION OF THE MS TAGGED COMPOSITE SPECIMEN .................................................................43
3.3 SPECIMEN PREPARATION FOR THE EXPERIMENT...............................................................................44
3.4 EXPERIMENTAL DESIGN ...................................................................................................................45
3.4.1 Description of the experiment .................................................................................45
3.4.1.1 Clamping fixture design ......................................................................................................48
3.4.1.2 Protective wood fixture design for the strengthen the gaussmeter probe.............................48
3.4.2 List of the equipment used .......................................................................................49
3.4.3 Equipment Calibration ............................................................................................50
3.4.4 LabView Virtual Instrument Environment...............................................................51
4 EXPERIMENTAL PROCEDURE AND RESULTS ......................................................................54
4.1 MAGNETIC ANNEALING....................................................................................................................54
4.2 TESTING PROCEDURE FOR MS TAGGED COMPOSITE BENDING EXPERIMENT.......................................54
4.3 RESULTS WITHOUT MAGNETIC ANNEALING......................................................................................55
4.3.1 Displacement and strain without magnetic annealing ............................................55
4.3.2 Magnetic flux density without magnetic annealing .................................................59
4.4 RESULTS WITH MAGNETIC ANNEALING BETWEEN LOADING-UNLOADING CYCLES............................63
4.4.1 Displacement and strain with magnetic annealing between loading-unloading
cycles .................................................................................................................................63
4.4.2 Magnetic flux density with magnetic annealing between loading-unloading cycles66
4.5 MAGNETIC FLUX DENSITY ANALYSIS ...............................................................................................70
4.6 PIEZOMAGNETIC STRESS AND STRAIN COEFFICIENTS ........................................................................72
4.6.1 Piezomagnetic stress coefficient, e31 without magnetic annealing ..........................73
4.6.2 Piezomagnetic stress coefficient, e31, with magnetic annealing...............................74
4.6.3 Piezomagnetic stress coefficient ,e31 ,using experimental magnetic flux density
without magnetic annealing and the 56-ply strain results model.........................................................75
4.6.4 Piezomagnetic stress coefficient, e31 using experimental magnetic flux density with
magnetic annealing between loading-unloading cycles and the 56-ply strain results model ..............77
4.6.5 Piezomagnetic strain coefficient d31 using experimental magnetic flux density
without magnetic annealing and 56-ply stress model ..........................................................................78
4.6.6 Piezomagnetic strain coefficient ,d31 ,for annealed specimen experiment and 56-ply
stress model .................................................................................................................................79
5 ANALYSIS AND DISCUSSION ......................................................................................................80
5.1 COMPARISON OF PIEZOMAGNETIC COEFFICIENTS FOR MAGNETOSTRICTIVE TAGGED COMPOSITES
WITH PREVIOUSLY PUBLISHED DATA ..............................................................................................................80
5.2 COMPARISON BETWEEN EXPERIMENTAL, DESIGN AND MODEL STRAIN FOR WITH AND WITHOUT
MAGNETIC ANNEALING...................................................................................................................................83
5.3 COMPARISON BETWEEN EXPERIMENTAL AND MODEL PIEZOMAGNETIC STRESS COEFFICIENT
WITHOUT MAGNETIC ANNEALING ...................................................................................................................84
5.4 COMPARISON BETWEEN EXPERIMENTAL AND MODEL PIEZOMAGNETIC STRESS COEFFICIENT WITH
MAGNETIC ANNEALING...................................................................................................................................84
6 CONCLUSIONS ................................................................................................................................86
7 BIBLIOGRAPHY ..............................................................................................................................89
8 APPENDIX.........................................................................................................................................91
LIST OF FIGURES
Figure 1 Magnetic flux in axial and thickness direction during axial loading of
magnetostrictive tagged composite specimens (White, 1999).................................... 3
Figure 2 Transverse (x-axis) magnetic flux density versus load: (a) neat resin;........... 4
Figure 3 Stress/magnetic flux density cyclic testing results: (a) without annealing
between cycles; (b) with annealing between cycles (Quattrone, Berman, and White,
1998) ......................................................................................................................... 5
Figure 4 Strain versus magnetic strength (Butler, 1988)............................................... 6
Figure 5 Relationships between a) magnetic flux density versus magnetic strength, b)
stress versus magnetic flux density, c) stress versus magnetic strength ..................... 6
Figure 6 Schematic of the static system ........................................................................ 9
Figure 7 Schematic of a generic lay-up....................................................................... 14
Figure 8 Balanced-orthotropic lay-up model used in the analysis .............................. 17
Figure 9 Cross-ply model used in the analysis............................................................ 20
Figure 10 Strain distribution of the 7, 14, 28 model with 0°/90° degree ply angle in
longitudinal direction................................................................................................ 26
Figure 11 Convergence of the strain result of the models to the experimental data ..... 28
Figure 12 Strain distribution for the 7, 14, 28 mathematical models with 0°/90° degree
ply angle in transversal direction. ............................................................................. 29
Figure 13 Stress – thickness distribution of the 7-ply balanced-orthotropic model with 0
degree fiber angle...................................................................................................... 30
Figure 14 Stress – thickness distribution in longitudinal direction for the 14-ply 0°/90°
degree fiber angle model........................................................................................... 31
Figure 15 Thickness distributions of longitudinal stress for 28-ply 0°/90° degree
fiber angle model. ..................................................................................................... 32
Figure 16 Thickness distributions of longitudinal stress for 56-ply 0°/90° degree
fiber angle model. ..................................................................................................... 32
Figure 17 Stress-thickness distribution for the 7, 14, 28-ply models with 0°/90°
degree fiber angle...................................................................................................... 33
Figure 18 Convergence of results on the specimen surface under maximum load
condition (Fmax = 58.9 N, Mmax = -88.29 Nm/m) as predicted by various models and
measured by experiment for stress results ................................................................ 35
Figure 19 Stress – thickness distribution in longitudinal direction for 14-ply 90°/0°
degree fiber angle model........................................................................................... 36
Figure 20 Stress - thickness distribution in longitudinal direction for the 28-ply
90°/0° degree fiber angle model ............................................................................... 36
Figure 21 Stress – thickness distribution in longitudinal direction for 56-ply 90°/0°
degree fiber angle model........................................................................................... 37
Figure 22 Stress-Thickness Distribution for the 7 ply 0 degree angle model and 14, 28-
ply models with 90°/0° degree fiber angle................................................................ 38
Figure 23 General view of the experiment set-up ......................................................... 45
Figure 24 Schematic of the experiment and data flow.................................................. 46
Figure 25 Position of the gaussmeter probe, strain gage, and displacement transducer47
Figure 26 Concept design of the protective wood fixture ............................................. 48
Figure 27 Overview of the assembly of magnetic sensor and carved wood ................. 49
Figure 28 Front panel of the program created for the MS composite beam
experiment using the LabView virtual instrument.................................................... 52
Figure 29 The logic flow of the program created for the MS composite beam
experiment using the LabView virtual instrument.................................................... 53
Figure 30 Mechanical data resulting from the experiments without magnetic
annealing between loading cycles: displacement vs. load ........................................ 56
Figure 31 Mechanical data resulting from the experiments without magnetic
annealing between loading cycles: strain vs. load .................................................... 58
Figure 32 Mechanical data resulting from the experiments without magnetic
annealing: strain vs. displacement ............................................................................ 58
Figure 33 Magnetic flux density responses, without magnetic annealing, to bending
strain in top surface of MS-tagged composite specimens: superposed 10 consecutive
cycles ................................................................................................................... 60
Figure 34 Mean value and standard deviation of the magnetic field ............................ 61
Figure 35 The resulted magnetic field as function of measured strain ......................... 61
Figure 36 Mean values of magnetic flux density and standard deviation versus strain
for the specimen without annealing .......................................................................... 63
Figure 37 Displacement results versus applied force for annealed specimen............... 65
Figure 38 Strain results for the MS composite specimen with magnetic annealing
between cycles .......................................................................................................... 66
Figure 39 Magnetic flux density results for specimen with magnetic annealing
between cycles .......................................................................................................... 68
Figure 40 Mean value and standard deviation of magnetic flux density results vs. load
for the specimen with magnetic annealing between cycles ...................................... 68
Figure 41 Experimental magnetic flux density results versus strain with magnetic
annealing of the specimen between cycles ............................................................... 69
Figure 42 Mean value and standard deviation of magnetic flux density results vs. strain
for the specimen with magnetic annealing between cycles ...................................... 69
Figure 43 Magnetic flux density in relationship with model longitudinal stresses on the
surface of the composite specimen model ................................................................ 70
Figure 44 Standard deviation of the stress from the mean value versus magnetic flux
density ................................................................................................................... 71
Figure 45 Magnetic flux density versus strain on the top surface of the model
material considered and 0°/90° degree fiber angle without magnetic annealing...... 71
Figure 46 Plot of the average magnetic flux density vs. strain showing the trend lines in
determination of the piezomagnetic stress coefficient e31 for non -annealed specimen
case 74
Figure 47 Plot of the average magnetic flux density vs. strain showing the trend lines in
determination of the piezomagnetic stress coefficient e31 of magnetic annealing of
the specimen.............................................................................................................. 75
Figure 48 Plot of the average magnetic flux density with no magnetic annealing
specimen vs. 56-ply model strain showing the trend lines in determination of the
piezomagnetic stress coefficient e31.......................................................................... 76
Figure 49 Plot of the average magnetic flux density with magnetic annealing vs. 56-ply
model strain showing the trend lines in determination of the piezomagnetic stress
coefficient e31 ............................................................................................................ 77
Figure 50 Trend line for calculating the piezomagnetic strain coefficient.................... 78
Figure 51 Trend line for computing the piezomagnetic strain coefficient d31 for the case
of magnetic annealing of the specimen between cycles ........................................... 79
LIST OF TABLES
Table 1 Dimensions of the MS tagged composite specimen ....................................... 9
Table 2 Mechanical properties considered for the magnetostrictive-tagged woven
composite material specimen.................................................................................... 11
Table 3 Lay-up of the composite models for 14, 28 and 56-ply models ................... 21
Table 4 Comparison of the longitudinal strain on the specimen surface under
maximum load condition (Fmax = 58.9 N, Mmax = - 88.29 Nm/m) as predicted by
various models and measured by experiment ........................................................... 27
Table 5 Stress results on the surface for 7 balanced-orthotropic, 14, 28-ply cross-ply
models 34
Table 6 Stress results on the surface for 7, 14, and 28 ply 0 and 90°/0° cross-ply
models 38
Table 7 List of the equipment used in the magnetostrictive composite beam
experiment................................................................................................................. 50
Table 8 The displacement data without magnetic annealing ..................................... 56
Table 9 Strain data without magnetic annealing........................................................ 57
Table 10 Magnetic flux density results without magnetic annealing........................... 59
Table 11 Displacement results with annealing between loading-unloading cycles..... 64
Table 12 Strain results with annealing between loading-unloading cycles ................. 65
Table 13 Experimental magnetic flux density with magnetic annealing between
loading-unloading cycles .......................................................................................... 67
Table 14 Summary of magnetostrictive coefficient for MS tagged composites, as
determined in the present work and by previous investigators................................. 81
Table 15 Comparison of the design, model and measured strain ................................ 83
1 INTRODUCTION
In recent years, numerous applications in civil engineering construction used
composite materials. The increase of the usage of the composite materials in civil
construction imposed problems of evaluating the in-service composite civil engineering
structures. The evaluation of the composite is a wide area in which engineers and
researchers have proposed several technologies, for in-service Non-destructive evaluation
(NDE). Conventional NDE methods, initially developed for metallic structures, have
been shown to be less effective in monitoring composite structures due to the
micromechanical complexity of the composite material. New NDE technologies are
required. The NDE technology analyzed in this paper is proposed for the inspection of
advanced composite by using the magnetostrictive particle tagging technique.
Magnetostrictive-tagged composites permit: (a) distributed sensory properties; (b) non-
contact damage detection; and (c) straight forward manufacturing implementation.
1.1 PREVIOUS WORK ON MAGNETOSTRICTIVE TAGGED COMPOSITES FOR
STRUCTURAL HEALTH MONITORING (TENSILE EXPERIMENTS)
Terfenol-D is a magnetic anisotropy-compensated alloy TbxDy1-xFe2 that
shows a strong magnetostrictive behavior. The name Terfenol-D represents the
composition of the material and the original name of the Navy Laboratory at which the
work begun. Ter represents Terbium, Fe from iron, nol Naval Ordnance Laboratory and
D Dysprosium. A magnetostrictive (MS) material, such as Terfenol-D, produces
magnetic field when subjected to mechanical strain. This phenomenon is known as the
“converse magnetostrictive effect”. White and collaborators (White, Albers and
Quattrone, 1996; White and Brouwers, 1998) did extensive work on magnetostrictive-
tagged composites under axial loading. White (1999) reviewed the magnetostrictive
tagging methodology of composites for structural health monitoring and gave an update
on recent results. An experiment in which MS-tagged composite specimens were
subjected to uniaxial tension in a testing machine was presented. Neat resin specimens
tagged 2.24% by volume with magnetostrictive Terfenol-D powder were used. The
resulting magnetic field was measured in both the axial and thickness directions (Figure
1). Trovillion et al. (1999) studied the magnetic characteristics of neat resin and glass-
fiber-reinforced magnetostrictive composites subjected to axial load. The fiber reinforced
polymer composite (FRP) specimens consisted of 4 layers of continuous strand glass mat
fibers embedded in a polyester resin. Trovillion et al. (1999) studied the magnetic
characteristics of neat resin and glass-fiber reinforced magnetostrictive composites
subjected to axial load. The fiber reinforced polymer composite (FRP) specimens
consisted of 4 layers of continuous strand glass mat fibers embedded in a polyester resin.
Figure 1 Magnetic flux in axial and thickness direction during axial loading of
magnetostrictive tagged composite specimens (White, 1999)
The top lamina of the composite was impregnated with Terfenol-D powder at a
volume fraction of 2.24% for that lamina. The specimens were subjected to uniaxial
loading under load control at a rate of 0.02 kN/s.
Hall-effect device were used to measure the magnetic field response to subsequent
loading and unloading. As seen in Figure 2, both measuring devices gave similar results.
Magnetic annealing (i.e. rearranging the magnetic dipoles chains of the magnetostrictive
molecules through application of strong magnetic field) was performed by applying a
magnetic field through the thickness of the specimen using a pair of 800 Gauss
permanent magnets.
Nersessian N. and Carman G.P. presented at ASME conference in Orlando results
of tests for five different volume fraction composites namely 10 %, 20%, 30 %, 40 % and
50 %. The composite were tested under constant magnetic field with varying the
mechanical load and constant mechanical load with varying magnetic field conditions.
Results for the constant magnetic filed test indicated that modulus generally increases
with volume fraction and increasing H/Hmax. For low fields, initial dip is noticed in
modulus attributed to domains becoming more mobile at lower magnetic field levels.
Results presented for the constant load test show a strong dependence of strain output on
applied pre-stress.
(a)
0.0 0.2 0.4 0.6 0.8 1.00.0
0.5
1.0
1.5
2.0
2.5 Gauss Probe Hall Effect Chip
Mag
netic
Flu
x D
ensi
ty (
gaus
s)
Load (kN)
(b)
0.0 0.5 1.0 1.5 2.00.00
0.25
0.50
0.75
Ma
gn
etic
Flu
x D
en
sity
(g
auss
)
Load (kN)
Gauss Probe Hall Effect Chip
Figure 2 Transverse (x-axis) magnetic flux density versus load: (a) neat resin;
(b) composite sample (Trovillion et al, 1999)
Quattrone, Berman, and White (1998) studied the magnetic response repeatability
of MS tagged composites under cyclic loading. Terfenol-D active-tagged composites
were subjected to uniaxial tension and the magnetic response in the axial direction under
repeated loading and unloading was measured. Two types of experiments were
performed: without magnetic annealing between loading cycles (Figure 3a) and with
annealing (Figure 3b).
(a)
Slope 45/10 mG/MPa
(b)
Slope 140/10 mG/MPa
Figure 3 Stress/magnetic flux density cyclic testing results: (a) without annealing
between cycles; (b) with annealing between cycles (Quattrone, Berman, and
White, 1998)
Krishnamurthy, Anjanappa, and Wang (1999) consider health-monitoring detection
of delaminations in composite materials using an excitation coil and a sensing coil. The
open-circuit voltage induced in the sensing coil is proportional to the stress generated in
the magnetostrictive layer by the presence of the delamination.
1.2 CHARACTERISTICS OF ETREMA’S TERFENOL-D MAGNETOSTRICTIVE
MATERIAL
An application manual for designing the magnetostrictive transducers using Etrema
Terfenol-D was prepared by J. Butler (1988). A physical description of the
magnetostrictive material, with emphasis on Etrema Terfenol-D, was given. The nominal
Etrema Terfenol-D strain versus magnetic strength, H, as illustrated in this manual, is
shown in Figure 4.
Figure 4 Strain versus magnetic strength (Butler, 1988)
The theoretical stress versus magnetic flux density, stress versus magnetic strength
and magnetic flux density versus magnetic strength curves are presented in Figure 5.
Butler (1988) also provides an analysis of the properties of Etrema Terfenol-D with
emphasis on strain, stress and magnetic field relationships. Some design considerations
and fundamental concepts for designing magnetostrictive transducers are also presented.
Figure 5 Relationships between a) magnetic flux density versus magnetic strength, b)
stress versus magnetic flux density, c) stress versus magnetic strength
1.3 PRESENT INVESTIGATION
In the present investigation, studies of the theoretical and experimental of stress-
strain and magnetic flux density response of a woven tagged composite subjected to
bending loading were performed. Several two-dimensional simulation models for strain
and stress prediction in the composite were investigated. The analysis was performed
under two separate assumptions regarding woven composites modeling: (a) balanced-
orthotropic equivalent and (b) cross-ply equivalent. The experimental investigation
considers the bending moment created by the loading and unloading of a simply
supported MS-tagged composite beam using several weights. At each loading and
unloading step, the magnetic flux density induced by the converse magnetostrictive effect
was measured. Comparison between the experimental and analysis results are then
presented. Based on the comparison of experimental and analysis data, piezomagnetic
stress and strain coefficients under bending are calculated. Comparison of these
coefficients with the results obtained by other investigators for axial loading is
performed. It is shown that, although exact comparison is not fully possible since bending
and axial stress distributions are essentially different, the results are similar. The
piezomagnetic stress and strain coefficients determined in this paper for MS tagged
composite under bending can be used in industry as reference data for design of novel
NDE devices based in the magnetostrictive effect in MS tagged composites.
2 ANALYSIS OF A MAGNETOSTRICTIVE COMPOSITE BEAM
The analytical work performed on the MS tagged woven composite specimen
consisted of: (a) loading analysis; (b) micromechanical analysis; (c) lamination analysis.
One balanced-orthotropic and several cross-ply lamination models were used. Converge
analysis of the strain and stress predictions versus number of plies in the lamination
model were also performed.
A magnetostrictive composite stress-strain model has been developed for
simulating the behavior of the woven magnetostrictive composite. Two dimensional
mathematical model study was performed for the 7 ply balanced-orthotropic model, and
14, 28 and 56-ply multidimensional composite model with 0°/90° and 90°/0° degree.
Influence of the Terfenol-D was considered for the first and last ply of the 7-ply model,
in first 2 plies and last 2 plies for the 14-ply model, for the first 4 plies and last 4 plies of
the 28-ply model and the 8 plies for the 56-ply model.
2.1 STATIC ANALYSIS OF SIMPLY SUPPORTED COMPOSITE BEAM UNDER
CENTRAL LOAD
To determine the bending moment generated by the centrally-placed load, a static
analysis of simply supported composite beam was performed. In this analysis, the
dimensions and mechanical properties of the woven composite presented in Table 1 were
used. A schematic of the system and how the load is applied is shown in Figure 6. The
loading and unloading of the specimen was assumed to be performed using 2-kg weights.
Table 1 Dimensions of the MS tagged composite specimen
Characteristics Dimension [mm]
Span, L 600
Width, w 100
Thickness, t 6.5
Thickness of the lamina 0.92857
Composite Material Plate
Strain Gages
Force given by the bricks
2-Kg bricks
L
w
Figure 6 Schematic of the static system
The force exerted by the one brick is calculated considering the weight of 2 kg and
gravitational acceleration of 9.81 m/s2. Thus the force of the brick is
F = 2.09.81 = 19.62 N (1)
The moment per unit width exerted by the weights is:
F LM
4 w
(2)
Where L is the actual length of the beam, w- the width of the beam, F is the forced
applied by one weight. During the loading cycle, up to three weights were sequentially
put on the specimen and then removed. The maximum force value, corresponding to
three weights, was Fmax = 58.9 N. The corresponding moment per unit width was Mmax =
88.29 Nm/m.
2.2 ANALYSIS OF A MAGNETOSTRICTIVE LAMINATED COMPOSITE
2.2.1 Micromechanics analysis of the MS tagged composite specimen
Since a woven layer has fibers in two directions, warp and fill, the analysis of a
woven composite layer cannot be directly performed through the Classical Lamination
Theory (CLT), which assumes fibers aligned with just one direction. An equivalence
principle needs to be applied. Tsai (1992) suggested that the predictions of elastic
constants and strength of woven composite could be made using classical micro and
macro mechanics with appropriate empirical correction factors (Tsai, 1992, page 7-14).
One approach is to replace a woven composite layer with two equivalent conventional
layers representing the weave and the warp of the original fabric. The micro-mechanics
stiffness and strength formulas are then applied to the equivalent plies. One shortcoming
of this approach is that the order of 0 and 90 plies may strongly affect the results.
Another approach is the replace the balanced woven composite layer by an orthotropic
layer with averaged properties. For a woven composite having fibers aligned with the
loading axes, this hypothesis yields balanced-orthotropic behavior. In our analysis, both
approaches were taken. In either case, a conventional micromechanics analysis (Jones,
1999) was first applied to determine the basic MS composite properties from the
properties of its constituents, given in Table 2.
Table 2 Mechanical properties considered for the magnetostrictive-tagged woven
composite material specimen
Mechanical properties Dimension Young’s modulus of the fiber, Ef
1 72.4 GPa
Young’s modulus of the matrix, Em1 3.25 GPa
Young’s modulus for the Terfenol-D, EMS2 30 GPa
Poisson ratio of the fiber, f1 0.2
Poisson ratio of the matrix, m1 0.3
Fiber weight fraction, wf 0.3
Fiber density, f1 2.54 g/ml
Matrix density, m1 1.18 g/ml
Terfenol-D density, MS3 9.25·103 kg/m3
Thermal expansion coefficient for the fiber, αf1 5·10-6 m/m per ºC
Thermal expansion coefficient for the matrix, αm1 3.0·10-6 m/m per ºC
Terfenol-D weight fraction 15
Terfenol-D particle size4 38-40 μm
Note: 1Malik, 1992; 2Butler, 1988; 3De Laicheisserie, 1993, Trovillion et al., 1999.
The fiber volume fraction vf is calculated as (Malik, 1992, page 81, equation 2.5):
f
ff
f f
f m
w
vw 1 w
(3)
For the values given in Table 2, the fiber volume fraction values yield vf = 0.16.
Since the sum of the volume fractions is equal to 1, the volume fraction for the
matrix is vm= 1-vf for the plies which does not include the Terfenol-D, and vm = 1- vf -
vMS for the plies that contain MS material. The basic moduli of elasticity in the material
axes were calculated following (Jones) 1999 as:
1 f f m m MS MS
1 1 1 12 f f m m MS MS
E E v E v E v
E [(E ) v (E ) v (E ) v ]
(4)
The ply Poisson’s ratio was calculated as:
12 f f m m MS MS
221 12
1
v v v
E
E
(5)
Shear modulus for the fiber, matrix and Terfenol-D for each ply were defined as:
f
ff
EG
2 1
,
mm
m
EG
2 1
,
MSMS
MS
EG
2 1
(6)
Following Jones (1999, page 134), we calculate the shear modulus for the plies as:
f m MS12
f m MS m _ MS f MS MS f m
G G GG
v G G v G G v G G
(7)
Thermal expansion coefficients for the ply were:
f1 f f m1 m m MS MS MS1
f f m m MS MS
E v E v E v
E v E v E v
(8)
2 f f1 f m m1 m MS MS MS 1 121 v 1 v 1 v (9)
and
1
2
[ ]
(10)
2.2.2 Lamination Analysis
The elastic behavior of multidirectional plies can be described in terms of the
stiffness matrix, the compliance matrix, (Jones, 1975). The goal is to determine the
strain-stress distribution in the material. The plane stress stiffness matrix [Q] of a
orthotropic composite ply is computed as:
1 21 1
12 21 12 21
12 2 2
12 21 12 21
12
E E0
1 1
E E[Q] 0
1 1
0 0 G
(11)
Where E1 and E2 are given by Equation (4). The composite is assumed to consist of a lay-
up of multiple plies. The lay-up schematic for generic composite is presented in Figure 7
where x-y are the loading axes. In our case, the angle of the plies is either 0 or 90-degree.
y
x
z
θ 1
θ k
θ n
Figure 7 Schematic of a generic lay-up
Considering the transformation matrix:
2 2
2 2
2 2
cos sin 2 cos sin
[T] sin cos 2 cos sin
cos sin cos sin cos sin
(12)
The stiffness matrix in loading axes [Q]is:
T1 1[Q] [T] [Q] [T] (13)
Knowing the stiffness matrix, one calculates the extensional stiffness matrix [A],
coupling stiffness matrix [B], bending stiffness matrix [D] for the composite laminate:
k 1 k kk
2 2k k 1k 1
k
3 3k k 1k 1
k
[A] [Q] (z z )
1[B] [ [Q] (z z )]
2
1[D] [ [Q] (z z )]
3
(14)
The variable zk represents the distance from the midplane to the bottom of the kth
lamina; zk-1 represents the distance from the midplane to the top of the kth lamina. These
distances were calculate using the formula:
layerk
Nz h (k )
2
(15)
In matrix form the general equation is:
0N A B
M B D
(16)
Where {N} is the load vector, {M} is the moment vector, {ε0} is the mid-surface strain
vector, and {κ} is the curvature vector. For the design purposes, knowing the loading, the
state of deformation general solution is:
10 A B N
B D M
(17)
Next, the effect of thermal conditions is analyzed by considering only the thermal effect,
i.e. without loading F=0. The loads derived from the above-mentioned conditions are:
layerN
k 1T k 1 k k 1 k 1k 1
[N] [Q] (z z ) T
,
layerN 2 2k 1
k 1T k 1k 1
z z[M] [Q] ( ) T
2
(18)
Here, the term k 1 represents the coefficient of thermal expansion in the loading axes,
which is calculated using the transformation matrix and the coefficients of thermal
expansion in the material axes:
T
k 1k 1 k 1{ } [T] { } (19)
The loads and moments due to thermal effect:
layerN
k 1T k 1 k k 1 k 1k 1
{N} [Q] { } (z z ) T
(20)
layerN 2 2k k 1
k 1T k 1 k 1k 1
z z{M} [Q] { } ( ) T
2
(21)
The load and moment vector are composed as:
T
T
N N N
M M M
(22)
A general loading vector it is defined as:
0
1
2
0
1
2
N
N
N{NM}
M
M
M
(23)
Solving the state of deformation Equation (17) yields the loading axes strain on the top
and bottom of each ply is calculated.
topk 1 0 k 1
0 kbotk 1
{ } { } z { }
{ } { } z { }
(24)
Using the strains in loading axes, the strains in material axes are calculated for the top
and the bottom ply:
T1
k 1{ } [T] { } (25)
Consequently, the stresses in the longitudinal (L) and transversal (T) directions at the top
and the bottom of each ply are:
k 1top k 1 topk 1{ } [Q] { }
(26)
k 1 k 1bot k 1 bot{ } [Q] { }
(27)
2.3 7-PLY BALANCED-ORTHOTROPIC MODEL
X
Y
Z
Ply 1
Ply n
Ply k
Figure 8 Balanced-orthotropic lay-up model used in the analysis
The balanced-orthotropic model, presented in Figure 8, uses the assumption that a
balanced woven composite can be represented by a composite having averaged
properties. Using the longitudinal and transverse moduli for a uni-directional fiber
composite layer we calculated the averaged properties:
1 2 L T
1E E (E E )
2
(28)
The MS Terfenol-D volume fraction and Poisson’s ratio were
vMS = 2.24% and νMS = 0.3 for the plies with Terfenol-D, and zero for the other plies.
Two indexes were defined in order to make a distinction between the laminas with
and without Terfenol-D, Nlayer=7, Nlayers=6.. Performing the CLT analysis yields the
balanced-orthotropic (BO) matrices:
7 6
6 7BO
6
6.596 10 7.291 10 0
[A] 7.291 10 6.596 10 0
0 0 9.668 10
(N/m) (29)
12 13
13 13BO
3.638 10 2.27 10 0
[B] 2.274 10 9.09 10 0
0 0 0
(N) (30)
BO
244.47 25.897 0
[D] 25.897 190.99 0
0 0 34.346
(N·m) (31)
The thermal effect was included for complete analysis procedure but since the cure
temperature was the same with environmental temperature, k 1T 0 , then all the stresses
due to thermal effects are zero.
Considering only the bending moment, the load vector is zero and the moment
vector is:
vector
M
{M } 0
0
(Nm/m) (32)
The values for this loading vector are:
0
0
0 (N / m)NM
88.29 (N)
0
0
(33)
The strain and curvature vector is determined then using the general state of
deformation equation mentioned above. Thus the strain and curvatures are:
0
0
0
0
and
0.366
0.050
0
(1/m) (34)
The values for the stresses in longitudinal direction for the top and bottom layer
are:
L _ top
12.72
8.72
5.23
1.74
1.74
5.23
9.08
(MPa)
L _ bot
9.08
5.23
1.74
1.74
5.23
8.72
12.72
(MPa) (35)
The strains in the longitudinal direction for the top and bottom layer are:
top
1347
962
577
192
192
577
962
(με)
bot
962
577
192
192
577
962
1347
(με) (36)
2.4 14-PLY CROSS-PLY ANGLE MODEL
X
Y
Z
Ply 1
Ply n
Ply k
Figure 9 Cross-ply model used in the analysis
In the cross-ply analysis (Figure 9), each woven composite layer is approximated
by a couple of unidirectional cross-ply layers. The cross-ply layer can be either 0/90or
90/0. Since the choice between 0/90and 90/0 directly affects the accuracy of the
results on the specimen surfaces, we initially considered both cases in our analysis.
Additionally, the number of the layers was gradually increased from 14 to 28 and 56, and
a convergence study was performed. During this convergence study, as the number of
layers were doubled, the thickness of the layer was correspondingly halved, such that the
overall thickness of the specimen was maintained. The difference in 14, 28 and 56-ply
mathematical models procedure from the balanced-orthotropic model is that the Young’s
modulus of elasticity in longitudinal and transversal direction were not averaged The 28-
ply and 56-ply models were obtained by subsequent subdivision of the 0/90 layers and
application of symmetry principles. The ply angle distributions for the 14, 28 and 56-ply
models are presented in Table 3:
Table 3 Lay-up of the composite models for 14, 28 and 56-ply models
Number of Layers Lay-up
14 [(0/90)3/0]S
28 [(0/90)7/90]S
56 [(0/90)14]S
Terfenol-D is assumed present in the first two layers and last two layers and the for
the Young’s modulus values in the longitudinal and transversal directions are:
E1 = 14.96 GPa, E2 = 3.94 GPa. The Young’s modulus of elasticity values for the
layers that does not contain magnetostrictive material are: E1 = 14.31 GPa, E2 = 3.84
GPa. The A, B and D matrices for the 14-ply model were:
7 6 11
6 7 914
11 9 6
6.6 10 7.29 10 2.51 10
[A] 7.29 10 5.60 10 1.83 10
2.51 10 1.83 10 9.67 10
(N/m) (37)
12 13
13 1314
3.64 10 2.27 10 0
[B] 2.27 10 9.09 10 0
0 0 0
(N) (38)
14
244.47 25.9 0
[D] 25.9 190.99 0 (N m)
0 0 34.35
(39)
Stress and strain results in longitudinal direction for the 14-ply 0°/90° cross-ply
laminate, at the top and bottom of the plies, are:
L _ top
18.01
3.95
12.32
2.57
7.39
1.28
2.46
0.00
- 0.64
- 4.93
- 1.92
- 9.85
- 3.29
-15.44
(MPa),
L _ bot
15.44
3.29
9.85
1.92
4.93
0.64
0.00
- 2.46
- 1.28
- 7.39
- 2.57
-12.32
- 3.95
-18.01
(MPa),
top
1191
1021
851
680
510
340
170
0
-170
-340
-510
-680
-851
-1021
(με),
bot
1021
851
680
510
340
170
0
- 170
- 340
- 510
- 680
- 851
-1021
-1191
(με) (40)
For the 90°/0° cross-ply laminate the stress and strain results are:
L _ top
5.94
19.81
4.14
12.64
2.48
6.32
0.83
0.00
- 3.16
- 1.66
- 9.48
- 3.31
-16.51
- 5.10
(MPa),
L _ bot
5.10
16.51
3.31
9.48
1.66
3.16
0.00
- 0.83
- 6.32
- 2.48
-12.64
- 4.14
-19.81
- 5.94
(MPa),
top
1524
1307
1089
871
653
436
218
0
- 218
- 436
- 653
- 871
-1089
-1307
(με),
bot
1307
1089
871
653
436
218
0
- 218
- 436
- 653
- 871
-1089
-1307
-1524
(με) (41)
2.5 28-PLY CROSS-PLY ANGLE MODEL
The 28-ply model is computed in a similar manner with the 14-ply model. The changes
occur in the indices, the ply angle, and the thickness of the lamina. The A, B and D
matrices were:
7 6 11
6 7 914
11 9 6
6.1 10 7.29 10 1.38 10
[A] 7.29 10 6.1 10 3.18 10
1.38 10 2.18 10 8.97 10
(N/m) (42)
12 13 13
13 13 1314
13 13
3.64 10 1.25 10 1.05 10
[B] 2.27 10 9.09 10 1.05 10
1.05 10 1.05 10 101.23
(N) (43)
14
231.19 25.9 0
[D] 25.9 204.27 0 (N m)
0 0 30.59
(44)
The stress and strain results are presented next:
L _ top
19.06
4.53
16.33
3.83
13.03
3.06
10.42
2.38
7.82
1.70
5.21
1.02
2.61
0.34
0.00
- 1.30
- 0.68
- 3.91
- 1.36
- 6.52
- 2.04
- 9.12
- 2.72
-11.73
- 3.49
-14.97
- 4.18
-17.69
L _ bot
17.69
4.18
14.97
3.49
11.73
2.72
9.12
2.04
6.52
1.36
3.91
0.68
1.30
0.00
- 0.34
- 2.61
- 1.02
- 5.21
- 1.70
- 7.82
- 2.38
-10.42
- 3.06
-13.03
- 3.83
-16.33
- 4.53
-19.06
top
1259
1169
1079
989
899
809
719
629
539
449
259
269
179
89
0
-89
-179
-269
-359
-449
-539
-629
-719
-809
-899
-989
-1079
-1169
bot
1169
1079
980
899
809
719
629
539
449
359
269
179
89
0
-89
-179
-269
-359
-449
-539
-629
-719
-809
-899
-989
-1079
-1169
-1259
(45)
2.6 56-PLY CROSS-PLY ANGLE MODEL
The 56-ply model was obtained by subsequent subdivision of the 0 and 90 layers
and application of symmetry principles. The changes appear in setting the indexes and the
thickness of the lamina. The A, B and D matrices for the 56-ply model were:
7 6 11
6 7 956
11 9 6
6.22 10 7.29 10 2.82 10
[A] = 7.29 10 5.97 10 2.06 10 (N/m)
2.82 10 2.06 10 9.67 10
(46)
-13
-1356
13
-72.14 -3.98 10 0
[B] = -3.98 10 72.14 0 (N)
0 0 9.09 10
(47)
56
224.47 25.90 0
[D] = 25.90 211 0 (N m)
0 0 34.35
(48)
The maximum strain on the surface of the specimen model was found to be:
= 1296 (49)
The corresponding maximum stress was:
= 19.6 MPa (50)
2.7 CONVERGENCE ANALYSIS
The convergence of the strain and stress distribution for various models was
studied. Attention was focused on the longitudinal stress and strain under maximum load
conditions (Fmax = 58.9 N, Mmax = - 88.29 Nm/m). The strain values predicted on the top
and bottom surfaces were also compared with the experimental value measured during
the tests described in Chapter 4.
2.7.1 Convergence of the strain
Figure 10 shows the distribution of the longitudinal strain for the 7-ply balanced-
orthotropic model and the 14-ply and 28-ply 0/90 cross-ply models. (The 56-ply results
were not plotted on Figure 10 to avoid cluttering the drawing). The sign convention used
in Figure 10 is negative to the left and positive to the right.
14 plies 0/90 deg 28 plies 0/90 deg
7-ply balanced-orthotropic model
Figure 10 Strain distribution of the 7, 14, 28 model with 0°/90° degree ply angle in
longitudinal direction
The study shows that the largest strain value was predicted by the 7-ply balanced-
orthotropic model. This value is also very close to the experimental value (1347 με vs.
1333 με). The convergence of the strain distribution for the cross-ply models is apparent
as the models are becoming more refined i.e. strain distribution is getting closer to the
experimental data as the number of plies employed in the model is increased. The strain
values predicted by each model, at the surface of the composite, and comparison with
experimental data, are presented in Table 4.
Table 4 Comparison of the longitudinal strain on the specimen surface under
maximum load condition (Fmax = 58.9 N, Mmax = - 88.29 Nm/m) as predicted
by various models and measured by experiment
Strain [με]
7-ply balanced-orthotropic
model
14-ply 0/90 model
28-ply 0/90 model
56-ply 0/90 model
Experimental
1347 1191 1259 1296 1333
The reason for the difference between the balanced-orthotropic model and the other
models is that the balanced-orthotropic model considers each of the plies at the same
angle. Thus, the strain that could be carried is larger than the more realistic models in
which the strains are lower because the angle of ply is alternating from 0 to 90 degree.
The results show that the plies with transverse fiber to the longitudinal direction of
measuring (90 degree fiber angle) shrink and oppose the longitudinal stress direction. The
convergence of the strain data of the models is presented in Figure 11.
Str
ain
[m
icro
str
ain
]
1000
1055
1110
1165
1220
1275
1330
1385
Number of the plies
Strain response of the models for surface Experimental Baseline Strain
Strain response for quasi-isotropic model
7 28 5614
Figure 11 Convergence of the strain result of the models to the experimental data
The figure shows that the maximum strain given by the 7-ply balanced-orthotropic
model, is approximately the same as the experimentally obtained strain. The graph also
shows that the strain prediction by the cross-ply models improves with increasing number
of layers.
Balanced –orthotropic model
14-ply 0/90 deg model
28-ply 0/90 deg model
Figure 12 Strain distribution for the 7, 14, 28 mathematical models with 0°/90° degree
ply angle in transversal direction.
The strain distribution for each model in transversal direction is presented in Figure
12.The strains on the left side of the symmetry line represent positive strains and on the
right side represent negative strains. This case shows the strain distribution for the 14 and
28 ply models have close results. It is also observed that for the 7-ply model the strains
are larger.
2.7.2 Convergence of stress
For the 7-ply balanced-orthotropic model, the longitudinal stress distribution is
presented in the Figure 11. It can be noticed the stress distribution is rather linear, in
accordance with the balanced-orthotropic assumption. The slight discontinuity between
the outer first and last layers and the other layers is due to the outer layers being stiffer
due to presence of MS tagging in the modified layers, layer 0 and 6.
Stresses -Thickness Distribution in Longitudinal Direction for 7 plies
Figure 13 Stress – thickness distribution of the 7-ply balanced-orthotropic model with 0
degree fiber angle.
The stress distribution for the 14-ply 0°/90° cross-ply model is presented in Figure
15.
Stress - Thickness Distribution for 14 plies 0/90 degree fiber angle model
Figure 14 Stress – thickness distribution in longitudinal direction for the 14-ply 0°/90°
degree fiber angle model
In this model, the stress distribution is different then the stress distribution for the
7-ply balanced-orthotropic model.It can be noticed the stress in the layers with 90-degree
fiber angle is lower than the stress in the layers with 0-degree fiber angle. For the layers
that contain the magnetostrictive material an increased stress is noticed. The cause of this
is the increased stiffness due to Terfenol-D. Similar model behavior is observed for 28
ply and 56-ply 0°/90° cross-ply models as can be shown in Figure 15 and 16.
Stress - Thickness Distribution for 28 plies 0/90 degree fiber angle model
Figure 15 Thickness distributions of longitudinal stress for 28-ply 0°/90° degree fiber
angle model.
Stress - Thickness Distribution for 7, 14, 28, 56 plies models
Figure 16 Thickness distributions of longitudinal stress for 56-ply 0°/90° degree fiber
angle model.
The longitudinal stress distribution for the 14, 28-ply 0/90 cross-ply model is
presented in Figure 17. In the cross-ply models, the stress distribution is substantially
different then the stress distribution from the 7-ply balanced-orthotropic model. The
stresses in the layers with 90-degree fiber angle are substantially smaller. This produce
alternating changes from high stress to low stress as the layer stiffness change from the
high to low in accordance with the 0 and 90 orientations of the cross-ply laminate.
7 p lie s b a lan ce d -o rth o tro p ic m o d e l
1 4 p lie s 0 /9 0 d eg m o d e l
2 8 p lie s 0 /9 0 d eg m o d e l
Figure 17 Stress-thickness distribution for the 7, 14, 28-ply models with 0°/90° degree
fiber angle
The 56 ply 0°/90° degree fiber angle model is presented separately since the graph
would have been hard to visualize and the stress distribution for each model would have
been too hard to follow if they were superimposed on Figure 17. The results reveal that
the layers that contain Terfenol-D are stiffer than the other layers.
The 7 ply balanced-orthotropic model stress distribution shows that the model does
not accurately simulate the real woven composite material because larger stress values
were observed than in the cross-ply. This statement is made from a stress distribution
point of view. Depending on the fiber orientation and the stress amplitude, the stress
carried is consequently proportional. For the 0 degree fiber orientation the stress is
larger. For the fiber oriented to a 90-degree angle to the longitudinal direction, the stress
amplitude is small. However, this shows that even the transversal fiber carry load. For
analysis purposes, Table 4 presents the largest stresses on the surface for each model.
Table 5 Stress results on the surface for 7 balanced-orthotropic, 14, 28-ply cross-
ply models
7 ply balanced-orthotropic model
[MPa]
14 ply 0°/90° model [MPa]
28 ply 0°/90° model [MPa]
56 ply 0°/90° model [MPa]
12.72 18.01 19.06 19.63
Table 5 shows that the longitudinal stresses obtained by the 14, 28, and 56-ply
models are larger than the stresses obtained with the balanced-orthotropic model. These
results are due to increased stiffness in the longitudinal direction and less stiffness in
transverse direction typical of cross-ply models. In the case of the balanced-orthotropic
model the stiffness in both direction are the same, but of lower value than the longitudinal
cross-ply stiffness..
Figure 18 presents the results of Table 5 in graphical form. It can be appreciated
that the balanced-orthotropic model grossly underestimates the stresses, whereas the
cross-ply models shown a definite convergence. It can be estimated that, by further
increasing the number of layers in the model, convergence towards an asymptotic value
of around 20 MPa would be obtained.
0
5
10
15
20
25
0 7 14 28 56
Number of the layers
Lo
ng
itu
din
al s
tres
s [M
Pa
]
balanced-orthotropic model cross-ply models
Figure 18 Convergence of results on the specimen surface under maximum load
condition (Fmax = 58.9 N, Mmax = -88.29 Nm/m) as predicted by various
models and measured by experiment for stress results
Stress analysis for 90/0cross-ply models was also performed. As before, the
maximum load case (Fmax = 58.9 N, Mmax = - 88.29 Nm/m) was considered. The next
graphs refer to the mathematical models with 90°/0° degree fiber angle. In Figure 19, 20
and 21 the 14, 28 and 56-ply models with 90°/0° degree fiber angle are presented.
Figure 19 Stress – thickness distribution in longitudinal direction for 14-ply 90°/0°
degree fiber angle model
Figure 20 Stress - thickness distribution in longitudinal direction for the 28-ply 90°/0°
degree fiber angle model
Stress - Thickness Distribution in Longitudinal Direction for 56 plies 90/0 Degree Fiber Angle Model
Figure 21 Stress – thickness distribution in longitudinal direction for 56-ply 90°/0°
degree fiber angle model
In Figure 22 the stress distributions for the 14 and 28 cross-ply models with 90°/0°
degree fiber angle are presented. 56-ply model is not included to not complicate the graph
and make it hard to visualize the stress – thickness distribution. In this graph, the 7-ply
balanced-orthotropic model is also represented. The thickness distribution for the
balanced-orthotropic model is similar to 0°/90° models stress distribution. Here is visible,
in addition, the influence of the Terfenol-D in increasing the stiffness of the material with
results in larger stress
7-ply balanced-orthotropicmodel
14-ply 90/0 deg model28-0ply90/0 deg model
Figure 22 Stress-Thickness Distribution for the 7 ply 0 degree angle model and 14,
28-ply models with 90°/0° degree fiber angle
For analysis purposes Table 6 presents the largest stresses on the surface for each
model. In this case, the comparison shows that the stress carried in the transverse
direction, when the top ply has 90 degree ply angle, is smaller than the stress given by
the balanced-orthotropic model.
Table 6 Stress results on the surface for 7, 14, and 28 ply 0 and 90°/0° cross-ply
models
7 ply balanced-orthotropic model [MPa]
14 ply 90°/0° model [MPa]
28 ply 90°/0° model [MPa]
56 ply 90°/0° model [MPa]
12.72 5.94 5.55 5.4
3 MAGNETOSTRICTIVE COMPOSITE BEAM EXPERIMENT
The experimental results alluded to the previous section were obtained during a
carefully conducted experiments, as described next.
3.1 SPECIMEN DIMENSIONAL DESIGN
For pre-design purpose, the mechanical properties of the specimen were evaluated
with a very simplified theory that took into account the following data:
1 Volume fraction of the fiber: vf = 0.5
2 Volume fraction of the matrix vm = 1- vf = 0.5.
3 Young’s modulus of the E-glass fiber the: 9
fE 72.4 10 Pa
4 Young’s modulus of the resin 9
mE 3.25 10 Pa .
We assumed an average Young’s modulus for random fiber composite calculated
with (Malik, 1992, equation 3.42, page 130):
11 22
3 5E E E
8 8
(51)
E11 and E22 are calculated using the rule of mixture for longitudinal modulus, E11,
and transverse modulus E22 (Malik, 1992, equations 3.26,3.27, page 123).
11 f f m mE E v E v (52)
f m22
f m m f
E EE
E v E v
(53)
The result is 9E 17.808 10 Pa .
The shear modulus for the composite (Malik, 1992, equation 3.43, page 130)
material is assumed to be:
11 22
1 1G E E
8 4
(54)
The result is 9G 6.061 10 Pa .
Simple statics was used to determine the bending moment of a simply supported
composite beam centrally loaded by a concentrated force. To dimension the specimen for
further computations, some mechanical and dimensional quantities were assumed. The
equivalent Young’s modulus, E, calculated above was considered. The width was
measured on the composite specimen, w = 100 mm. The maximum deflection was
considered not to exceed the value of qmax = 36 mm based work previously done in the
area of materials testing.
To determine the thickness of the composite, the length, L to thickness, t, ratio was
calculated:
1
3max
max
2 w E qk
F
,
Lk
t
(55)
The maximum force was calculated based on the force of one weight and the
number of weights to be used in the experiment. The maximum force to be used in the
experiment was considered to be the force exerted by ten weights, Fmax = 196.2 N. The
thickness is then
max
a
3 F kt
E w
(56)
t = 5.051 mm.
The length of the specimen was calculated as:
L t k (57)
with L = 449.9 mm, thus t = 6.5 mm.
Furthermore the length of the specimen is adjusted to L = 579.6 mm. Rounding up,
the final length for following calculations is L = 600 mm.
The design dimensions of the specimen we choose to be:
Length – 1000 (mm) (58)
Width – 100 (mm) (59)
Thickness – 6.75 (mm) (60)
The next step was to calculate the moment of inertia for determining the deflection.
The moment of inertia was calculated using the well-known formula for a rectangle:
3w tI
12
(61)
The result is: 3 4I 2.563 10 mm .
Then the deflection is
3
F
LFq
E I 24
(62)
The result is
F
4.104
8.208
12.312
16.417
q [mm]20.521
24.625
28.729
32.833
36.937
for
19.62
39.24
58.86
78.48
F (N)98.1
117.72
137.34
156.96
176.58
(63)
At this point, the maximum stress and strain were calculated. Because of symmetry, the
stress and strain have same absolute value when equally distanced from mid-plane. The
values for the strain are:
83.6
167.5
250.8
334.4
(MPa)417.9
501.5
585.1
668.7
752.3
445
889
1334
1778
( )2223
2668
3112
3557
4002
(64)
The maximum stress was max
tM
2I
and the maximum strain was max
max E
.
Assuming the flexural strength to be 6
FS 150 10 Pa and a safety factor of SF=1.5,
admissible tensile stress is calculated as
F
a
S
SF
(65)
The result is 6
a 100 10 Pa . The admissible strain value is
a
a E
(66)
εa = 5617 με. (67)
Comparison of the result with the strain predicted above shows that adequate safety
is built into the specimen. These simple calculations were used to ensure that the
specimen was adequate for experimentation.
3.2 DESCRIPTION OF THE MS TAGGED COMPOSITE SPECIMEN
An MS tagged composite specimen was fabricated at Reichhold Chemicals
(Raleigh, N.C.) by binding 7 layers of fiberglass woven roving 36oz./sq.yd. and Atlac
580-05 Urethane-modified Vinyl Ester Resins. The specimen was 1000 mm long, and
had a 100 mm by 6.5 mm cross section. MS Terfenol-D tagging powder was used in the
two outside layers in the middle 500 mm of the span. Out of 1000 grams of resin, 250
grams had the MS powder. The resin was cured with 1 % MEKP (methyl ethyl ketone
peroxide) at room temperature for approximately 90 minutes. The target weight fraction
of the glass fibers in the composite was wf = 30 %. The weight fraction of MS Terfenol-D
tagging in the resin was 15 %.
3.3 SPECIMEN PREPARATION FOR THE EXPERIMENT
The specimen was instrumented with strain gages in the mid-span section on the
upper and lower surfaces. The specimen surface was prepared for bonding the strain
gages using the procedures given by Measurements Group (Raleigh, N.C.). The bonding
area was made planar using abrasive sandpaper. Next, the surface was cleaned using
Measurements Group surface conditioner M-Prep Conditioner A (a water based acidic
surface cleaner). Measurements Group strain gages types CEA-06-125UT-120 were
applied with M-Bond AE-10 Kit gage adhesive. To protect the strain gages and reduce
their exposure to the environment, the strain gages were covered with one sided Scotch
tape.
3.4 EXPERIMENTAL DESIGN
3.4.1 Description of the experiment
LakeshoreGaussmeter
StrainIndicator
MagnetostrictiveComposite
National InstrumentAmplifier
SolartronDisplacementTransducer
Figure 23 General view of the experiment set-up
The overall configuration of the experiment can be seen in Figure 23. The
experimental setup permitted the simultaneous measurement of beam deflection,
mechanical strains, and magnetic response of the magnetostrictive (MS) tagged beam.
Data flow through the data acquisition and processing modules is presented schematically
in Figure 24. The MS-tagged composite beam was supported on concrete blocks (500
mm equivalent span) and loaded gradually with an incremental number of clay weights (2
kgf = 19.6 N each). Strain gages were placed on both the upper and lower surfaces of the
specimen at the mid-span and were connected in a half bridge configuration to the strain
indicator. The magnetic flux density produced from the magnetostrictive particles was
measured by the gaussmeter.
Magnetic Sensor
Strain Gages
Strain Indicator
Gaussmeter
SCXI Unit DAQ PCLVDT
Displacement Transducer
Figure 24 Schematic of the experiment and data flow
For correlation purposes, the mid-span displacement was also measured. An LVDT
displacement transducer and a non-magnetic (aluminum and brass) clamping fixture were
used. Details of the mid-span instrumentation are given in Figure 25. Initial trials showed
that the strain gauge and LVDT electromagnetic fields did not influence the Gaussmeter
reading of the magnetostrictively induced magnetic field.
The MS tagged composite beam, supported on concrete blocks, was loaded
gradually with 2-kg weights. In order to avoid excessive strain in the composite material,
the number of loading weights was limited to three. The test procedure was as follow:
load the specimen, acquire data and repeat until all the three weights were on the
specimen. Next, reverse the procedure and perform unloading. The strain gages were
connected to a strain indicator, which provided the actual strain value at top and bottom
surfaces. At the same time, the displacements were measured with an LVDT
displacement transducer. Hence, the relationship between loads, displacements and
strains could be established. Simultaneously, the magnetic field developed by the MS
active tagging particles was detected using the gaussmeter. The gaussmeter provided the
value of the magnetic flux density read by the probe on the surface of the MS tagged
composite material. The data was collected using National Instruments LabView
software and associate hardware consisting of a SCXI signal-conditioning module, and
the Gateway computer. The information was processed using National Instruments
LabView software. The data collection was developed for simultaneous on-line operation
with the strain gages, displacement transducer and the gaussmeter. At each loading step,
complete data collection was performed. The configuration of the strain gages,
displacement transducer and position of the gaussmeter probe are presented in Figure 25.
Strain Gage
Gaussmeter Probe
The rod of the Displacement Transducer
Clamping Fixture
Figure 25 Position of the gaussmeter probe, strain gage, and displacement
transducer
3.4.1.1 Clamping fixture design
A clamping device was designed in order to achieve a better position of the
displacement transducer on the magnetostrictive composite material .The fixture is
composed of 2 aluminum parts (see Figure 25), which are kept attached to the composite
material with a brass bolt and nut. A brass bolt was chosen for use in order to diminish
the magnetic influence of all the components to the gaussmeter probe.
3.4.1.2 Protective wood fixture design for the strengthen the gaussmeter probe
A special fixture had to be constructed to ensure proper and consistent alignment of
the gaussmeter probe with respect to the composite surface. Its design is presented in
Figure 26:
Wood
Sensor Tip2 mm Sensor Rod
Figure 26 Concept design of the protective wood fixture
Considering the fragility of the gaussmeter probe and the advice from Lakeshore
Cryotonics, a wood fixture for the protection of the probe tip was designed and
fabricated. The distance from the tip of the sensor to the upper surface of the specimen
was 2 mm. The open area was covered with clear tape. The sensor rod was attached to the
wood fixture using Pro Seal Blue RTV Silicone made by Pacer Technology (Rancho
Cucamonga, CA.). A view of this apparatus is given in the Figure 27.
Sensor Tip
Machine Carved Wood
Sensor Rod
Figure 27 Overview of the assembly of magnetic sensor and carved wood
3.4.2 List of the equipment used
In this experiment, many different pieces of equipment were used to achieve the
simultaneous measurements of beam deflection, mechanical strains, and magnetic
response of the MS tagging particles. The equipment list is presented in Table 7.
Table 7 List of the equipment used in the magnetostrictive composite beam
experiment
Name Model Manufacturer
Strain Gages CEA-06-125UT-120 Measurement Group, Inc.
Strain Indicator P-3500 Measurement Group, Inc.
Gaussmeter Model 450 Lakeshore Cryotonics, Inc.
LVDT displacement transducer
B-50 Solartron Company
SCXI amplifier unit SCXI-1000 National Instruments Company
LabView NI Professional Measurement Suite
National Instruments Company
PCMCIA card DAQcard-AI-16E-4 16 channel
National Instruments Company
Composite Material Specimen
Reichold Chemicals
Permanent Magnets 6”x4”x1” Ceramic 8 Magnet Adams Magnetic Products
3.4.3 Equipment Calibration
In order to obtain valid data the equipment had to be calibrated. The calibrated
equipment was Solartron Displacement Transducer, Lakeshore Gaussmeter, National
Instruments Data Acquisition Card, and Measurement Group Strain Indicator.
Calibration of the displacement transducer implied establishing the relationship
between the measured displacements and the output voltage. The equation that defines
the displacement is:
outV
Displ (mm)2
(68)
For the Lakeshore gaussmeter, the calibration was achieved by setting the range
field of the magnetic field and corresponding output voltage. The range of the magnetic
field was established to be G06.0 . The corresponding output voltage for the maximum
and minimum values of the range field was V3 . The National Instruments data
acquisition card was calibrated using National Instrument software. For the Measurement
Group strain indicator the calibration required adjusting the reading of the output voltage
with a calibration coefficient for the real strain. This calibration factor was 0.45.
3.4.4 LabView Virtual Instrument Environment
The LabView visual programming language software used in the experiments was
purchased from National Instruments Company. For processing and visualizing the data,
we created a virtual instrument visual interface using the LabView programming
language. The front panel and the logic flow are shown in the Figure 28 and 29.
The front panel is shown in the Figure 28.
Figure 28 Front panel of the program created for the MS composite beam experiment
using the LabView virtual instrument
The front panel shows the virtual instrument for recording the increase and
decrease of the loading force, and the reading channels for displacement transducer,
gaussmeter, and strain gages. Figure 28 shows three graphs: (a) applied force versus
measured displacement; (b) milistrains given by the strain indicator versus applied force;
and (c) magnetic field read by the gaussmeter versus applied force. Another value
presented on the front panel is the brick weight, in case different types of weights are
used. Total force applied represents the force applied by the weights present at the time of
measurement. The logic flow of the software is presented in the Figure 29.
Figure 29 The logic flow of the program created for the MS composite beam
experiment using the LabView virtual instrument.
The data flows from the left to right. There are two “while loops” which control all
the flowing data in the program.The principal while loop is required by the language
programming in LabView. The inside loop contains the elements of creating , formatting
and saving of the data for each type of reading: displacements, magnetic
fields,milistrains. On the left are the input elements for the data. These elements are the
inputs from the data acquisition card. On the right, inside the while loop, are the elements
which read the data from the saved files in order to display, in the form of the graph, the
information on the front panel. There are some logic loops which control the logic of
taking and displaying the measurements.
4 EXPERIMENTAL PROCEDURE AND RESULTS
4.1 MAGNETIC ANNEALING
Before detailing the experimental procedure, we need to define the term “magnetic
annealing”. Magnetic annealing consists of applying a strong magnetic field that aligns
the magnetic domains of the MS tagging particles. Thus, the initial response under load is
enhanced. We used a pair of permanent magnets (Adams Magnetic Products, Garland,
TX) generating 740 Gauss magnetic flux density. The magnetic annealing was applied
with the specimen in the no-load position. During the set of experiments with magnetic-
annealing, we applied the magnetic annealing between every loading cycle.
4.2 TESTING PROCEDURE FOR MS TAGGED COMPOSITE BENDING EXPERIMENT
The MS-tagged composite beam, supported on concrete blocks (500 mm equivalent
span), was loaded and unloaded gradually with 2-kg weights. In order to avoid excessive
strain in the composite material, the number of loading weights was limited to three. The
test procedure was as follow: gradually load the specimen, acquire data and repeat until
all the weights were on the specimen. Next, reverse the procedure and perform gradual
unloading. The strain gages were connected to a strain indicator, which provided the
actual strain value at top and bottom surfaces. At the same time, the displacements were
measured with an LVDT displacement transducer. Hence, the relationship between loads,
displacements and strains could be established. Simultaneously, the magnetic field
developed by the MS active tagging particles was detected using the gaussmeter. The
gaussmeter provided the value of the magnetic flux density read by the probe on the
surface of the MS tagged composite material. The data was collected using National
Instruments LabView software and associated hardware consisting of a SCXI signal-
conditioning module, and the Gateway computer. The information was processed using
National Instruments LabView software. The data collection was developed for
simultaneous on-line operation with the strain gages, displacement transducer and the
gaussmeter. At each loading step, complete data collection was performed.
4.3 RESULTS WITHOUT MAGNETIC ANNEALING
In the first set of experiments the acquisition of the data was performed without
annealing the specimen between cycles. Displacement, strain and magnetic field levels
were also analyzed.
4.3.1 Displacement and strain without magnetic annealing
Table 7 shows the displacement results obtained for the experiment without
magnetic annealing and the force generated by the weights in loading and unloading
procedure.
Table 8 The displacement data without magnetic annealing
Displacement [mm]
Cycle 1
Cycle 2
Cycle 3
Cycle 4
Cycle 5
Cycle 6
Cycle 7
Cycle 8
Cycle 9
Cycle10
0 0.221 0.211 0.229 0.235 0.236 0.243 0.242 0.248 0.251 0.269
19.62 3.097 3.122 3.082 3.057 3.105 3.104 3.116 3.122 3.128 3.128
39.24 5.436 5.567 5.431 5.363 5.571 5.529 5.549 5.581 5.481 5.448
58.86 8.296 8.492 8.232 8.168 8.466 8.356 8.484 8.521 8.231 8.187
39.24 5.645 5.693 5.53 5.04 5.676 5.637 5.695 53579 3.201 3.213
19.62 3.202 3.225 3.161 3.134 3.2 3.205 3.204 3.203 3.201 3.213
0 0.257 0.256 0.256 0.262 0.249 0.269 0.268 0.272 0.291 0.283
Next, the dependency of the displacement with the applied force is presented in
Figure 28. It can be seen that there is a linear relationship between displacement and
force. It must be emphasis that this data depends on the accuracy of the data acqusition
card and National Instrument amplifier.
Figure 30 Mechanical data resulting from the experiments without magnetic
annealing between loading cycles: displacement vs. load
Good repeatability of the readings during loading and unloading was observed.
Both of these observations validate the experimental set-up. Table 8 presents the strain
data obtained. The precision of the voltage output of the Measurement Group Strain
Indicator was lower than National Instruments devices. For a very small strain displayed
on the strain indicator for instance 2-3 microstrains for zero loads, the voltage output had
the same value of noise. Thus a perfect zero output couldn’t be obtained. The error of the
output voltage versus real strains is lower than 2%. The linear relationship between strain
and force is presented in the Figure 29.
Table 9 Strain data without magnetic annealing
Strain ε [milistrains]
Cycle 1 Cycle 2 Cycle 3 Cycle 4 Cycle 5 Cycle 6 Cycle 7 Cycle 8 Cycle 9 Cycle10 Average strain
Standard deviation
%
0 0 0 0 0 0 0 0 0 0 0 0 0
19.62 443.9 461.8 466.6 454.6 469.7 468.7 456.6 463.5 454.6 467.7 460.7 1.78
39.24 858.1 883.6 872.2 853.2 888.1 879.1 870.8 871.2 877.7 886.3 874.0 1.30
58.86 1342.2 1347.1 1345.0 1309.5 1343.7 1318.1 1330.2 1339.9 1326.8 1332.3 1333.4 0.93
39.24 875.3 906.4 888.8 877.4 900.8 895.0 888.1 895.3 888.1 897.7 891.2 1.10
19.62 459.4 473.9 468.7 463.2 477.0 463.9 467.7 473.2 470.8 475.2 469.2 1.21
0 0 0 0 0 0 0 0 0 0 0 0 0
Figure 31 Mechanical data resulting from the experiments without magnetic
annealing between loading cycles: strain vs. load
By analyzing the strain data, linearity and repeatability of the strain versus force
data can be seen. The strain and displacement linear relationship is presented in the
Figure 30, which shows a linear dependency of the strain with displacement for a
specimen subjected to loading.
Figure 32 Mechanical data resulting from the experiments without magnetic
annealing: strain vs. displacement
4.3.2 Magnetic flux density without magnetic annealing
The magnetic flux density response of the specimen is presented in Table 10. The
specimen at the time of data acquisition had not been annealed for more than two days.
Thus, the results are indicative of the response of a field-deployed MS-tagged composite
material. The small difference between readings in different cycles illustrates the
deviation expected from the use of general-purpose magnetic flux density measurement.
The data acquired by equipment depends on the position of the gaussmeter probe, the
distance from the sensor tip to the surface of the specimen, and the environment. These
observations indicate that non-annealed MS-tagged composites present satisfactory
repeatability and small hysteresis in bending. White and Brouwers (1998) made similar
observations for axially loaded specimens and axially measured magnetic fields.
Table 10 Magnetic flux density results without magnetic annealing
Magnetic Flux Density B [mG]
Cycle 1 Cycle 2 Cycle 3 Cycle 4 Cycle 5 Cycle 6 Cycle 7 Cycle 8 Cycle 9 Cycle10
Average value
Standard deviation %
0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0
19.62 4.73 4.79 5.67 5.57 5.39 5.25 5.41 3.36 3.88 3.92 4.79 0.16
39.24 9.48 9.96 10.68 10.64 9.92 10.39 8.96 8.96 7.98 7.73 9.56 3.97
58.86 16.60 16.53 16.63 16.05 16.74 16.37 16.85 15.15 15.02 14.43 16.03 5.05
39.24 12.27 11.08 10.96 10.65 11.41 10.23 11.11 9.26 8.15 8.86 10.39 11.54
19.62 6.62 5.85 6.31 5.94 6.79 5.3 6.14 4.07 3.49 2.69 5.31 25.23
0 0.87 1.6 0.62 0.81 0.58 0.29 1.09 0.81 1.06 0.89 0.86 38.37
The zero load reading was substracted from the other values in order to eliminate
the influence of the environmental magnetic field. Ten cycles were considered in order to
see the repeatability of the magnetic flux results. Some difference in magnetic flux
density readings resulted between cycles were noticed. Though, the strain and
displacement reading were almost identical. This difference between cycles in magnetic
flux density readings arises from the fact that the magnitude of the magnetic flux density
is very small, of order of miligauss, and hence the influence of the environmental
magnetism has an impact on the stress-induced magnetic flux density readings. This
influence made the gaussmeter display to constantly fluctuate during readings. Graphical
representation of the magnetic flux density versus applied force is given in Figure 33.
The standard deviation of the magnetic flux density is presented in the Figure 34.
0
2
4
6
8
10
12
14
16
18
0 20 40 60 80
Force [N]
cycle 1cycle 2cycle 3cycle 4cycle 5cycle 6cycle 7cycle 8cycle 9cycle 10M
agn
etic
Flu
x D
ensi
ty B
[mG
]
Figure 33 Magnetic flux density responses, without magnetic annealing, to bending
strain in top surface of MS-tagged composite specimens: superposed 10
consecutive cycles
Figure 34 Mean value and standard deviation of the magnetic field
One can also see in Figure 33 that the magnetic field has a small hysteresis. This
phenomenon is present because of changes in the microscopic magnetic status of the
tagging particles (White and Brouwers, 97-98, pp 8).
Figure 35 The resulted magnetic field as function of measured strain
Figure 35 shows the raw data of magnetic flux density vs. strain obtained during the
ten cycles. Figure 35 shows that strain can be predicted when the magnitude of the
magnetic flux density B is known. At the time the experiment was performed and the data
was acquired, the specimen was not annealed for more than two days. Thus, the results
are indicative of the response of a field-deployed MS-tagged composite material. The
small difference between readings in different cycles illustrates the deviation expected
from the use of general-purpose magnetic flux density measurement equipment.
Figure 36 shows the average-trend correlation between the magnetic flux density
and strain for the specimen without annealing. The data presented are the mean values
and standard deviation of the experimental results. These observations indicate that non-
annealed MS-tagged composites present satisfactory repeatability and small hysteresis in
bending. White and Brouwers (1998) made similar observations for axially loaded
specimens and axially measured magnetic fields.
Figure 36 Mean values of magnetic flux density and standard deviation versus strain
for the specimen without annealing
4.4 RESULTS WITH MAGNETIC ANNEALING BETWEEN LOADING-UNLOADING
CYCLES
4.4.1 Displacement and strain with magnetic annealing between loading-unloading
cycles
A new set of experimental data was obtained by applying magnetic annealing
between cycles. Ten loading-unloading cycles were conducted, with magnetic annealing
between each cycle. The annealing procedure was performed with 2 magnets with
average magnetic field magnitude of 740 G. Adams Magnetic Products (Garland, TX)
supplied the magnets for this experiment. The procedure followed was analogous to that
previously use, with exception that annealing between cycles (load and unload the
specimen with all the weights) was performed. The annealing time was on average 1.5
minutes. Displacement, strain and magnetic field levels were collected and analyzed. The
strain and displacement results showed a very good repeatability with less than 1.7%
standard deviation. The displacement results are presented in Table 11.
Table 11 Displacement results with annealing between loading-unloading cycles
Displacement [mm]
Cycle 1 Cycle 2 Cycle 3 Cycle 4 Cycle 5 Cycle 6 Cycle 7 Cycle 8 Cycle 9 Cycle10
0 0.215 0.308 0.418 0.395 0.368 0.277 0.357 0.343 0.278 0.356
19.62 2.927 3.004 3.049 3.109 2.898 2.964 3.027 3.007 2.895 3.052
39.24 5.479 5.611 5.589 5.64 5.433 5.557 5.604 5.56 5.405 5.637
58.86 8.03 8.467 8.318 8.487 8.259 8.411 8.475 7.925 8.135 8.442
39.24 5.588 5.765 5.739 5.738 5.585 5.691 5.75 5.68 5.527 5.753
19.62 3.029 3.146 3.16 3.212 3.078 3.101 3.131 3.111 3.03 3.168
0 0.35 0.414 0.515 0.517 0.405 0.363 0.426 0.419 0.406 0.471
The displacement results are presented graphically in Figure 35. The graph shows
the linearity of the displacement with loading and unloading. The strain results are
presented in the Table 11.
Figure 37 Displacement results versus applied force for annealed specimen
Table 12 Strain results with annealing between loading-unloading cycles
Strains ε [milistrains]
Cycle 1 Cycle 2 Cycle 3 Cycle 4 Cycle 5 Cycle 6 Cycle 7 Cycle 8 Cycle 9 Cycle10
Average value
Standard deviation,
%
0 0 0 0 0 0 0 0 0 0 0 0 0
19.62 440.4 432.1 432.5 443.5 434.9 437.0 437.3 436.6 439.0 434.9 436.8 0.80
39.24 886.7 882.6 879.1 882.6 877.4 875.3 874.3 869.4 881.9 880.1 878.9 0.55
58.86 1310.2 1359.5 1323.0 1365.4 1344.0 1340.9 1336.4 1296.4 1245.4 1359.9 1337.1 1.6
39.24 895.3 896.0 895.3 894.3 878.4 890.1 876.7 880.1 889.5 898.8 889.4 0.91
19.62 447.0 452.8 447.0 457.0 444.6 448.0 447.3 450.1 445.9 452.5 449.2 0.84
0 0 0 0 0 0 0 0 0 0 0 0 0
It should be noted that the strains in Table 12 are very similar to those in Table 9,
indicating that magnetic annealing does not influence the mechanical response. The
linearity of the strain with applied force is presented in Figure 38.
0
200
400
600 800
1000
1200 1400
1600
0 20 40 60 80
Force [N]
Str
ain
[m
icro
stra
ins]
cycle 1 cycle 2 cycle 3 cycle 4 cycle 5 cycle 6 cycle 7 cycle 8 cycle 9 cycle 10
Figure 38 Strain results for the MS composite specimen with magnetic annealing
between cycles
4.4.2 Magnetic flux density with magnetic annealing between loading-unloading cycles
The experimental magnetic field data are given in Table 13. The 10 cycles show
that the measurements are repeatable and consistent. As in the experiment without
magnetic annealing, the small difference between readings in different cycles illustrates
the deviation to be expected from the use of general-purpose magnetic flux density
measurement equipment.
Table 13 Experimental magnetic flux density with magnetic annealing between
loading-unloading cycles
Magnetic Flux Density B [mG]
Cycle 1
Cycle 2
Cycle 3
Cycle 4
Cycle 5
Cycle 6
Cycle 7
Cycle 8
Cycle 9
Cycle10 Average value
Standard deviation
%
0 0.05 0.00 1.6 0.70 -0.25 -0.05 -0.05 0.47 0.33 0.64 0.3 0
19.62 11.03 8.18 9.98 8.42 11.92 16.18 14.99 17.10 14.88 15.29 12.80 24.37
39.24 23.88 20.63 50.45 16.45 20.43 24.14 22.89 25.5 22.64 23.02 21.99 11.09
58.86 32.59 28.19 29.18 26.75 29.05 32..5 31.23 33.91 30.65 30.37 31.43 7.85
39.24 33.79 28.69 29.38 26.37 28.81 32.93 31.54 34.94 30.56 30.13 30.71 8.04
19.62 32.84 27.57 27.63 25.83 30.02 33.88 33.42 35.11 30.87 30.36 30.75 9.46
0 36.34 30.27 29.95 20.33 24.03 29.49 29.72 30.06 26.20 25.29 28.17 14.73
The plot of magnetic flux density versus force is given in Figure 39. It can be seen
that, on the increasing branch of the curve, the magnetic field response is almost linear.
The maximum magnetic flux values are about double the values obtained without
annealing (32 mG vs. 16 mG, respectively). This proves that magnetic annealing
increases the initial magnetostrictive response. The increasing branch also shows a slight
negative curvature, i.e., the marginal response per unit load becomes less pronounced as
the load increases. On the decreasing branch of the curve, the marginal response is very
small, and significant hysteresis is present. These observations indicate that annealed
MS-tagged composites possess a higher initial response, but also a very pronounced
hysteresis. Mean average and standard deviation of the magnetic flux density are
presented in Figure 38.
Figure 39 Magnetic flux density results for specimen with magnetic annealing
between cycles
Figure 40 Mean value and standard deviation of magnetic flux density results vs. load
for the specimen with magnetic annealing between cycles
Figure 41 shows the correlation between the magnetic flux density and strain for
the specimen with annealing between cycles. Mean average and standard deviation of the
magnetic flux density is presented in Figure 42.
Figure 41 Experimental magnetic flux density results versus strain with magnetic
annealing of the specimen between cycles
Figure 42 Mean value and standard deviation of magnetic flux density results vs.
strain for the specimen with magnetic annealing between cycles
4.5 MAGNETIC FLUX DENSITY ANALYSIS
Based on the stress results of the model and experimental magnetic flux density
data Figure 43 shows the relationship between two variables. The graph shows
convergence of stress results for the models without averaging the Young’s Modulus.
Figure 43 Magnetic flux density in relationship with model longitudinal stresses on the
surface of the composite specimen model
The results show linearity between longitudinal stress obtained using bending
moment of the weights involved in experiment and magnetic flux density (B) measured
during the experiment. For purpose of calculating the magnetostrictive strain, coefficient
standard deviation of the stress from average has been produced. This is presented in
Figure 44.
Figure 44 Standard deviation of the stress from the mean value versus magnetic flux
density
Similar behavior is observed in the magnetic flux density versus calculated strain of
the models. This is presented in following figure, Figure 45.
Figure 45 Magnetic flux density versus strain on the top surface of the model material
considered and 0°/90° degree fiber angle without magnetic annealing.
4.6 PIEZOMAGNETIC STRESS AND STRAIN COEFFICIENTS
For monolithic magnetostrictive Terfenol-D material, the piezomagnetic
coefficients, e31, and d31 are well known and available from standard references (Butler,
1988; De Lacheisserie, 1993). These coefficients correlate the magnetic flux density, B3,
with the strain, ε1, and stress, σ1. For magnetostrictively tagged composites, such simple
relationships cannot be directly established, since the composite contains three separate
phases (fiber, resin, and magnetostrictive tagging powder) of which only one is
magnetostrictive. However, effective piezomagnetic coefficients can be established to
allow a macroscale relationship between the recorded magnetic flux density and the strain
and stress in the tagged composite. To this purpose, we have calculated both the overall
e31 and the d31 coefficients, as shown in the following sections. The e31 coefficient was
calculated because it can be determined directly, from the measured strain and magnetic
flux. The d31 coefficient was calculated in order to facilitate comparison with the d31 data
reported in the literature for axially loaded MS composite specimens. It should be noted
that the d31 coefficient cannot be determined directly from experimental data alone, since
no simple device exist to directly measure the stress inside an composite beam in
bending. We determined the d31 coefficient using the stress predicted by the lamination
analysis, and the magnetic flux density measured during the experiments. Details of this
process are given next.
4.6.1 Piezomagnetic stress coefficient, e31 without magnetic annealing
By using the mean value of magnetic field versus strain for the loading steps,
piezomagnetic strain coefficient was calculated. The trend line helps to determine this
coefficient as a ratio of magnetic flux density B, and strain .
According to De Lacheisserie (1993), the piezomagnetic stress coefficient e31,
defines the relationship between the applied strain, ε1, and the resulting flux density B3, in
the form:
3 31B d (69)
3 31 εB e (70)
The Figure 46 shows the trend line and the point that was chosen to determine the
piezomagnetic stress coefficient. The trend line was considered only for the loading steps.
Thus, the piezomagnetic stress coefficient e31, for the case when the specimen is not
magnetically annealed, is:
-431
N13.7 10
A me
(71)
0
2
4
6
8
10
12
14
16
18
0 500 1000 1500
Experimental Strain [microstrains]
Mag
net
ic F
lux
Den
sity
B [
mG
]
Slope 13.6/1000 mG/με
Figure 46 Plot of the average magnetic flux density vs. strain showing the trend lines
in determination of the piezomagnetic stress coefficient e31 for non -
annealed specimen case
4.6.2 Piezomagnetic stress coefficient, e31, with magnetic annealing
Piezomagnetic stress coefficient e31 was calculated using the average of the
magnetic flux density function of the strain experimentally obtained when annealing the
specimen between cycles.
0
5
10 15
20 25
30
35
40
0 500 1000 1500
Experimental Strain ε [microstrains]
Exp
erim
enta
l M
agn
etic
Fie
ld B
[m
G]
Slope 24.5/1000 mG/με
Figure 47 Plot of the average magnetic flux density vs. strain showing the trend lines
in determination of the piezomagnetic stress coefficient e31 of magnetic
annealing of the specimen
Considering the plot in Figure 47 the piezomagnetic stress coefficient e31 for the
case when the specimen is annealed we obtained:
431
N24.5 10
A me
(72)
4.6.3 Piezomagnetic stress coefficient ,e31 ,using experimental magnetic flux density
without magnetic annealing and the 56-ply strain results model
We calculated the piezomagnetic stress coefficient, e31, using the plot of the
average values of the experimentally determined magnetic flux density, B3, and the
calculated strain ε1. In our investigation, we used several lamination models (7-ply
through 56-ply), and compared their predicted strain with experimentally measured
values. This showed that the 56-ply model was most adequate.
An analysis of the piezomagnetic stress coefficient, when correlating between the
experimental magnetic flux density results of non-annealed specimen case and the strain
results of the 56-ply model, is now shown. For the piezomagnetic stress coefficient e31 we
used:
0 2 4 6 8
10 12 14 16 18
0 200 400 600 800 1000 1200 1400
56-ply model strain [microstrains]
Exp
erim
enta
l M
agn
etic
Flu
x
Den
sity
B [
mG
]
Figure 48 Plot of the average magnetic flux density with no magnetic annealing
specimen vs. 56-ply model strain showing the trend lines in determination
of the piezomagnetic stress coefficient e31.
Thus, the values for the piezomagnetic stress coefficient e31 was found as:
431
N13.25 10
A me
(73)
4.6.4 Piezomagnetic stress coefficient, e31 using experimental magnetic flux density
with magnetic annealing between loading-unloading cycles and the 56-ply strain
results model
In a similar way with previous coefficient determination, the correlation between
experimental magnetic flux density data for the specimen annealed between cycles and
the strain data of the 56-ply model was determined. The value for the coefficient is then:
431
N25 10
A me
(74)
56-ply Model Strain [microstrains]
0 5
10 15 20 25 30 35
0 200 400 600 800 1000 1200 1400
Exp
erim
enta
l M
agn
etic
Flu
x D
ensi
ty B
[m
G]
Figure 49 Plot of the average magnetic flux density with magnetic annealing vs. 56-
ply model strain showing the trend lines in determination of the
piezomagnetic stress coefficient e31
4.6.5 Piezomagnetic strain coefficient d31 using experimental magnetic flux density
without magnetic annealing and 56-ply stress model
The piezomagnetic strain coefficient d31 was determined using the experimental
magnetic flux density of the not annealed specimen and the 56-ply model stress results
(Figure 50). Based on the trend line in Figure 50 the piezomagnetic strain coefficient, d31
was calculated by considering the ratio of the magnetic flux density to strain. From the
graph, we calculated the piezomagnetic strain coefficient d31 for the not annealed
specimen to be:
1431
m10.2 10
Ad
(75)
0 2
4 6 8
10 12 14 16 18
0 5 10 15 20 25 56-ply model Stress [MPa]
Exp
erim
enta
ly M
agn
etic
Flu
x D
ensi
ty B
[m
G]
Figure 50 Trend line for calculating the piezomagnetic strain coefficient
4.6.6 Piezomagnetic strain coefficient ,d31 ,for annealed specimen experiment and 56-
ply stress model
In a similar way, the piezomagnetic strain coefficient is determined for the case of
magnetic annealing of the specimen between cycles. As stated before, the piezomagnetic
coefficient d31 was considered to be the ratio between experimentally obtained magnetic
flux density B and 56-ply model stress. Considering the graph above for the
piezomagnetic strain coefficient the value found is:
1431
m15.6 10
Ad
(76)
Figure 51 Trend line for computing the piezomagnetic strain coefficient d31 for the
case of magnetic annealing of the specimen between cycles
5 ANALYSIS AND DISCUSSION
5.1 COMPARISON OF PIEZOMAGNETIC COEFFICIENTS FOR MAGNETOSTRICTIVE
TAGGED COMPOSITES WITH PREVIOUSLY PUBLISHED DATA
The present authors used the experimentally-derived curves reported by other
investigators (Quatttrone, Berman, and White, 1996) to identify average values of stress-
magnetic flux density coefficients as depicted in Figure 3. These coefficients express the
magnetic flux density developed when a given axial stress is applied to the MS tagged
neat-resin composite specimen (Please recall that the piezomagnetic stress coefficient
defines magnetic flux density per unit strain, and vice-versa). The values of these
coefficients were entered in the last two columns of Table 14. Table 14 shows that the
“raw” piezomagnetic strain coefficients obtained in our composite experiments are about
4 to 10 times lower then those published by Quattrone, Berman, and White (1998) for
neat-resin tension specimens.
Table 14 Summary of magnetostrictive coefficient for MS tagged composites, as
determined in the present work and by previous investigators
Bending Series Tensile Series
Not annealed Annealed Not annealed Annealed
Composite type Woven Woven Neat resin Neat resin
Chained No No Yes Yes
Measured B Field Direction Transverse Transverse Axial Axial
Piezomagnetic Stress Coefficient (N/Am) e31 =13..710-4 e31 = 24.510-4 X X
Raw d31 = 10.210-14 d31 = 15.610-14
Corrected for MS volume
fraction d31 = 33.710-14 d31 = 51.510-14
Note: The tensile series for neat resin not annealed was published by Trovillion, J et all
(1999). The tensile series data for annealed neat resin was published by Quattrone, Berman, and
White (1998). The bending series refers data were published by Giurgiutiu, et al. (1999). The
piezomagnetic coefficients were calculated by the authors of the present report from the data
published in the references cited above.
The reasons for this difference are:
1. Different type if stress distribution, bending vs. axial. The axial stress
distribution is uniform across section, while the bending distribution is
linearly varying from maximum positive to negative values. Hence, only a
small portion of the composite (outer fibers) is actually loaded to the
maximum stress. Thus, for the same stress values in the outer fibers, a
tensile specimen will produce much more magnetism than bending
specimen. Since the gaussmeter probe was measuring the overall response,
it results that its readings are expected to be lower in the bending specimen
than in the tensile specimen.
2. Previous investigators, performing tensile experiments, measured the
magnetic flux density, B3, parallel to the applied stress, σ3. In our bending
experiments, the applied load generated stress in the σ1, which is
perpendicular to the direction of the measured B3.
3. In the tensile net-resin specimens used by the previous investigators, the
Terfenol-D particles were dispersed throughout the neat-resin body,
whereas in our composite bending experiments an important portion of the
specimen was occupied by the reinforcing fibers. Hence, the Terfenol-D
particles could only be dispersed in the 81% volume allocated to resin.
Additionally, Terfenol-rich resin was applied only in the outer layers of the
composite lay-up. In essence, this means that, in our composite bending
specimens, the Terfenol volume ratio was 0.68%, whereas in neat-resin
tensile specimens it was 2.24%, which is 3.3 times higher.
When the difference in MS tagging volume fraction was compensated for, the
values of piezomangetic strain coefficients measured by us in bending without annealing
between cycles, became comparable with those measured by Quattrone, Berman and
White (1996) under axial loading and similar non-annealing conditions. This shows
convergence of the two investigations and increased confidence in the industrial
application potential of this technology.
5.2 COMPARISON BETWEEN EXPERIMENTAL, DESIGN AND MODEL STRAIN FOR
WITH AND WITHOUT MAGNETIC ANNEALING
Since the 56-ply model 0°/90° degree fiber angle was considered to be the closest
in simulation of the composite material, the results of the model are included in
comparison with classical theoretical prediction and experimental strain obtained.
Comparing the information about the strain, results and theoretical values, we can
conclude that the results meet the predicted behavior of the composite material specimen.
The comparison is presented in the Table 15.
Table 15 Comparison of the design, model and measured strain
Experimental Strain [με] με με
Non Annealed Annealed
0 0 0 0 0
19.62 445 432 460 436
39.24 889 842 874 878
58.86 1334 1296 1333 1337
39.24 889 842 891 889
19.62 445 432 469 449
0 0 0 0 0
As can be seen for both of annealing and not annealing, the results of the predicted
and mathematical model are close to experimental results. These comparisons consider
the loading and unloading of only three weights, and for the mathematical model used, 56
ply model with 0°/90° degree fiber angle.
5.3 COMPARISON BETWEEN EXPERIMENTAL AND MODEL PIEZOMAGNETIC
STRESS COEFFICIENT WITHOUT MAGNETIC ANNEALING
The purpose of comparing the piezomagnetic coefficients is to check how close is
the model to the experimental data. In determining the mathematical piezomagnetic stress
coefficient, the magnetic flux density experimentally obtained has been used with model
strain calculated. The model strain calculated was considered for the 56 ply with 0°/90°
degree ply angle. For the experiment, considering the non-annealed specimen the
piezomagnetic stress coefficient has been obtained (64):
431
N13.7 10
A me
. As
previously shown for the 56-ply model has been obtained (66):
431
N13.25 10
A me
.
By comparison, both coefficients, the 56 ply mathematical model proves to give good
level of accuracy in prediction.
5.4 COMPARISON BETWEEN EXPERIMENTAL AND MODEL PIEZOMAGNETIC
STRESS COEFFICIENT WITH MAGNETIC ANNEALING
Here a comparison between piezomagnetic stress coefficient what was obtained for
the 56-ply model and the one for the material with annealing between loading cycles is
presented. As before, the experimental piezomagnetic stress coefficient was calculated as
the ratio of the value of magnetic flux density and 1000 strains. This fraction was
calculated using the graph presented in Figure 47. Using the value for the piezomagnetic
stress coefficient of the 56-ply model, which was obtained in paragraph 4.5.4,
431
Ne 25 10
A m
to the value for the composite material (on the surface) that was
obtained at paragraph 4.5.2,
431
Ne 24.5 10
A m
, we observe that the results are very
close. In this case the difference between piezomagnetic coefficients is small. The reason
for this is that the magnetic flux density (B) is different from the not annealed case. The
prediction of the mathematical model when is used experimental magnetic field data is
very good.
6 CONCLUSIONS
This thesis has presented data obtained from the experimental study of
magnetostrictive-tagged fiber reinforced composites under bending (flexural) load and
from two-dimensional mathematical models. A brief review of the state of the art
revealed that no previous work on bending of magnetostrictive-tagged composites has
been published yet.
A magnetostrictive-tagged composite bending specimen was designed, fabricated,
and tested. An analysis procedure of laminate woven composites was applied to predict
the state of stress and strain in a simple supported composite beam specimen.
Convergence of results was studied using an increasing number of layers from 7 to 56.
The main equipment and the experimental setup and testing procedure have been
described. Test data was obtained with and without annealing between loading cycles.
The results were processed and presented in terms of magnetic response vs. strain and vs.
predicted stress. For tests without annealing between cycles, the results showed linearity
and low hysteresis. For tests with annealing between cycles, an increase in response and
pronounced hysteresis were both observed.
The piezomagnetic stress coefficient, e31, was calculated using measured magnetic
flux vs. measured strain for both annealed and non-annealed experiments. The
piezomagnetic strain coefficient, d31, was also calculated using the measured magnetic
flux density and the stress predicted by a 56-ply model with 0/90 cross-ply layers.
Comparison between the present bending results and the previously published results for
axial loading was performed.
The values of these coefficients are presented in chapter 5. Piezomagnetic stress
coefficient d31 has been calculated using the mathematical model for future reference.
The strain data used are from the 56-ply model with 0°/90° degree fiber angle.
This paper has proved that the strain vs. magnetic flux density coefficient for
magnetostrictive-tagged composites is portable from neat-resin composite to fiber
reinforced composites and can be used as a “master” design coefficient. This concept, not
previously reported in literature, resolves a major issue related to the design of
magnetostrictive-tagged composites with various fiber - reinforcement architecture,
which share a common design criteria base on strain or deflection constraints. Further
work needs to be performed to identify the variation of strain vs. magnetic field
coefficient with magnetostrictive tagging density and to develop a three dimensional
mathematical model. On this line, one has to verify whether the saturation phenomenon
previously reported in neat-resin magnetostrictive-tagged composites, also appears in
fiber-reinforced composites, and to find tagging density values at which it becomes
present. This investigation has shown that a clear relationship can be established between
the magnetic flux density measured at the surface of a MS tagged composite and the
stress state inside the specimen. Previous investigators had proved this relationship for
axial loading. In this paper, for the first time, MS tagged composite results under bending
loading are presented. Both axial results (presented by previous investigators) and
bending results (present in this paper) indicate a good consistency and repeatability of the
magnetic flux response resulting from loading and unloading of the MS-tagged
composite.
This paper also showed the concept that magnetostrictive tagging can be applied
sparsely, i.e., only in critical places. In our specimen, magnetostrictive tagging was
applied only in the outer layer and only in the mid-span region. This concept allows
tailoring of the magnetostrictive tagging and results in considerable cost saving and
improved application efficiency.
The impact of using MS-tagged composite for in-service NDE of composite
structure could be very significant. Unlike conventional strain gages, which need to be
permanently applied and wired on the structure, the MS-tagging method can be done by
simple surface scanning using a traveling magnetic flux gage. This opens the opportunity
for measuring stress patterns in large structural areas to identify stress distributions and
“hot spot”.
At present, the MS-tagging NDE methodology is still in its infancy. Considerable
work needs still to be done before large-scale industrial implementation is possible.
Further research work needs to cover: (a) development of portable, hand-held equipment
for magnetic flux measurement of large structural areas using surface scanning
techniques; (b) further characterization of magnetic response of MS-tagged composite of
various tagging densities and distributions; (c) evaluation of long-term behavior under
various environmental conditions
7 BIBLIOGRAPHY
Adly, A.A., I.D. Mayergoyz, and A. Bergqvist, (1991), "Preisach Modeling of Magnetostrictive
Hysteresis " Journal of Applied Physics, Vol 69, No. 8, 1991, pp 5777-5779.
Brouwers, F.,(1998), "Magnetostrictive Tagging of Composite Materials for Structural Health
Monitoring," Master's Thesis, University of Illinois (Urbana-Champaign), Department of
Aeronautical and Astronautical Engineering, October 1998.
Butler, J. L. (1988) “Application Manual for the Design of Etrema Terfenol-D Magnetotrictive
Transducers”, Etrema Products Inc., 1988, pp 1- 10.
De Lacheisserie, E. du T. (1993) “Magnetostriction- Theory and Applications of
Magnetoelasticity”, CRC Press, 1993, pp. 53-56 and 345-347.
Engdahl, G., and L. Svensson, "Simulation of the Magnetostrictive Performance of Terfenol-D in
Mechanical Devices," Journal of Applied Physics, Vol 63, No. 8, 1988, pp 3924-3926.
Etrema Products, Inc., 2500 North Loop Drive, Ames, IA 50010.
Giurgiutiu,V.; Jichi F.; Rogers, C.A.; Quattrone, R.; Berman, J.B. (1999a) “Experimental Study
of Mangetostrictive Tagged Composite Strain Sensing Response for Structure Health
Monitoring”, 2nd International Workshop of Structural Health Monitoring, Sept. 8-10,
1999, Stanford University, pp. 690-699.
Giurgiutiu, V., Jichi, F., and Rogers, C. A. (1999b) “Design, Fabrication and Testing of a
Magnetostrictive-Tagged Fiber-Reinforced Composite Beam”, Paper # USC-ME-LAMSS-
99-106, June 30, 1999.
Hommema, J.A., (1999) "Magnetomechanical Behavior of Terfenol-D Particulate Composites,"
Master's Thesis, University of Illinois (Urbana-Champaign), Department of Theoretical and
Applied Mechanics, October 1999.
Jones, R.M. (1999) “Mechanics of Composite Materials”, Hemisphere Publishing
Corp.,1999
Malik, P.K. (1993) “Fiber-Reinforced Composites. Materials, Manufacturing and Design”,
second edition, revised and expanded, 1993, Marcel Dekker Inc.
Mitrovic, M., M. Robert, and G. Carman (1995), "Magnetostrictive Composite Material Systems
Analytical/Experimental," Proceedings of the American Society for Composites Tenth
Technical Conference, Technomic Publishing Co., Inc., Lancaster, PA, 1995, pp 85-94.
Nersessian, N., Carman, G., (2000), “ Magneto-mechanical Characterization of Mangetostrictive
Composite”, The 2000 ASME International Mechanical Engineering Congress and
Exposition, Nov.5-10, 2000, Orlando, Florida, pp 139
Pinkerton, F.E., T.W. Capehart, J.F. Herbst, E.G. Brewer, and C.B. Murphy, (1997),
“Magnetostrictive SmFe2/metal Composites,” Applied Physics Letters, 70:2601-2603.
Quattrone R., J.B. Berman, and S.R. White, “Self-Monitoring Structures Containing
Magnetostrictive Composite,” Proceedings, 21st Army Science Conference, June 15-
17,1998, Norfolk, VA, pp 21-26.
Quattrone, R.F., J.B. Berman, Z. Chaudhry, V. Giurgiutiu, C.A. Rogers, and S.R. White,(1998),
“Investigation of Active Tagged Composites for Army Infrastructure Applications,” CERL
TR 98/022, ADA344919, April 1998.
Quattrone, R., Berman, J. B., and White, S. R. (1998) “Self-Monitoring Structures Containing
Magnetostrictive Composite”, Proceedings, 21st Army Science Conference, June 15-
17,1998, Norfolk, VA, pp. 21-26.
Trovillion, J., Kamphaus J., Quattrone R., Berman J. (1999) “Structural Integrity Monitoring
Using Smart Magnetostrictive Composites”, Proceedings, International Composites
EXPO’99, Cincinnati, OH, 1999, session 22-D, pp. 1-6.
White, S. R.; Albers, R. G., (1996) "Magnetostrictive Tagging of Composite Materials for
Structural Health Monitoring," USACERL Final Technical Paper, Feb., 1996.
White, S.R., and Brouwers, F. (1998) “Smart Materials for Infrastructure Applications”,
Technical Paper, USACERL Contract # DACA88-97-K-0001, June 24, 1998.
White, S. R. (1999) “Magnetostrictive Tagging of Composites for Health Monitoring”,
Proceedings, 1999 International Composites Expo, Cincinnati, OH, May 10-12, 1999, pp.
22E1 – 22E6.
White, S. R.; Albers, R. G.; Quattrone, R. F., (1996) "Tagging of Composite
Materials with Terfenol-D Magnetostrictive Particles for Structural Health Monitoring,"
ASME 1996 Mechanics and Materials Summer Meeting, June 12-14, 1996, Johns
Hopkins Univ., Baltimore, MD.