Professional Documents
Culture Documents
IN LAYERED MANUFACTURING
A DISSERTATION
OF STANFORD UNIVERSITY
DOCTOR OF PHILOSOPHY
Xiaochun Li
June 2001
© Copyright by Xiaochun Li 2001
ii
I certify that I have read this dissertation and that in my opinion it is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.
Fritz Prinz
(Principal Adviser)
Thomas W. Kenny
Drew V. Nelson
iii
Abstract
Layered Manufacturing can be applied to build "smart" parts with sensors,
integrated circuits, and actuators placed within the component. Embedded sen-
sors can be used to gain data for validating or improving designs during the pro-
totype stage or to obtain information on the performance and structural integrity
of components in service. Techniques for embedding fiber optic sensors in met-
als, polymers, and ceramics have been investigated. Embedding optical fibers
into metals is especially challenging because engineering alloys tend to exhibit
high melting temperatures. In the present research an embedding sequence was
developed capable of embedding fiber sensors into parts made of metal alloys
with high melting temperatures.
Fiber Bragg Grating (FBG) sensors were selected as the most promising sensor
candidate. The embedded FBG sensors were characterized for temperature and
strain measurements. The embedded FBG sensors in nickel and stainless steel
provided high sensitivity, good accuracy, and high temperature capacity for tem-
perature measurements. Temperature sensitivity approximately 100% higher
than that of bare FBGs was demonstrated. For strain measurements, the sensors
embedded in metal and polyurethane yielded high sensitivity, accuracy, and lin-
earity. The sensitivity of the embedded FBGs was in good agreement with that
of bare FBGs. Moreover, a decoupling technique for embedded FBG sensors was
developed to separate temperature and strain effects. The embedded FBG sen-
sors were used to monitor the accumulation of residual stresses during the laser-
assisted Layered Manufacturing, to measure the strain field in layered materials,
to measure pressure, and to monitor temperature and strain simultaneously.
iv
New techniques have been developed for temperature and strain measurements
of rotating components with FBG sensors embedded or attached to the surface.
Tunable laser diodes were incorporated into the sensing system for monitoring
the Bragg grating wavelength, and thus the temperature or strain was deter-
mined. The non-contact sensing system provides a new sensing tool that is supe-
rior to slip rings. Moreover, the new techniques provide a new health monitoring
methodology for rotating structures, especially those exposed to hostile environ-
ments, such as blades in gas turbine engines.
v
Acknowledgements
I would like to thank my advisor, Professor Fritz Prinz for his support and
advice during the past years. I am grateful to have the opportunity to learn the
novel manufacturing technologies and to reshape my future research direction
under the guidance of his superior vision.
I would also like to thank Professors Thomas Kenny and Drew Nelson for
serving on my reading committee. I must thank Professor David Barnett for his
insightful comments on my research. I owe many thanks to Professor Stephen
Tsai, who chaired my Defense Committee.
vi
Contents
Abstract.....................................................................................................iv
Acknowledgements.....................................................................................vi
List of Tables.............................................................................................xi
List of Figures...........................................................................................xii
1.1. Overview..............................................................................................1
vii
2.3.3. Fiber optic sensors embedding..........................................................16
3.1. Introduction........................................................................................22
3.2.1. Cleaning...................................................................................24
3.2.2. Sputtering.................................................................................26
3.2.3. Electroplating...........................................................................28
3.5. Summary.............................................................................................35
viii
4.4.2. Strain response..........................................................................53
5.5. Summary.............................................................................................76
6.1. Introduction........................................................................................78
ix
6.4.5. Low sensing speed....................................................................93
6.5. Summary.............................................................................................97
7.1. Introduction......................................................................................100
7.3.4. Properties................................................................................111
7.5. Summary...........................................................................................121
Bibliography.......................................................................................126
x
List of Tables
Table 3.1. Deposition parameters for Ti and Ni film....................................................27
xi
List of Figures
Fig.2.1. Layered Manufacturing system........................................................3
xii
Fig.4.12. Two FBGs for decoupling of temperature and strain effects...........46
Fig.4.19. Setup of four-point bending test and cross section of the beam.......54
Fig.5.2. Path sets of the laser cladding on the stainless steel beam................62
xiii
Fig.5.11. Pressure sensor with an a/t ratio of 15...........................................71
Fig.6.1. Sensing system for rotating parts with embedded FBG sensors........80
Fig.6.8 Geometric parameters for the rotary plate with embedded FBG.........87
xiv
Fig.6.19. Remote sensing system for rotating parts......................................98
xv
Chapter 1: Introduction
1.1. Overview
Layered Manufacturing can build functional “smart” parts with sensors, inte-
grated circuits, complete functional assemblies, and actuators placed within the
structure and fully embedded. These sensors can be used to gain data for validat-
ing or improving designs during the prototype stage or to obtain information on
the performance and structural integrity of functional components in service. A
sequence for embedding fiber optic sensors in engineered structures is devel-
oped. Fiber Bragg Grating (FBG) sensors, embedded in metallic, nickel and
stainless steel, polymers, and ceramic structures, are characterized for tempera-
ture and strain measurements. Moreover, a decoupling technique that is capable
of separating temperature and strain effects is also developed. This dissertation
also presents various applications of the embedded sensors, for instance, the
monitoring of residual strain during the laser-assisted Layered Manufacturing,
the measuring of the strain field in layered materials, and the measuring of dif-
ferential pressures.
The dissertation also discusses a new technique based on FBG sensors that are
embedded or attached to the surface, which allows the measurement of the tem-
perature and strain for rotating components. The new techniques provide a new
health monitoring methodology for rotating structures, especially those exposed
to hostile environments, such as blades in gas turbine engines.
1
expansion (CTE), high yield strength, good toughness, and high wear resistance.
This dissertation presents the investigation of laser-deposited composites of
INVAR and TiC we undertook to address these issues.
2
Chapter 2: Background
Decreasing product development cycle times and increasing product complex-
ity are crucial for any country or corporation to remain successful in global
competition. In the last decade, “Layered Manufacturing” or “Solid Free-form
Fabrication” has emerged as a popular manufacturing direction to accelerate
product creation. Layered Manufacturing can build parts that have traditionally
been impossible to build because of their complex shapes or variety in materials.
A number of Solid Freeform Fabrication processes have been developed [1-8]. A
general system is schematically presented in Figure 2.1 [1].
Layered Manufacturing can build functional “smart” parts with sensors, inte-
grated circuits, complete functional assemblies, and actuators placed within the
3
structure and fully embedded. In particular, sensors embedded within the struc-
tural materials add intelligence to structures and enable real-time monitoring at
some critical locations not accessible to ordinary sensors, which must be
attached to the surface. Moreover, embedded sensors are also protected from
damage caused by extraneous environmental effects. These sensors can be used
to gain data for validating or improving designs during the prototype stage or to
obtain information on the performance and structural integrity of functional
components in service. The capability to obtain such information is important to
many industries. Examples include the manufacturing industry (molds, dies,
drilling bits, etc.), the aerospace industry (components of jet engines), the oil
industry (drilling equipment), the power industry (vessels and pipes), the auto-
motive industry (components of motors), and the construction industry (struc-
tural components in buildings).
4
2.1.1. Stereolithography (SLA)
Stereolithography is currently the most widely used SFF RP technology. It was
first introduced into market by 3D Systems. SLA uses an ultraviolet laser to
selectively solidify photocurable resins in a vat of liquid photocurable polymer,
building shapes directly from a computerized model [9-12]. Parts are built up on
an elevator platform that incrementally lowers the part into the vat by the dis-
tance of the designated layered thickness. A laser is then guided to scan across
the surface selectively. The step is repeated until the final geometry is achieved.
Stereolithography can build very complex shapes. However, supports are
required for large overhanging features since the initial thin layers can break off
as the part moves down into the liquid. Normally, these supports are made of
thin wall sections, which can be removed easily after the part is completed.
5
tively scans designated areas to form shapes from the solid decomposition prod-
ucts of the gas precursor. SALD involves only a substrate, a reactant gas or
gases, and a scanning laser beam. Examples of materials deposited this way
include carbon, silicon carbide, titanium oxide, and silicon nitride [25-27].
Unique combinations of materials can be achieved in the final parts with avail-
able gas precursors. We can also tailor local microstructures by carefully con-
trolling processing temperature, gas partial pressure, and other variables,
thereby fabricating a structure comprising multiple materials [28].
6
features include a conformal cooling channel, embedded copper, tough core
(INVAR), and a hard surface (stainless steel)
7
MD*, "recursive mask and deposit", is a layered manufacturing process that
was developed prior to SDM and that lacks the advantages of SDM. In MD* lay-
ers of material are sprayed through disposable masks and parts are thereby cre-
ated [35]. The MD* process can fabricate parts from sprayed zinc, steel, or
nylon materials, but the mask generation and placement steps are cumbersome.
Disadvantages of the MD* method are the unreliability of complete automation
of the process, the presence of stairlike steps in the resulting parts, and the rela-
tive inaccuracy of parts produced this way [30].
8
ander Nickel’s Ph.D thesis [38]. Nickel studied thermal stresses in SDM using a
combination of analytical modeling, finite element modeling, and experiments.
Initially he developed an analytical model to predict the overall part warpage.
Then he used both a finite element model and experiments to investigate how
the pattern used to deposit a layer influences the substrate warpage and to inves-
tigate the interlayer surface defect known as the Christmas Tree Step. Since this
approach needs significant cooling time between depositions, it is too slow for
the rapid prototyping process. Moreover, accurate models for the computation of
residual stresses are difficult to obtain.
In one earlier study, Nau embedded strain gauges into tensile and bending
specimens, which were fabricated via the SLA process, and compared their out-
puts to these of external strain gauges during testing [43].
9
and PZT ceramic/polymer composites (for actuators and sensors) free of the
constraints of the design complexity [44]. Many novel and complex composite
structures, including volume fraction gradients (VFG), staggered rods, radial
and curved composites, and actuator designs such as tubes, spirals and telescop-
ing structures were made using the flexibility provided by the FDC technique
[45-47].
Denham et al. and Calvert et al. studied the fabrication of polymers and com-
posites containing embedded sensors, which were manufactured in-situ by an
modified FDM process [48-50]. In collaboration with Advanced Ceramics
Research Corp. of Tucson, Arizona, they developed a reactive extrusion method
to build parts by the extrusion of a material stream through a fine needle. A cur-
ing reaction at the deposition site allowed solid components to be formed. This
approach can be readily modified to allow deposition of several materials in the
same part and hence can be used to fabricate intelligent materials containing
embedded sensors and actuators. Examples were demonstrated, such as the
application of this method to an epoxy part containing an optical fiber sensor
and polyvinylidenefluoride piezoelectric films [48]. Piezoelectric polyvi-
nylidenefluoride films were also embedded between layers of composite lami-
nates and cured. The films retained much of their piezoelectric response upon
cooling even though the cure temperature exceeded the melting point of the
film. Embedded sensors were used to monitor the curing of reinforced low tem-
perature curing epoxy films sandwiched between polycarbonate sheets and sub-
jected to an intermittent external load perpendicular to the surface [49]. Epoxy
parts were freeformed with lines of conducting carbon-filled polymer written
into the structure during forming. Parts were also made with sensors distributed
across the surface [50].
10
also used SALD to make the embedded thermocouple devices in situ and
SALDVI to fabricate the macro-components. The fabricated in situ thermocou-
ples responded sensitively to temperature variation. Furthermore, the electric
signal of the embedded thermocouple was stable and reproducible in response to
thermal cycling. However, if the thermocouple was not embedded in the SiC
matrix, the signal was not stable mainly due to the oxidation of the thermocou-
ple elements.
Luo et al. embedded sensors into the core and cavity of a rapid resin mold for
direct measurement of pressure and temperature for the purpose of improving
the injection molding process [55]. Using the information collected by embed-
ded sensors, they developed a control system to ensure the quality of products.
MD* and SDM were used to build “smart” objects with embedded electronics,
sensors, and actuators. Beck et al. and Prinz et al. investigated the feasibility of
manufacturing mechatronics and forming smart structures via the MD* process
[56, 57]. An early part was a "Simon" game packaged within a brick of polymer
material [56]. Sprayed zinc traces on two planar layers were connected to indi-
vidual integrated circuits and passive components. A castable polymer material
was used to make the housing in an aluminum mold.
Fasching et al. produced “smart” coatings that contain embedded sensors [58].
Using masking techniques, they spray-formed sensors, such as thermocouples,
humidity sensors, and strain gages, as well as sensor arrays directly on coatings.
The integration of sensors into thermally sprayed coatings can provide feedback
about the functional status and operating history of the coatings as well as that
of the coated structures and surrounding environments.
11
wearable computer built using Shape Deposition Manufacturing was the
"VuMan 3R" or "VuMan SDM" [61]. Two circuit boards connected by pins and
pin receptacles were sandwiched between three solid layers of polymer [62].
Buttons, LEDs, and a jack for the attachment of a head-mounted display consti-
tuted the user interface for the computer. A socket for a PCMCIA card was
included to allow the computer to be reconfigured. Standard consumer batteries,
located in battery compartments on both ends of the housing, powered the com-
puter.
Cham et al. presented prototypes for biomemetic robotics to illustrate the pos-
sibilities for devices with embedded components and they addressed the issues
that constrain their process planning. They also discussed a combination of pro-
12
cess planning algorithms and manufacturing methods to support the design of
layered products with embedded components [65].
Bailey et al. used SDM to fabricate small scale robots from off-the-shelf struc-
tural materials, sensors and actuators, which were embedded directly into three-
dimensional structures without fasteners or connectors [66]. They also fabri-
cated structures with spatially varying material properties such as specific stiff-
ness and damping. In addition, they designed small robot limbs with locally
varying stiffness and embedded sensors and actuators and manufactured them
via SDM.
13
sor embedding and Fiber Bragg Grating sensor embedding have become popular.
This section discusses the different sensors employed for embedding.
14
repeatability, sensitivity and reliability. All devices survived and yielded repeat-
able responses to uniaxial tension loads applied over 10000 cycles.
15
try would eliminate wires and the stress on long wire runs between MEMS sen-
sors and processing microelectronics in harsh environments. Research in this
wireless area should facilitate the integration of smart sensing, control, and
actuation with the unprecedented capability to permanently embed telemetry as
a part of the functional components.
As early as 1978, Butter and Hocker developed a fiber optics strain gauge,
which was based on measuring the change in optical path length of an optical
fiber [77]. In 1991, the proceedings of “Fiber Optic Smart Structures and Skins
IV” edited by Richard and Udd, incorporated thirty-five papers that were
arranged in six sessions: embedding/materials/modeling; damage; high tempera-
ture; sensor system signal processing; smart material and structure integration;
and smart material and structure impelmentation concepts [78]. Spillman
reviewed the application of fiber optic embedded sensors as health monitoring
elements in so-called “smart” structures in 1994 [79]. The embedded fiber optic
16
sensors have been employed as damage sensors [80, 81]. These damage sensors
were applied to investigate the development of low velocity impact-induced
delamination in composite laminates with a single optical fiber embedded at the
laminate mid-plane. Lawrence applied embedded fiber optic strain sensors for
process monitoring of composites [82]. Foedinger et al. studied structural health
monitoring of filament-wound composite pressure vessels with embedded opti-
cal fiber sensors [83]. Kim et al studied the effect of embedded sensors on the
strength of composite laminates [84]. They performed in-plane compressive and
three-point bend tests to evaluate the strengths of composite laminates contain-
ing embedded sensors. The data showed that the sensors had little effect on the
strength.
More recently, Fiber Bragg Grating sensors have become popular for process
monitoring, especially for temperature and strain measurements [85-87]. Thus,
the principle of Fiber Bragg Grating sensors is discussed.
17
length-coding manner [76, 90]. FBGs have been considered suitable for measur-
ing static and dynamic fields, such as temperature, strain, and pressure [91].
Broad-band
light
Single
λb Fiber Bragg Grating
d
ncore
L
Reflected light Transmitted
light
λ 0 = 2n 0 d 0 (2.1)
∆λ = 2n 0 ∆d + 2d 0 ∆n (2.2)
which leads to
∆λ 1- dn
------- = λ 0 α + ---- --------0 (2.3)
∆T n 0 dT
18
Wagreich’s paper [94] and from Kim’s dissertation [95] are selected for theoret-
ical consideration due to the lack of data from manufacturers of fiber Bragg
The strain dependence of the Bragg wavelength arises from the physical elon-
gation of the sensor and the change in refractive index due to photoelastic
effects. The wavelength shift due to axial strain can be obtained by equation
∆λ = ( 1 – p e )λ 0 ε (2.4)
where ε is the applied strain, and p e is the photoelastic coefficient term given by
2
n0 (2.5)
pe = ------- [ p 12 – ν ( p11 + p 12 ) ]
2
where the p i,j coefficients are the Pockel's coefficient of the strain-optic tensor,
and ν is Poisson's ratio. Some typical values are listed as: ν=0.2, p 11 =0.113,
p 12 =0.252, n 0 =1.46. Thus, the value of ∆λ/ε is approximately 0.001051 nm/µε at
1300 nm and 0.001254 nm/µε at 1550 nm under axial strain.
Kim studied birefringence in a fiber core due to unequal transverse strains and
expressed the response of the two individual Bragg peaks, which are induced by
the birefringence, to applied strains and temperatures [95] as:
2 2 2 2
∆λ n n n n dn-
1- -----
---------2 = 1 – ----0- p 12 ε 1 – ----0- p 11 ε 2 – ----0- p 12 ε 3 + ----0- ( p 11 + 2p 12 )α + ---- ∆T (2.6)
λ0 2 2 2 2 n 0 dT
2 2 2 2
∆λ n n n n dn-
1- -----
---------3 = 1 – ----0- p 12 ε 1 – ----0- p 12 ε 2 – ----0- p 11 ε 3 + ----0- ( p 11 + 2p 12 )α + ---- ∆T (2.7)
λ0 2 2 2 2 n 0 dT
where ε 1 is axial strain, and ε 2 and ε 3 are transverse strains. Thus, under unequal
transverse strains without axial strain or temperature change, the sensitivity of
19
the two peaks from a fiber Bragg grating sensor with an original Bragg wave-
length of 1300 nm can be given by:
The wavelength shifts, ∆λ 2 or 3 , under transverse strains are much lower than
∆λ 1 under a same strain value in the axial direction.
and
∆λ dif
------------ = Ks ( ε2 – ε3 ) (2.11)
λ0
where ∆λ avg is the average wavelength change, ∆λ dif is the differential wave-
lengths resulting from birefringence, and K 1 , K h , K T, and K s can be expressed
as:
2 2
n0 n0 (C – C )
s
K 1 = 1 – ----- p 12 + ----- ( p 11 + p 12 ) --------------------s
2 2 C (2.12)
2G + -----s
ν
s
C
---- ( ν + 1 )α – C ----- ( ν s + 1 )α s
n
2
s 1 dn n
2
ν ν
s
K T = ----0- ( p11 + 2p12 )α + ----- ------ + ----0- ( p 11 + p 12 ) ---------------------------------------------------------------
- (2.13)
2 n 0 dT 2 C
s
2G + -----s
ν
2
n0 4G ( 1 – ν )
K s = ----- ( p 12 – p 11 ) -------------------------------------s (2.14)
4 G + ( 3 – 4ν )G
20
where G, C, ν, and α are shear modulus, Lame’s constant, Poisson’s ratio, and
21
Chapter 3: Sensor Embedding
This chapter discusses a method for embedding fiber optic sensors into metal-
lic structures, as well as into polymers and ceramics, in Layered Manufacturing.
3.1. Introduction
Layered Manufacturing processes provide capabilities not easily achieved
with conventional manufacturing processes. Examples include the ability to cre-
ate complex three-dimensional shapes with internal cooling channels, to produce
parts with continuously varying material properties and parts with embedded
mechanical and electrical components.
Since the sensors are embedded into functional metallic structures, a non-
obtrusive embeddability is a necessity for maintaining the integrity of the func-
tional metallic structures. The sensors ought to be small and rugged inside the
22
metal matrix. Past research has demonstrated the feasibility of embedding thin-
film sensors into metallic structures [63]. However, the insulation layer in thin-
film sensor embedding is prone to cracks induced by stress during the high tem-
perature material processing. Thus, new sensors are needed to improve the
embeddability.
Recently, fiber optic sensors and communication links have been identified as
promising candidates for integration into structural materials [76]. These sen-
sors offer immunity to electromagnetic interference, non-obtrusive embeddabil-
ity, good resistance to a hostile environment, and high bandwidth. Unlike thin-
film sensors, these sensors do not need electrical insulation since they are not
electrically conductive.
The embedding of fiber optic sensors has been investigated intensively for
composites. However, only a few papers demonstrate techniques for embedding
these sensors in metal parts [97-99]. Tayler et al. and Lee et al. embedded a fiber
Fabry-Perot Interferometer into aluminum, whose melting point is 660 °C, by
casting it in a graphite mold [97,98]. Baldini et al. used gold coating on optical
fibers before placing these fibers into titanium matrix composites produced by
arc spraying [99]. However, none of these embedding techniques are successful
enough to useful for the embedding of fiber optic sensors during the production
of 3-D full dense metallic structures in Layered Manufacturing.
23
rial, for instance acrylate, reduces light loss caused by microbending that occurs
when the fibers are pressed against rough surfaces, and provides protection to
the fragile cores and claddings. The diameter is 9 µm for the core of a single-
mode fiber while 125 µm for the cladding.
Coating
Cladding
Core
3.2.1. Cleaning
The cleanliness of the substrates is important for the later electroplating. Thus
the substrates were rinsed in borothene, acetone and isopropanol to remove
salts, oils, and grease. Then the substrates were blown dry with dry nitrogen,
24
rinsed with de-ionized water, and dried again. The above steps were repeated
once more, and then the substrates were immersed in a 1:15 solution of micro-
detergent in de-ionized water before the substrates were placed in an ultrasonic
bath for five minutes. Finally the substrates were rinsed with copious amounts
of de-ionized water and blown dry with dry nitrogen.
Ni electroplating
Substrate
The cleaning of the optical fibers was much simpler since they came coated.
Mechanical stripper is a standard tool for stripping the polymer coating. How-
ever, fiber optic sensors were available without any coating in the area of the
sensing element. The interface between the cladding of the sensing area and the
neighboring polymer coating, as shown in Figure 3.1, created a stiffness mis-
match and since the fragile sensing element lacked polymer coating, it easily
broke during the mechanical stripping. Moreover, there were residues of the
polymer coating left on the surface of the cladding. These residues tended to
induce microbending and debonding during the later high-temperature embed-
ding. Thus, a chemical stripping process was used. The optical fibers were
immersed in a bath of acetone for about ten minutes. The polymer coatings then
25
detached from the cladding. The process created no stresses and no residues, and
the sensing elements remained undamaged.
3.2.2. Sputtering
Since an optical fiber is basically made of SiO 2 , an electrically conductive
coating is needed to serve as a “seed” for electroplating. This conductive layer
should be thin and capable of bonding well with the optical fiber. Moreover, this
thin layer should be fabricated at a low temperature to avoid damage in the opti-
cal fiber. We selected sputtering to deposit the metallic thin film.
26
enhanced process, provides the impinging species to activate interfacial bond-
ing, thereby giving better adhesion than evaporative deposition.
pressure as low as 10 -6 Pa. Bare optical fibers were then sputter-coated with a
thin titanium film (< 0.1 µm) to enhance adhesion. Afterwards a thin nickel film
(< 0.2 µm) was sputter-coated over the thin titanium film. The process parame-
ters for the sputtering deposition are described in detail in Table 3.1. A short
period (< 10 min.) of Ni thin-film sputtering was preferred since the Ni target
became hotter and hotter during the sputtering. Hotter plasma in the chamber
might have resulted in microcracks in the optical fibers because of a temperature
gradient produced by it.
27
3.2.3. Electroplating
Electroplating is a process of producing a coating by electrolysis. Nickel was selected
because it bonded well to the subsequently deposited stainless steel layer. Ini-
tially, the exposed stainless steel surface was mechanically roughened by sand-
blasting in order to improve the mechanical interlocking with the nickel layer.
The substrate was then transferred immediately to the electroplating bath for
the growth of a nickel layer of 0.5 mm to 1.0 mm thick. To ensure low residual
stresses from the electroplating process, a Barrett SN solution (Nickel sulfa-
mate) plating bath was used. The chemical composition of the bath is listed in
Table 3.4, while the deposition process parameters are presented in Table 3.5.
The temperature of the plating bath was controlled at 49°C. A DC power sup-
ply was used to control the current density. At the first hour of plating, the cur-
rent density was set at 0.215 mA/mm 2 . Then, the current density was raised to
0.646 mA/mm 2 . At the end of the deposition sequence, the substrate was
retracted from the bath and all insulating layers (photoresist, resin, and kapton
tape) were removed. Finally, the substrate was rinsed in distilled water and
blown dry.
28
Reverse current density 0.215mA/mm2 (two minutes)
Forward current density 0.215mA/mm2 (two minutes)
Temperature 27 °C
Table 3.2: Electroplating parameters for “seed” Ni layer
29
3.2.4. Laser cladding
The optical fiber is now ready to be electroplated into a thicker nickel layer or
embedded in the other metallic structures by high-temperature processing (cast-
ing, welding, or arc spraying, etc.). In this study, a 2400W, CW Nd:YAG laser
was used to fuse metal powders. Stainless steel powders were pre-placed on the
surface of the nickel layer. The laser power was controlled, and the deposition
apparatus moved across the surface of the substrate by a four-degrees-of-free-
dom robotic manipulator.
The main issues are the quality of the bond between the nickel layer and the
substrate, the thermal stress induced by a large temperature gradient that exists
during layered deposition, and the fragile points where the optical fibers enter
and leave the metal part. During the manufacturing cycle, two types of defects,
delamination and cracking, will damage the optical sensors and lower the
strength of the structure. Delamination or micro cracks can be induced by inade-
quate cleaning of the optical fibers and the substrates. Figure 3.4(a) shows a
crack at the interface of an optical fiber and the electroplated nickel while Fig-
ure 3.4(b) displays delamination due to inadequate cleaning of the substrate.
Delamination or cracking may also occur during or after high-temperature
embedding. Laser cladding, which is similar to a welding process, was used in
this study. The cladding parameters for nickel are similar to those of stainless
steel, except the molten nickel weld pool is more difficult to control. Adequate
mixing of the molten stainless steel powders and nickel is crucial for a good
30
bonding. Figure 3.4(c) shows a crack at the interface of the laser-deposited
stainless steel and the electroplated nickel. Figure 3.4(d) shows delamination,
cracks and voids due to inadequate cleaning, inadequate mixing of molten metal,
and thermal stress.
Part Support
material material
Deposit (part)
Shape
Shape
Metallic thin film
Insulating coated fiber optic
layer sensor
Deposit (support)
Ni electroplating
Since SDM is an additive and subtractive process, the points where fiber optic
sensors enter and leave the metallic structure need extra protection. In particu-
lar, optical fibers are very fragile during the "shaping" process because chips
generated during the machining processes can easily cut off exposed optical
fibers. One way to prevent this problem is to a nickel coating on the extra fiber
31
leads (normally more than 10 mm in length) outside the ingress or egress points,
and used a polyethylene furcation tube with 1.2 mm diameter to protect the other
optical fibers. We can also cut away all extra fiber leads outside the parts. This
requires a non-contact sensing system to launch light into the optical fibers.
This non-contact sensing technique is discussed in Chapter 6.
Electroplated nickel
27 µm
Electroplated nickel
µ
Optical fiber
Electroplated 2 mm
µ nickel
Stainless steel substrate
(c)
32
3.4. Sensor Embedding in Other Materials
We also investigated the feasibility of embedding fiber optic sensors in other
structural materials, such as polymers and ceramics, during Shape Deposition
Manufacturing. These sensors, if successfully embedded, can be used to evalu-
ate the validation of a design and the integrity of a structure. The sensors can be
interconnected into a real-time monitoring network to detect structural weak-
nesses, thereby becoming a basis in the development of smart polymer and
ceramic materials.
Nickel
33
Ceramic materials pose more challenges to the embedding of fiber optic sen-
sors. A high-temperature process (normally above 1200 °C) is needed to produce
dense ceramic parts. The high-temperature means that silica optical fiber cannot
be used. Sapphire optical fibers may allow effective long-term sensor operation
at temperatures above 1700 °C, at which other conventional sensors either do not
perform as well as they do at low temperatures or at which they degrade unac-
ceptably. Sapphire optical fibers typically are manufactured by either edge-
defined film growth or crystalline draw methods. Since the refractive index of
sapphire is only slightly higher than that of sintered alumina, alumina is selected
to serve as a cladding for sapphire fibers, thereby forming a sapphire waveguide.
Alumina is often used in electrical and electronic applications because it is an
excellent insulator and is relatively inexpensive. However it is also used in
structural applications because of its good mechanical properties. Alumina is
highly chemically resistant, and since it is an oxide ceramic it is oxidation resis-
tant, which makes it ideal for use in harsh chemical environments.
34
ing. Afterwards, the alumina part was heated up to 600 °C for debinding. The
heating process burned out the polymer remaining in the ceramic, eliminating
the possibility of defects and cracks resulting from trapped polymer. Finally the
temperature was raised to 1300 °C for the sintering process, which made the
ceramic part hard and dense. The sintering process was performed under a nitro-
gen atmosphere with a pressure of 1 atm.
3.5. Summary
Layered Manufacturing can be used to integrate fiber optic sensors during the
production of tooling and structural components. Sensors can be placed close to
points of interest prior to enclosure. This chapter presented a method for embed-
ding fiber optic sensors in metallic structures by laser-assisted Shape Deposition
Manufacturing. Manufacturing issues, such as integrity of the structures with
optical fiber sensors, were discussed. Feasibility of embedding sensors in poly-
mer and ceramic materials were also investigated.
35
Chapter 4: Characterization of Embedded
Sensors
This chapter presents results for the characterization of embedded Fiber Bragg
Grating (FBG) sensors. Microstructures of electroplated Ni, laser-deposited
stainless steel, and Ni in the embedded structures are presented. Temperature
and strain responses of the embedded sensors are discussed.
36
half of the reflected peak was captured with an optical spectrum analyzer with a
resolution of +/-0.001 nm. The Bragg wavelength shifts were used to determine
linearity with axial strains.
Figure 4.2 shows results of calibration for an FBG with a Bragg wavelength of
1300.017 nm (from Blue Road Research, OR) and one with a Bragg wavelength
of 1534.568 nm (from O/E Land, Inc., Canada).
Weight
FBG 0f 1300nm under axial strain FBG of 1534nm under axial strain
2.5 2.5
Wavelength shift, nm
Wavelength shift, nm
1.5 1.5
1 1
0.5 0.5
0 0
0 500 1000 1500 2000 0 500 1000 1500 2000
37
The sensitivity is in good agreement with a value of 1.192x10 -3 nm/µε for an
FBG of 1550 nm and a value of 1.0x10 -3 nm/µε for an FBG of 1300 nm [92].
Oven
Broad band LED
2 x 2 fiber coupler
FBG
OSA
Figure 4.4 indicates that the Bragg wavelength shift is linear to temperature
change, and the sensitivity is determined to be 9.861x10 -3 nm/ºC for the FBG
with a Bragg wavelength of 1300.017 nm and 1.342x10 -2 nm/ºC for the FBG
with a Bragg wavelength of 1534.568nm. The sensitivity of the thermal
responses of the FBGs are slightly different from the theoretical values at Chap-
ter 2: 1.051x10 -2 nm/ºC @1300nm and 1.254x10 -2 nm/ºC. The difference may be
due to an inaccurate value of dn 0 /dT in theoretical model. It is difficult to obtain
38
the accurate value since manufacturers do not provide this data for FBGs in this
study.
6 12
y = 1.342E-02x - 5.450E-02
Wavelength shift, nm
Wavelength shift, nm
y = 9.861E-03x - 2.099E-04
5 10 R2 = 9.996E-01
R2 = 1.000E+00
4 8
3 6
2 4
1 2
0 0
0 200 400 600 0.0 500.0 1000.0
A high reflectivity is always desirable for an FBG sensing system. Figure 4.5
presents how temperature affected the reflectivity of an FBG when it was placed
in an oven for a long time. A recoverable decrease in reflectivity was induced
when the temperature was less than 550 ºC. The decrease became permanent,
however, when the temperature was more than 550 ºC. Finally, the reflectivity
dropped to almost 0% when the temperature was held at approximately 860 ºC
for more than 20 minutes. These results suggest that the FBG should be used at
temperatures lower than 550 ºC to keep its reflectivity intact. These effects of
temperatures were also verified by other reports [76, 90].
39
Temperature and time effects for FBG of 1534.568nm
100 1000
Temperature change, C
Reflectivity, % 80 800
60 600
40 400
20 200
0 0
0 100 200 300 400
Time, min
Reflectivity, % Temperature change, C
125 µm
40
Microscope (TEM), the microstructure of electroplated nickel was revealed, and
it is shown in Figure 4.7. The grain size is approximately 150 nm. The mechani-
cal properties of nickel are significantly improved if it has such a fine grains.
Through tensile tests, the yield strength, tensile strength, and strain to failure of
the electroplated nickel were found to be 928.72 MPa, 946.65 MPa, and 4.6%,
respectively. Moreover, the Young’s modulus, Poisson’s ratio, and CTE for Ni
are 214GPa, 0.31, and 12.9x10 -6 /ºC respectively. The high mechanical strength
plus the corrosion resistance of the electroplated nickel make it suitable for sen-
sors in hostile environments, such as sensors monitoring temperature or pressure
in oil wells or the deep sea.
150 nm
with a Bragg wavelength of 1300.017 nm and 1.212x10 -2 nm/µε for the FBG
41
with a Bragg wavelength of 1534.568nm. The sensitivity of the thermal
responses of the embedded FBGs are slightly different from the theoretical val-
@1300nm and 1.296x10 -2 nm/ µε @1550 nm. It suggests that the bonding
between nickel and optical fiber is strong.
Wavelength shift, nm
Wavelength shift, nm
Figure 4.9 presents two spectral responses for two different temperatures. At
room temperature, 20.4 °C, which was measured by thermocouple, the peak
wavelength of the reflected light for the Ni-coated FBG was about 1299.940 nm.
This wavelength was only slightly shifted from 1299.990 nm, which was mea-
42
sured before the electroplating process. This results suggests that the low tem-
perature embedding processes, sputtering and electroplating, produce almost no
residual stress in the Ni-coated FBGs. At a temperature of 250 °C, the peak
wavelength was at about 1304.74 nm. In a span of 230 °C, the wavelength
shifted approximately 4.65 nm.
4.50E-06
4.00E-06
3.50E-06
Amplititute, nW
3.00E-06
2.50E-06
2.00E-06
1.50E-06
1.00E-06
5.00E-07
0.00E+00
1298.00 1300.00 1302.00 1304.00 1306.00 1308.00
Wavelength, nm
Figure 4.10 presents detailed data for two FBGs, one with a Bragg wavelength
of 1299.940 nm and the other with a Bragg wavelength of 1546.345 nm. Temper-
ature change is plotted against wavelength shift. The best-fit line for the two
parameters shows high linearity. As shown by the best-fit equations, 0.0212 nm/
°C is the temperature sensitivity for the Ni-coated FBG with the Bragg wave-
length of 1299.94 nm while 0.0257 nm/°C is that for the embedded FBG with the
Bragg wavelength of 1546.345 nm. The values for the temperature sensitivity
are approximately twice as high as those of the bare FBGs measured before elec-
troplating. Clearly, nickel coating increases sensitivity of the FBGs due to its
higher CTE. The theoretical temperature sensitivity can be obtained by the equa-
tion (2.10)-(2.13) as: 0.0188 nm/°C @1299.94nm and 0.0224 nm/°C
@1546.345nm, which are more than 10% off from the experimental ones. The
43
difference may be due to the selected values of n 0 , dn 0 /dT, Pockel’s constants,
and a constant CTE of nickel.
Wavelength shift, nm
Wavelength shift, nm
6 12
y = 0 .0 21 2x - 0 .02 86
10 y = 0 .0 25 7x + 0 .0 14 9
R 2 = 0.9 98 9 R 2 = 0 .9 99 6
4 8
6
2 4
2
0 0
0 100 200 300 0 100 200 300 400
44
A sudden temperature drop appeared and then the temperature rapidly increased
again. When more coolant was sprayed onto the surface, the temperature experi-
enced another sudden drop, whereas the wavelength curve was much smoother.
While the temperature change decreased to 26 °C, the wavelength shift dropped
to almost zero. The wavelength shift should be about 0.52 nm, which corre-
sponds to a 26 °C temperature change if a temperature sensitivity value of 0.021
nm/°C is applied. The difference between the expected and the observed wave-
length shift is possibly due to the thermal stress induced by the large tempera-
ture gradient at the moment when coolant is sprayed on the top surface of the
sample.
300 6
Temperature Change
250 Wavelength Change 5
Temperature change, (°C)
200 4
Dl (nm)
150 3
100 2
50 1
0 0
-50 -1
0 500 1000 1500 2000 2500 3000
Time (s)
At the second thermal cycle, the wavelength shift obviously lagged behind the
temperature change until the temperature change reached the peak at about 260
°C, where the temperature sensitivity became about 0.021 nm/°C again. It is
possible that the sample experienced some annealing when it was heated for 10
minutes. After the second peak in the temperature change, the temperature sen-
sitivity well matched 0.021 nm/°C until the coolant was applied again. The tem-
45
perature gradient might induce compressive residual strains, which shifted the
wavelength down almost 1 nm when the sample was cooled to room tempera-
ture. Thus, a temperature gradient applied to the embedded FBG sensor could
cause errors if the FBG is used to measure temperatures. A new design of
embedded FBG sensors that can separate the effect of temperature from that of
mechanical strain is highly desirable.
Embedded FBG 1
Single λb1
For the embedded FBG, the Bragg wavelength shift, ∆λ 1 , can be induced by
both temperature and strain and it can be given by equation (2.10) while for the
free-floating FBG, the Bragg wavelength shift, ∆λ 2 , responds only to tempera-
ture and is given by equation (2.3). Under axial stress and temperature loads, the
strain ε and temperature T can be determined.
To prove the validity of this design, two FBGs with initial Bragg wavelengths
of 1300.102 nm and 1299.978 nm, manufactured by Blue Road Research, were
used to form a decoupling sensor. A stainless steel tube with an inner diameter
of 170 µ m and a wall thickness of 150 µ m was used to host the FBG with a
46
Bragg wavelength of 1299.978 nm. The tube was sealed at one end and was then
electroplated with the other FBG with a Bragg wavelength of 1300.102 nm on a
150 mm x 12.7 mm x 9.2 mm stainless steel beam. The thickness of the electro-
plated nickel layer was controlled to be approximately 2.0 mm.
To calibrate the decoupling sensor, the sensor was first placed in an oven for
thermal tests, and then it was tested under four-point bending at room tempera-
ture. Wavelength shifts from both FBGs were measured. Figure 4.13 shows the
test results. Clearly, both FBGs responded to thermal loads, but with different
sensitivity. The thermal sensitivity of the embedded FBG is in good agreement
with that of previously characterized Ni-coated FBG. Moreover, the free-float-
ing FBG behaved in the same way as a bare FBG under thermal loads. However,
when the beam was under strain loads, the free-floating FBG showed no wave-
length shifts while the embedded FBG acted almost the same as the Ni-coated
FBGs tested previously.
The different behavior for the two FBGs in this decoupling sensor suggests
that this sensor is capable of separating the effects of temperature and strain.
This method can, hence, be used to monitor temperature and strain in a func-
tional structure during manufacturing in-situ or during service of the structure.
47
4.5
4 y = 0.021x - 0.0047
3.5
Wavelength shift, nm
3
Free-floated
2.5
FBG
2
Embedded FBG
1.5
1
y = 0.0099x - 0.011
0.5
0
0 50 100 150 200 250
T emperature change, C
1.1
0.7 Free-floated
FBG
0.5
Embedded
0.3 FBG
-0.1
0.0 200.0 400.0 600.0 800.0 1000.0
Strain, x1e-6
Fig.4.13. Calibration of decoupling sensor under thermal and axial strain loads
48
were mounted, mechanically polished down to 1µm diamond paste, and etched
in Marble's reagent, a mixture of 4g CuSO 4 , 20ml HCl, and 20ml water. A light
etch easily revealed the subgrains of the deposited stainless steel, but nickel is
slowly attacked by the dilute HCl in Marble's reagent. Further etching was
therefore needed to reveal the nickel microstructure as well as to allow the use
of polarized light during optical microscopy.
Figure 4.14 shows a typical microstructure across the entire embedding struc-
ture, starting with the layers from the stainless steel substrate at the bottom and
ending with the deposited stainless steel on the top. The sample was heavily
etched for this photo to show the nickel microstructure. The boundary between
the stainless steel substrate and the electrolytically deposited nickel was very
distinct, showing little evidence of inter-diffusion at this boundary. The depos-
ited stainless steel/nickel interface was layered and uneven, showing evidence
of diffusion.
When lightly etched with Marble's reagent, the subgrains of the deposited
stainless steel were revealed. The grain structure appeared fibrous, with long,
parallel rods, and is known as a coarse dendritic grain structure. The direction of
growth is perpendicular to the deposited surface in Figure 4.15. The dendrites
are typical of a cast or deposited material, and the grains reach lengths of 100 to
several hundred microns. Deposited stainless steel etched in Marble's reagent
shows the long, parallel cellular packets, or dendritic structure, of the subgrains.
Two of the grains have been outlined for easy viewing.
The electrolytically deposited nickel grains are still fine (~50µm diameter)
and exhibit equiaxial shape close to the nickel/SS substrate interface, as shown
in Figure 4.16. They gradually become large and irregular in the interior of the
sample, as shown in Figure 4.17.
Figure 4.18 shows that the grain size of the nickel is decreasing from about
200 µm near the interface between the laser-deposited stainless steel (at right)
and the nickel (at left) to about 10 µm at a distance of 3 mm. Clearly, there is a
thermal effect to the originally fine grains of the electroplated nickel during the
49
laser-cladding process. Obviously, the electroplated nickel recrystallizes during
and after the laser cladding because the laser melts part of the nickel layer and
mixes it with molten stainless steel powder.
Nickel
Optical fiber
50
Fig.4.16. Small Ni grains close to the interface between nickel and stainless
steel substrate
51
Microstructures of nickel
after embedding
1200 µm
52
4.4.2. Strain response
Ni-coated FBGs transferred strains well according to previous characterization
However it still remains uncertain how well the FBGs embedded in stainless
steel structures will perform. Particularly, we do not know the extent to which
thermal stresses and residual stresses from laser cladding will impair bonding
between the nickel layer and the stainless steel substrate, between the optical
fiber and nickel layer, and between the nickel layer and laser cladded stainless
steel layer. From the micrographs of the embedding structures, no obvious
delamination was observed.
53
F F Strain gauge
40
80
SS304 19.73
13.0
Ni
3.7
1.6
2.0 10.8
1.6
Fig.4.19. Setup of four-point bending test and cross section of the beam
Figure 4.20 shows the results when the beam was still in the elastic regime.
Strain readings and wavelength shifts were plotted in comparison with nominal
strains at the location of the FBG sensor. The nominal strains were calculated
from elastic beam theory since the displacement and the load from the 4PBT rig
were recorded. Best-fit lines are drawn to show the correlation with the nominal
strains.
The strains from the strain gauge, which was placed on the top surface of the
beam, are linear to the nominal strains with a coefficient of 3.1451, which is in
good agreement with the predicted value of 3.1765, which is the ratio of the dis-
tances of the strain gauge and the FBG from the neutral axis according to elastic
beam theory. Moreover, the embedded FBG demonstrates a strain sensitivity of
1.245x10 -3 nm/µm, which is close to the 1.296x10 -3 nm/µm predicted from the-
54
ory. We can also calculated the “measured strain” in the FBG location from the
strain gauge measurement. The “measured strain” in FBG location is plotted
against the FBG wavelength shift, as shown in Figure 4.21. The sensitivity is
shown as 1.2245x10 -3 nm/µm, which is also in good agreement with the theoret-
Elastic regime
-1600 -1
-1400 y = 3.145E+00x + 1.452E+00
Strain from Strain
Wavelength shift,
R2 = 9.999E-01 -0.8
-1200
gauge, 1e-6
-1000 -0.6
nm
-800
-600 -0.4
-400 y = 1.245E-03x - 4.405E-03
-0.2
-200 R2 = 9.980E-01
0 0
0 -100 -200 -300 -400 -500 -600
Norminal strain at FBG sensor location, 1e-6
-2
y = 1.2245E-03x - 8.5403E-03
Wavelength shifts, nm
2
-1.5 R = 9.9912E-01
-1
-0.5
0
0 -200 -400 -600 -800 -1000 -1200 -1400
Measured Strain (corrected from the strain gauge for the
FBG location), 1e-6
Fig.4.21. Measured strain (corrected from the strain gauge for the FBG location)
versus FBG wavelength shifts
55
Figure 4.22. presents the wavelength shifts against the strain readings from the
strain gauge. The higher strain readings, up to 5700 µm, show that the beam was
in plastic deformation. The wavelength shifts are linear to the strains through all
elastic and plastic regimes. The linear correlation suggests that the embedded
FBG functioned well for strain measurements.
56
measured by the strain gauge on the top surface. To obtain the correlation
between wavelength shifts and the theoretical strains, a best-fit line was used.
Figure 4.25 presents a typical test result for an FBG with a Bragg wavelength of
1551.114 nm. The sensitivity of 1.215x10 -3 nm/µε matches that of the bare FBG
well. The test results suggest that the bonding between polyurethane and the
optical fibers is strong and that FBGs embedded in polyurethane are capable of
measuring strains accurately.
14.000
Wavelength shift, nm
40
3 4.55 6
FBG
Strain Gauge
80
57
Polyurethane embedded FBG
6
Wavelength shift, nm
y = 6.075E-04x + 1.642E-03
5 R2 = 9.998E-01
4
3
2
1
0
0 2000 4000 6000 8000 10000
Strain, 1e-6
4.7. Summary
Fiber Bragg Grating sensors were embedded in metallic, nickel and stainless
steel, polymers, and ceramic structures. The embedded FBG sensors were char-
acterized for temperature and strain measurements.
58
The embedded FBG sensors in nickel and stainless steel provided high sensi-
tivity, good accuracy, and high temperature capacity for temperature measure-
ments. A temperature sensitivity more than 100% higher than that of bare FBGs
was demonstrated. The increased sensitivity was mainly due to the higher CTE
of nickel than that of the optical fiber. For temperature measurements, the
embedded FBG sensor yielded good accuracy and linearity. A main cause for
errors would be possible temperature gradients in the metallic structures if the
embedded sensors are under rapid thermal cycles.
However, sapphire fiber were not able to maintain their integrity in alumina.
We believe that the shrinkage and stresses during sintering damaged the fibers.
59
Chapter 5: Applications of Embedded
Sensors
This chapter investigates the potential applications of embedded Fiber Bragg
Grating sensors. The embedded sensors can be used to monitor the accumulation
of residual strains during the laser-assisted Shape Deposition Manufacturing.
They can also be used to measure the strain field in layered materials and to
measure temperature and strain simultaneously during Shape Deposition Manu-
facturing. Finally, embedded sensors can be used for pressure sensing.
We made samples with embedded FBGs to measure strains. Each FBG was
embedded in a 2.8 mm thick nickel layer, which was electroplated on stainless
steel beams with a dimension of 9.3 mm x 24.5 mm x 127 mm. The FBG was
carefully placed in the middle of the beam. The beam was bolted down on an
aluminum plate. A laser was then used to deposit layers of stainless steel on top
60
of the embedded sensor. Figure 5.1 depicts the process. Powders were preplaced
on the substrate before laser was turned on.
la se r
F B G S lo c a tio n
Ni
To monitor residual strains during the laser layered deposition, a serial set of
paths that the laser followed to deposit the material were designed. Figure 5.2
shows the paths, which are projected for better viewing. The powders are not
shown. The arrowhead solid lines are laser melting paths, and the dashed lines
are paths that laser ia made to scan at a high enough speed (300 mm/s) to keep it
from creating a melting pool. After it scanned every set of paths, the laser was
stopped, and the stainless steel beam was cooled to room temperature by air. The
Bragg wavelength was then measured by OSA. During the process, no birefrin-
gence-induced two-peaks were found. This might due to the significantly less
magnitude in transverse strains or little difference in transverse strain compo-
nents ε 2 and ε 3 . Alexander Nickel [38] applied Finite Element Analysis to calcu-
late the residual stresses during laser-assisted layered manufacturing and
showed that ε 2 and ε 3 are almost 1 order smaller than ε 1 in beam tests. Moreover,
the sensors were embedded 2.8 mm away from the first laser-deposited layer.
Elastic behavior in the sensor area was analytically predicted. Thus, the wave-
length shifts measured might be used to determine ε 1 approximately.
61
Path set 9
Path set 8
Path set 7
Path set 6
Path set 5
Path set 4
Path set 3
Fig.5.2. Path sets of the laser cladding on the stainless steel beam
Figure 5.3 presents the wavelength shifts due to residual strains generated by
laser cladding after each path set. Obviously, all residual strains are in compres-
sion. Since the laser-deposited layers were at a much higher temperature than
the nickel layer, more compressive plastic deformation occurred in the laser-
deposited layers due to constraints by the cooler nickel layer and substrate.
After cooling, the laser-deposited layers were thus in tension, and a compressive
strain was generated in the nickel layer, in which the sensors were embedded. It
should be noted that residual stresses generated during any Layered Manufactur-
ing occur in highly complicated 3-D manner.
62
5.2. Strain analysis with two embedded sensors
In additive layered processes, flexure and simple extension can coexist in a
layered beam. With one embedded sensor, it is difficult, if not possible, to dis-
tinguish the strains introduced by pure flexure from those introduced by simple
extension. Thus we proposed a two-sensor technique and tested it to verify the
concept. Figure 5.4 presents a configuration of two embedded FBG sensors for
strain analysis of a beam. Y 1 , y 2 , and τ are the distance from the bottom to
FBG1, FBG2, and the neutral axis respectively.
-0.1
-0.2
-0.3
-0.4
-0.5
-0.6
-0.7
-0.8
-0.9
FBG1
Neutral axis
y1
τ y x
2
FBG2
63
If the beam is under unknown bending and stretching, a theoretical strain at
any point in the beam can be expressed as
ε xx = ε bxx ( flexure ) + ε txx ( extension ) (5.1)
( y1 – τ ) (5.2)
ε 1xx = ------------------ + ε txx
R
and in FBG2 by
( y2 – τ ) (5.3)
ε 2xx = ------------------ + ε txx
R
where R is the curvature of the beam under bending. From the equations of ε 1xx
and ε 2xx , we get
( y1 – y2 )
R = ------------------------------ (5.4))
( ε 1xx – ε 2xx )
Sensors FBG1 and FBG2 can be used to measure wavelength shifts, using the
equation
∆λ
ε = ------------------------- (5.5)
λ0 ( 1 – pe )
We can therefore determine ε 1xx and ε 2xx . The curvature R can then be calcu-
lated through the equation
( y1 – y2 )
R = ------------------------------------------------------------------ (5.6)
∆λ 1
-------------------------- ∆λ 2
- -------------------------- -
λ 01 ( 1 – p e ) – λ02 ( 1 – p e )
With known R, we can determine the axial strain both in flexure and simple
extension at any point in the beam. To verify the concept, two FBG sensors were
embedded in polyurethane and two were embedded in a metallic structure.
64
5.2.1. Two sensors in polyurethane
Two FBGs, one with an original Bragg wavelength of 1551.123 nm and the
other with an original Bragg wavelength of 1555.457 nm, were embedded in a
100 mm x 12.7 mm x 6 mm polyurethane beam. Y 1 and y 2 were set at 4.625 mm
and 1.625 mm, respectively, and we used four-point bending to applied loads.
OSA was used to measure the Bragg wavelengths. To verify the calculated R
from the Bragg wavelength shifts in the two FBGs, we used a mechanical dial
with a resolution of 12.7 µm to measure deflection at three different points. Fig-
ure 5.5 presents the principle of determining R through δ deflection and L, which
is the distance between the two outside points. R can be given by
2
L --- + δ 2deflection
2 (5.7)
R = ----------------------------------------------
2δ deflection
R R
L 2
R = --- + ( R – δ deflection )
2 2
L/2 2
δdeflection
Beam
Figure 5.6 shows the results of the Bragg wavelength shifts from the two FBGs
against δ deflection , which is called maximum deflection in the figure. The wave-
length shifts from both FBGs is approximately linear with δ deflection .
65
Wavelength shifts
Wavelength shift, nm
15 y = 2.6909x + 0.1323
10 R2 = 0.9939
5
w_shift1, nm
0
w_shift2, nm
-5
y = -3.1166x - 0.2035
-10 R2 = 0.9957
-15
0 1 2 3 4
Maximum Deflection, mm
6000
Radius of culvature, mm
R_dial, mm
5000
R_sensor
4000
3000
2000
1000
0
0 0.5 1 1.5 2 2.5 3 3.5
Deflection, mm
66
5.2.2. Two sensors in metallic structures
Next we tested the two-sensor technique in metallic structures. Two FBGs, one
with an original Bragg wavelength of 1300.014 nm and the other with an origi-
nal Bragg wavelength of 1300.017 nm, were embedded in a 100 mm x 12.7 mm
x 6 mm stainless steel beam, as shown in Figure 5.8. Four-point bending was
used to applied loads. As shown in Figure 5.9, the comparison of the R’s mea-
sured by the two-sensor technique with those measured by the mechanical dial
technique indicated good agreement, with an error less than 3.6%.
FB G 1
1 .1 0 m m th i c k N i
S ta in le s s s te e l 4 .6 5
4 .6 5
1 .1 0 m m th ic k N i
FB G 2
5000 R_dial, mm
Radius of culvature, mm
R_sensor
4000
3000
2000
1000
0
0 0.5 1 1.5 2 2.5 3 3.5
Deflection, mm
67
5.3. Embedded Sensors for Pressure Measurements
With embedded sensors, it is possible to develop a low-cost fabrication meth-
odology for the production of differential pressure sensors that function just as
well as currently available structures with built-in semiconductor membranes.
The basic concept is a tubular vessel with a membrane embedded. The mem-
brane has an FBG embedded close to the neutral axis to minimize the influence
of bending. The selection of suitable materials for diaphragms with embedded
strain gauges is crucial for the sensor’s performance.
∆P
68
where ν is Poisson's ratio and E is Young's modulus for the diaphragm. The
maximum strain at the circumference is also linear by relate to the pressure dif-
ference
3 ( 1 + υ )∆ P a 2
ε max = ----------------------------- --- (5.9)
8E t
The strain causes the Bragg wavelength to shift in the Fiber Bragg Grating
sensor, which is embedded in the thin diaphragm. The wavelength shift is deter-
mined by
∆λ = λ 0 ( 1 – p e )ε max (5.11)
69
gauge was embedded. In the other one, both a strain gauge and an FBG with an
original Bragg wavelength of 1533.478 nm were embedded. Since an optical
fiber has a diameter of 125 µm and a thin-film strain gauge is approximately 50
µm thick, the sensing point was actually slightly above the neutral axis.
Figure 5.11 presents the results from the testing of the pressure sensor with the
a/t ratio of 15. Strain is plotted against pressure. In the lower pressure region, a
linear relationship between pressure and strain is obtained. Approximately a
pressure of 105.91 Pa induces a strain of 1.0 µε, as shown in Figure 5.11(a).
However, the ratio of pressure to induced strain becomes larger in a higher pres-
sure range, as shown in Figure 5.11(b). The pressure sensor failed when the
pressure was raised to 3.5x10 5 Pa (3.5 bar) and the induced strain was 2229 µε.
Testing results of the embedded strain gauge for the pressure sensor with the a/
t ratio of 25 are shown in Figure 5.12. In the lower pressure region, a linear rela-
tionship between pressure and strain is obtained. Approximately a pressure of
37.676 Pa induces a strain of 1.0 µε, as shown in Figure 5.12(a). Again, the ratio
of pressure to induced strain seems to be larger in the higher pressure range, as
shown in Figure 5.12(b). The pressure sensor functioned well at a pressure of
1.6x10 5 Pa (1.6 bar) when the induced strain reached 3445 µε. This suggests that
pressure sensors with a higher a/t ratio can endure larger strain.
Embedded strain gauges function well for the pressure sensors. Embedded
FBGs were also tested for pressure measurements. Wavelength shifts were mea-
sured at different pressures, and the wavelength shift was plotted against the
strain measured by the embedded strain gauge, as shown in Figure 5.13. Wave-
length shifts follow the strain, thus pressure, linearly, at least for higher strains.
70
ing and after the layered deposition. With the decoupling technique described in
Chapter 4, we can monitor temperature and axial strain simultaneously by use of
two embedded FBGs. To determine 3-D strain components and temperature
simultaneously, two embedded FBGs oriented in different directions along with
a free-floating FBG might be necessary.
3.5E+04
3.0E+04
2.5E+04
Pressure, Pa
2.0E+04
1.5E+04
1.0E+04 y = 105.91x
R2 = 0.9986
5.0E+03
0.0E+00
0 100 200 300 400
Strain, ue
Pressure Vs Strain
for diam ter=30m m , thickness=1m m
4.0E+05
3.5E+05
Failure point
3.0E+05
Pressure, Pa
2.5E+05
2.0E+05
1.5E+05
1.0E+05
5.0E+04
0.0E+00
0 500 1000 1500 2000 2500
Strain, ue
71
Low e r pr e s s ur e r e gion
for diam e te r =50m m , thick ne s s =1m m
5.0E+04
4.0E+04
Pressure, Pa
3.0E+04
2.0E+04
1.0E+04 y = 37.676x
R2 = 0.9999
0.0E+00
0 200 400 600 800 1000 1200
Str ain, ue
1.8E+05
1.6E+05
1.4E+05
Pressure, Pa
1.2E+05
1.0E+05
8.0E+04
6.0E+04
4.0E+04
2.0E+04
0.0E+00
0 1000 2000 3000 4000
Strain, ue
72
Embedded FBG vs embedded strain gauge
Wavelength shift, nm
5 y = 0.00158x + 0.07184
4 R2 = 0.99892
3
2
1
0
0 1000 2000 3000
Strain from strain gauge, 1e-6
To monitor the strain and temperature in situ, we need a higher sampling fre-
quency than an Optical Spectrum Analyzer can provide since the thermal and
induced effects during Layered Manufacturing are transient. Thus a high-speed,
low-cost system was used to measure Bragg wavelength shifts of the two FBGs
simultaneously. Figure 5.14 depicts the new sensing system. A broad-band LED
emits light through a 2x2 fiber coupler and onto a 1x2 optical switch, whose out-
put points are connected to the two embedded FBGs. At every single moment,
the optical switch allows only one FBG to reflect light at its Bragg wavelength.
The reflected light then goes back through the 2x2 fiber coupler and onto a sec-
ond 2x2 fiber coupler, which divides the reflected light into two streams. In one
stream, a chirped grating, which is a wavelength-dependent filter, is used to fil-
ter the light. The filtered light then goes onto a photodetector. In the other
stream, light goes onto another photodetector directly and functions as a refer-
ence. The Bragg wavelength can be determined from the ratio of the signals
from the two detectors.
73
1x2
1300 nm 2x2 optical
Broad band LED fiber switch FBG 1
coupler
FBG 2
Filter function
Three different
Bragg wavelengths
transmitted through
the chirped grating
Chirped Detector 1
grating
A/D PC
2 x 2 fiber coupler
Detector 2
Figure 5.15 presents the spectrum curves of the LED, the direct output of the
chirped grating, and a normalized version of the chirped grating. Normalization
is important because the non-flat shape of the LED introduces nonlinearity into
the direct output curve for the chirped grating. The slope of the normalized
curve is determined from the curve in terms of the rise in percentage over the
run in nanometers. The slope is valid in a range of 9.611% to 89.84%, which
corresponds to 15.325 nm. Although the trend of the normalized curve is quite
linear, some fine fluctuations in the curve make the exact measurement at cer-
tain points difficult.
Since the FBGs used in this study come with a bandwidth larger than 0.15 nm,
some averaging of these fine fluctuations occurs during the measurements,
which improves the accuracy of the results. By careful calibration, a resolution
of 0.1 nm for the measurement of Bragg wavelengths can be achieved.
Three layers of stainless steel were deposited on top of beams with the decou-
pling sensors embedded. The dimensions of the beams are shown in Chapter 4.
74
Laser power were set to be 1920 W, and the scanning speed was 20 mm/s. After
each layer (1.1 mm thick), laser was turned off, and the beam was cooled to
room temperature naturally by air. Wavelength shifts were monitored for all
three layers in situ and plotted against time, as shown in Figure 5.16.
LED
Normalized
version of
the filter
Direct output
from the chirped grating
In all three layers, the wavelength shifts from the free-floating FBG, which
responds to temperature effects only, gradually increased when the laser was
scanning from one end to the other end of the beams. Peaks were found when the
laser was scanning on the surface immediately above the embedded sensors.
Then, the wavelength shifts are decreasing since the laser was scanning on the
surface that was further away from the sensors. After laser deposition, the air
cooled the beam, and the Bragg wavelength from the free-floating FBG gradu-
ally shifts back to its original one. Furthermore, the magnitude of the peaks in
75
wavelength shifts is decreasing from layer 1 to layer 3. This is because each
layer deposited on the surface of the beams increases the distance between the
sensors and the laser scanned surface. The experimental results verify that the
free-floating FBG can be used to determine temperatures in situ.
The wavelength shifts from the embedded FBG behave differently. For each
layer, a residual wavelength shift was found after the beam cooled to room tem-
perature, showing residual strains were induced by the laser deposition process.
In layer 1, residual wavelength shift is about -1.2 nm. However, much less resid-
ual wavelength shifts are induced in layer 2, -0.5 nm, and in layer 3, -0.1 nm.
Since the longitudinal strain in the beams dominates, the wavelength shifts can
be used to determine the magnitude of the strain approximately. Apparently, the
layer 1 contributed the largest residual strain while layer 3 induced the least
residual strain in the embedded sensors. This phenomena was also predicted
analytically and observed experimentally in Alexander Nickel’s dissertation
[38].
5.5. Summary
This chapter investigated the potential applications of the embedded FBG sen-
sors. The embedded sensors were used to monitor residual stresses during the
laser-assisted Shape Deposition Manufacturing. Embedded sensors were also
used to measure the strain field in layered materials. Pressure sensors based on
embedded strain gauges and FBGs were developed. The embedded sensors were
capable of monitoring temperature and strain simultaneously during laser-
assisted Layered Manufacturing.
76
8
7 Layer 1
Wavelength shift, nm
6
5
Temperature and Strain
4
3
2
Temperature only
1
0
-1 0 200 400 600 800 1000
-2
Time, s
8
Layer 2
Wavelength shift, nm
0
0 200 400 600 800 1000
-2
Time,s
8
Layer 3
Wavelength shift, nm
0
0 200 400 600 800 1000
-2
Time, s
77
Chapter 6: Optical Sensing for Rotating
Components
This chapter presents the development of new techniques of optically sensing
the temperature and strain of rotating components by use of embedded Fiber
Bragg Grating (FBG) sensors.
6.1. Introduction
In industry, a great deal of testing involves temperature and strain analysis for
rotary components, such as wheels, rotors, and blades. Particularly, a strong
motivation behind efforts to detect the temperature and strain patterns in the tur-
bine blades comes from the need to design blades that withstand higher combus-
tion temperatures. Engineers test and analyze new designs and conduct tests to
verify the quality of products coming off production lines. They may also run
computer simulations through finite element analysis and then compare those
results with actual values from physical tests under various loading conditions.
However, it is difficult to get data off rotating components from mounted or
embedded sensors.
Slip rings, with brushes connecting rotating sensors to the data collection unit,
are commonly used to collect data from spinning components. Unfortunately,
slip rings present several potential problems. Noise, limitation in the possible
number of simultaneous channels of data, and heavy weight are among the dom-
inant ones [100]. Electrical noise generated by the sliding brushes can corrupt
the data being collected. The number of channels that can be used at any one
time to collect data is limited by the fact that slip ring cannot become too large.
Thus engineers have to perform two or more test runs if a large number of strain
78
gauges or thermocouples are used. Moreover, multi-channel slip rings are gener-
ally heavy, which also limits their usefulness. These reasons are why most
research groups have employed stationary flat or mildly curved plates to study
effects of cooling and have obtained data by wire thermocouples when they have
studied turbine blades [101,102]. Telemetry systems have been investigated to
solve the problem [100]. However, it is difficult to implement an electronic sys-
tem in a hostile environment, such as high temperatures. Furthermore, telemetry
systems are also prone to electromagnetic interference.
Given the difficulties, the appeal of fiber optic sensors to collect strain and
temperature data in a non-contact mode is obvious. Many Fiber Bragg Gratings
can be inscribed in a single optical fiber. Thus, multiplexing becomes conve-
nient and the weight of all the sensing elements can be less than a few grams.
Such a system will be especially useful when the task is to obtain temperature
and strain information from rotating parts such as turbine blades, which are
working in a hostile environment.
79
PC DAQ
system Embedded fiber optic sensor
Rotating metallic
(Etalon or Fiber bragg grating)
parts
Photo-
detector
EDF
EDF
Optical Embedded optical
Wavelength Alignment
isolator fiber lead
Tunable laser Lens
Fig.6.1. Sensing system for rotating parts with embedded FBG sensors
At the Bragg wavelength, the intensity of the transmitted light drops to almost
zero. The load, which is linear to the Bragg wavelength shift, can then be deter-
mined. A PC data acquisition system can be used to collect data from the photo-
detector and from the tunable laser diode. This sensing system also works when
FBG sensors are mounted on the surfaces of rotating components.
The reason for using a tunable laser diode is to ensure an absolute measure-
ment since the measurands are wavelength-coded. In order to avoid the reflected
light from the surfaces of rotating components, it is better to collect the trans-
mitted light instead of the reflected light from the FBG sensor. The input light is
aligned, and since the spot of the focus on the surface of a rotary component is
larger than the optical fiber size, greater tolerance in alignment is allowed.
Greater tolerance prevents misalignment due to possible vibration generated by
the rapid spinning of the component.
80
was embedded into a stainless steel beam and a non-contact sensing system was
set up as shown in Figure 6.2.
The end surfaces of the embedded optical fiber leads were polished to increase
the transmission of light to the photodetector. A gap of about 30 mm was set
between the end surface of the beam and a pigtailed fiber collimator. The model
of the collimator was Newport F-COL-9-15-OPT-50 (single mode 9/125 µm with a
center wavelength of 1550 nm and an FC/SPC connector). The features of these colli-
mators include low insertion loss, low back reflection, a metal housing sleeve,
and a strain relief boot. The diameter of the collimated beam was as small as 450
µm. The pigtailed fiber collimators were mounted on optical mounts, which
were placed on an x-y-z microstation.The resolution of the microstation was 1
µm.
Commercially available external cavity diode lasers (ECDL) provide large and
continuous wavelength tuning and narrow line width (< 1MHz), which makes
them ideal for wavelength selections. In this work, an InGaAsP ECDL
(1.497~1.582 µm , New Focus Inc., Model 6200) was used to transmit the spec-
trally controlled light to the pigtailed fiber optic collimator. This ECDL had lin-
81
ewidths of approximately 300 kHz and operated single mode throughout the
tuning range. The principle of the external cavity diode laser is schematically
shown in Figure 6.3. A Fabry-Perot diode laser is anti-reflection coated on the
output facet to suppress oscillation between facets and is placed into a grazing
incidence Littman-Metcalf external cavity. The grating acts as a frequency
selective output coupler since the angle of the retroreflector determines the out-
put wavelength. A piezoelectrically controlled motor enables tuning over the
full range at a rate of up to 40 nm/min. In addition, a piezoelectric actuator is
used for wavelength fine tuning and rapid scanning over a small wavelength
range. The output light power can be controlled up to 2.2 mW. The voltage out-
put is linear to the wavelength of laser light and can be accessed through a BNC
connector. Figure 6.4. shows how the voltage responds to the wavelength. The
relationship can be described as
Motor
Laser diode
Piezoelectric actuator
Retroreflector
Collimating
lens
Laser output
82
1565
1560
1550
1545
1540
1535
1530
1525
2 2.5 3 3.5 4
Voltage from ECDL, V
Figure 6.5 shows a typical waveform of the voltages from the photo-detector
and the laser diode when the ECDL was tuning continuously. At room tempera-
ture, 22.1°C, the wavelength was found to be 1534.44 nm, which was obviously
shifted from the original Bragg wavelength of 1536.28nm by residual stresses
that resulted from laser deposition. This wavelength shift was verified with opti-
cal spectrum analyzer (OSA). When the temperature of the beam was gradually
raised to 85.35°C (measured by a K-type thermocouple), the Bragg wavelength
shifted to 1536.46 nm.
Initial experimental results showed that the proposed system was capable of
measuring loads on embedded fiber Bragg gratings in a non-contact fashion.
However, the signal level in the photodetector was very weak even when the
gain knob was set at 10M. Clearly, there was a need to identify what causes light
loss and then to solve the problem by implementing a new optical system that
eliminated or lessened them.
83
Voltage from photo-dector, V
2.5 0.012
2.3 0.006
0.003
2.1 0
0 5 10 15
Scanning time, s
A simple model, schematically depicted in Figure 6.6, shows how light loss
occurs. Since the diameters of the input light beam and optical fiber core are 450
µm and 9 µm respectively, the light intensity will theoretically drop by
10log(450/9) 2 dB, about 34 dB. When light is transmitted from the optical fiber
into the output collimator, the light loss will be determined by the distance, L,
and the aperture, sinθ, of the optical fiber. Given a 0.2 aperture for a single
mode fiber and an L of 20 mm, the light loss is approximately (by rule of thumb)
10log(Ltanθ/0.45) 2 or about 25dB. Since the photo sensitive area in the photo-
detector is about 100 µm, there is more light loss due to inadequate coupling
84
between the output collimator and the photodetector. Hence total light losses in
the sensing system can be up to 75 dB.
Clearly, the coupling efficiency must be improved. Figure 6.7 presents a new
optical system to solve the problem. Focuser 1 was designed to provide adequate
light coupling between the single-mode input fiber, which was connected to the
laser diode, and the embedded optical fiber. OZ Optics, Ltd. assembled the 0.79”
OD single-mode 1550 nm focuser with FC receptacle and desired aspheric. The
device, model HPUFO-23-1550-S-4-19.5-3.9AS, yielded a magnification of 4
and a 19.5 mm working distance. The focus spot diameter was set at 36 µm in
order to allow greater tolerance to the light coupling, which was crucial to offset
the vibrations while the part was rotating. Focuser 2 was used as a light collec-
tor instead of a focuser. The light coming from the sensor fiber core was col-
lected by a 0.79” OD focuser with FC receptacle and desired achromat, yielding
a magnification of 1 and a 40mm working distance. Again Focuser2, model
HPUFO-23-1550-S-1-40-20AC, was made by OZ Optics, Ltd. A multi-mode
fiber with a 100 µm core was connected to the receptacle to provide more toler-
ance for light coupling. The detected light loss was measured to be about 20 dB,
much better than the 75 dB light loss in the collimator-based light coupling sys-
tem.
85
plate was then assembled with a stepper motor, which could rotate up to 2200
rpm with adequate current input. Figure 6.9 presents the assembled rotary struc-
ture.
distance L
Output Output
θ 9 µm sensor fiber core
collimator light
beam
Focuser 2
(f=20mm, magnification=1, 40mm working distance)
Focuser 1
(f=3.9mm, magnification=4, 19.5mm working distance)
86
Symmetric mount with hole
(to be assembled with a motor) Rotating
axis
4 mm
e ad
fiber l
l
tica
Op G
FB
40
m mm
m 115
Fig.6.8 Geometric parameters for the rotary plate with embedded FBG
87
6.4.2. Characteristics of thermal response in stationary testing
In order to calibrate the embedded FBG sensors, the new optical coupling sys-
tem (Figure 6.7) was implemented to perform stationary testing under thermal
loads. The rotary plate was placed on top of a hot plate and the temperature was
controlled. Insulation materials were used to form an oven-like small cavity on
top of the hot plate. Four thermal couples were used to ensure a temperature dif-
ference of less than 0.2°C between the two surfaces of the rotary plate. Figure
6.10 shows the calibrated results. As predicted by the theoretical model for
embedded FBG sensors, the wavelength shifts were linear with temperature
changes.
18
16 y = 0.033713x - 0.107070
Wavelength shift, nm
14 R2 = 0.999669
12
10
8
6
4
2
0
0 100 200 300 400 500
Temperature change, C
88
6.4.3. Signal characteristics for rotating plates
When the plate is spinning, the laser beam is aligned with the embedded opti-
cal fiber lead for a period whose length depends on the spinning speed. Thus
pulse signals appear. The pulse signals contain the temperature and strain infor-
mation. Assuming Focuser 2 collects all the light coming out the sensor fiber
core in Figure 6.7, the characteristics of the pulse signals will be determined by
the coupling between the stationary Focuser 1 and the spinning fiber core. An
analogy is shown in Figure 6.11. Since the laser light is coming out of the sin-
gle-mode fiber, the light intensity at the focus spot (φ=36µm) after Focuser 1
should have a Gaussian distribution. Thus the pulse width t w and pulse level P l
can be approximately given in terms of speed ω, spinning radius r, and time t by
( 18 µ m – t ω r )
2
– --------------------------------------
2σ
2
36µ m P0
t w =-------------- P l = -------------- e (6.2)
ωr 2πσ
where t is the time and P 0 is a nominal parameter for the pulse level. Time t
starts at the moment the spinning smaller circle, φ=9 µm, enters the φ=36µm cir-
cle, and it stops at t w.
φ=36µm
φ=9µm
ω
r
Fig.6.11. Analogy to pulse generation for a spinning plate with embedded FBG
89
on the order of tens of microseconds for a spinning speed ranging from 60 rpm
to 2000 rpm. This is an important parameter from which one can determine the
sampling frequency for the data acquisition system. Moreover, no matter how
much the shape of the pulse changes due to the parameter t, the maximum pulse
level (when t ω r=18 µ m ) stays constant for all spinning speeds.
For the manufactured rotary plates in this study, r was set at 20 mm. However,
the two embedded fiber leads were placed slightly off the symmetric positions
along the rotating axis, so only one pulse was generated for one turn. If the
wavelength of the laser diode stays constant and the plate is rotating, the signals
in the photodetector can be depicted as in Figure 6.12.
Another interesting thing seen in the equation is that t w can be infinitely large
if the spinning radius is 0. This suggests that temperature and strain measure-
ments for rotating objects can be performed similar to the way stationary testing
is done if the embedded or attached fiber leads are placed on the rotating axis. In
this case, we can use collimators to make a long-range remote sensing system as
depicted in Figure 6.13.
Pulse signal
λ
t t
Fig.6.12. Signals for a constant wavelength laser beam during plate spinning
Since the fiber optic lead is placed on the rotating axis, a GRIN collimator,
which is connected to the embedded or attached optical fiber lead, receives the
input light from the input collimator in a fashion similar to that of stationary
optical coupling. Since there is no significant reflection from the optical cou-
90
pling system to disturb the reflected light at the Bragg wavelength, a photode-
tector can be used to collect the reflected light instead of the transmitted light
from the FBG sensors. Furthermore, it is highly advantageous to implement
multiplexing techniques to monitor temperature and strain at different points by
inscribing different Fiber Bragg Gratings in a single optical fiber. Multiplexing
results in a sensing system that is much lighter than the multi-channel slip ring
system traditionally used for these types of measurements.
Input collimator
GRIN collimator
Optical coupler FBG1 FBG2 FBG3...
Rotating axis
Photodetector
Time, t
Fig.6.13. Optical sensing system with optical fiber leads on rotating axis
Experiments were carried out to determine the pulse width and pulse level
under different spinning speeds for the non-contact sensing system. By fixing
the light from the ECDL at a wavelength of 1550 nm and a laser power of 1.8
mW, we found the pulse level and pulse width shown in Figure 6.14. The pulse
width was measured at the 50% pulse level. The experimental values for t w were
on the same order of magnitude as the theoretical ones. However, the experi-
mental data showed that higher spinning speeds led to lower maximum pulse
91
levels, while the theoretical model suggests that the maximum pulse level is
constant and only the pulse shape changes at different spinning speeds. The rea-
son for the discrepancy was that the sensitivity of the photodetector diminished
due to a pass filter effect created by the parallel gain resistor and the photodiode
capacitance. This was especially true at higher gain settings for the photodetec-
tor.
400 100
50% pulse width, us
Pulse signal voltage, m
80
Pulse width tw, u
0 0
0 500 1000 1500
Rotating speed, rpm
92
Pulse signal
λ
Pulse signal
λ
The strategy for the low-speed sensing is to detect missing pulses while tuning
the diode laser continuously. The time at the moment when pulses are missing
(or significantly compressed) is used to track the laser wavelength in order to
determine the Bragg wavelength. The strategy for the high-speed sensing is to
tune a laser diode at a very high speed so that one tuning cycle is completed in a
period shorter than the pulse width. The time at the moment corresponding to
the Bragg wavelength is then determined.
93
low. This technique is suitable for measuring low-frequency temperature or
strain loads with a tunable laser diode with a lower tuning speed, such as ECDL.
Experiments with ECDL were performed to test the thermal response of the
rotating plates. A heating gun was used to heat the plate while the plate was
rotating. For the results shown in Figure 6.16, the rotating speed was set at 600
rpm, i.e., the time for a pulse was 0.1 second. On the chart the sixth pulse was
significantly lower than the others due to the reflection of the FBG sensor at the
Bragg wavelength. Then the Bragg wavelength shift was determined, and thus
the temperature.
Since the sampling frequency was very high (normally > 100KHz) due to the
short period, i.e., small pulse width, there were many data points for drawing
one chart like that of Figure 6.16. But since only the pulse signal contains tem-
perature and strain information, we used trigger techniques to save only those
data points that were in the pulse itself together with the time of their collection.
Given a set rotating speed, the time increases linearly with the number of pulses
until it reaches the moment (pulse missing) at which the laser light is at the
Bragg wavelength of the sensor. The time between the pulse before and after the
Bragg wavelength is larger than the time between the other pulses, thus creating
a jump in the curve, as shown in Figure 6.17. By correlating t Bragg to the voltage
signals from the ECDL, we determined the Bragg wavelength.
94
0.03
0.025
Pulse signal, v
0.02
0.015
0.01
0.005
0
0 0.2 0.4 0.6 0.8
Scannng time, s
tbragg
Numbers of pulses
95
2
1.8
Wavelength shift, nm
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0 10 20 30 40 50 60
Time, s
We used an Altitun NYW-30 GCSR laser, which has three tuning sections: a
coupler section, a phase section, and a reflector section. The coupler section
provides coarse tuning, and the reflector and phase sections provide fine tuning.
The tuning range of the laser diode is from 1529 to 1561 nm. The line width of
the laser output is 25 MHz. This high-speed, non-contact sensing system is
96
sketched in Figure 6.19. The GCSR laser diode was used to launch light into the
embedded sensor, and the output signal, modulated by the temperature/strain,
was collected by a high speed (10 GHz) PIN photo-detector or an optical spec-
trum analyzer (OSA). Under a thermal load, the Fiber Bragg Grating reflected a
single wavelength; other light with different wavelengths went on through the
grating and into the photo-detector or OSA. Again, trigger techniques were used
to control the operation of the tunable laser diode. Since the output voltage from
the PIN photodetector was negative, the high-speed sensing signal in Figure
6.15 was actually reversed to negative values. Only after a down trigger signal
was detected was the laser diode controlled to tune a range of wavelengths at a
speed of 0.2 nm every 40 ns. During each tuning circle, data points were
recorded to determine the Bragg wavelength. The load was then calculated
according to the calibrated relationship between the Bragg wavelength shift and
the temperature change.
The experimental results are in Figure 6.20, which shows pictures scanned by
an OSA. The results clearly demonstrate the Bragg wavelength shifts after a
load was applied. A typical result in the optical power domain (from an oscillo-
scope connected to the PIN detector) is shown in Figure 6.21. The voltage was
negative and hence an inverted pulse signal was shown instead of the uninverted
pulse signal in Figure 6.15. All the results suggest that the high-speed non-con-
tact sensing system is feasible for monitoring the temperature and strain varia-
tions of rotating parts with fiber optical sensors embedded or attached to the
surface.
6.5. Summary
New techniques have been developed for temperature and strain measurements
of rotating components with Fiber Bragg Grating sensors embedded or attached
to the surface. Tunable laser diodes were incorporated into the sensing system
for monitoring the Bragg grating wavelength and thus the temperature or strain
was determined. Both low-speed and high-speed sensing methods were proposed
97
and tested. Experimental results verified the feasibility of the new methods.
Clearly, the non-contact sensing system provides a new sensing tool superior to
slip rings to optimize the design for turbine blades, vehicle wheels, and other
rotary tooling, etc. Moreover, with embedded sensor techniques, the new tech-
niques provide a new health monitoring methodology for rotating structures,
especially those exposed to hostile environments, such as blades in gas turbine
engines.
GCSR laser
EDFA
Fast tuning,
20nm/us
OSA High P(λ)
P(λ) speed
sampling
λ
λ scope
Fig.6.19. Remote sensing system for rotating parts with embedded fiber optical
sensors
Fig.6.20. Bragg wavelength shifts from an OSA for the rotating part
98
Reversed pulse
99
Chapter 7: Materials Issues for Layered
Manufacturing
This chapter discusses material issues in laser-assisted Layered Manufactur-
ing, and presents a new class of composites of INVAR and TiC designed to min-
imize thermal stresses in Layered Manufacturing. Experimental results on
mechanical and physical properties of the laser deposited composites of INVAR
and TiC are presented. Furthermore, the feasibility of laser-deposited stainless
steel 316L and tool steel H13, both mixed with varying amounts of TiC, is also
studied.
7.1. Introduction
For the fabrication of metallic parts, a laser was added to the SDM process in
the late 1990s and it is the laser we used in this study presented here [37]. The
laser is a 2.4 kW CW Nd:YAG laser, which is delivered to the focusing optics
through a 25 m long, 600 mm diameter fiber. The end of the fiber is equipped
with an end effector that can vary the focal length from 130 to 450 mm, with
corresponding changes in the focal spot diameter from 1.2 to 4.8 mm. In laser-
assisted SDM, the part material is deposited with an intense laser beam that is
focused onto the substrate, where it creates a melt pool. Powders are fed into the
melt pool through a simple gravity-driven, off-axis feed nozzle, and the powder
feed rates are controlled with an auger-style dry material feeder. Three powder
feeders are available for deposits of different materials. Both the laser end effec-
tor and the powder feed nozzle are mounted on the end of a four-degrees-of-
freedom robotic manipulator arm. The feed nozzle moves over the substrate
100
according to the deposition path generated by the process planner, depositing a
layer in near net shape. To obtain the final net shape, the layer is machined with
a five-axis mill or an electric-discharge machine (EDM). Figure 7.1 shows pic-
tures of the deposition apparatus mounted on the robot arm and three powder
feeders.
101
(b) Powder feeders
Since lighter powders tend to be deflected more by the inert gas and plasma, it
is difficult to control the composition of heterogeous structures accurately.
Moreover, as the schematic in Figure 7.2 shows, it is difficult to build a straight
wall via this direct powder injection technique: since the growth of material at
the wall occurs at a slant, more material is needed for the later subtractive shap-
ing process.
102
To solve the problem, we developed a Selective Powder Preplacing technique,
shown in Figure 7.3. The feed nozzle moves over the substrate according to the
deposition path generated by the process planner with the laser beam off, and a
powder layer with controlled composition is preplaced on the substrate in near
net shape.To keep air from being trapped inside the powders, a much lower flow
rate of nitrogen was applied. Then, the nitrogen flow rate was increased and a
laser was used to melt the preplaced powder layer. Extra powders were depos-
ited so that straight wall was maintained. Figure 7.4 shows artifacts--a turbine
blade and a step container--built with the selective powder preplacing technique.
In this new technique, laser energy absorption was higher since the laser was
melting the powder rather than the solid substrate directly. Moreover, the
remelting depth in the substrate was much less than it was in the direct powder
injection technique. The fact that there is less melting of the substrate may
prove to be advantageous in maintaining fine features created in the last layer
and in minimizing residual stresses in the structure. This technique may also
become complementary to Selective Laser Sintering (SLA) for building hetero-
geneous structures and functionally gradient materials (FGMs). FGMs are dis-
cussed later in this chapter
103
Fig.7.4. Artifacts built by Selective Powder Preplacing technique
104
•·A low coefficient of thermal expansion (CTE) over a wide
temperature range, since internal stresses that occur during
solidification and cool-down depend strongly on the CTE.
Low CTE materials, such as INVAR, are candidates for minimizing the build-
up of thermal stresses. INVAR is a 36% nickel-64% iron alloy with a very low
CTE: near zero below temperatures of 300 ºC. Above 300 ºC the yield strength
of INVAR decreases rapidly as shown in Figure 7.5. This means that during the
solidification and cool-down of deposited INVAR no elastic energy originating
from thermal stresses is stored in the material, because down to 300ºC, the
matrix is too soft to store a significant amount of elastic energy. Below this tem-
perature, the CTE is low enough to avoid the buildup of further residual stresses.
Early beam tests showed that deflection in an INVAR beam was one tenth of the
deflection in a SS 316L beam [37]. However, INVAR does not satisfy the
requirement of Rapid Tooling due to its low yield strength, low wear resistance,
and low corrosion resistance, especially at elevated temperatures. Its strength,
wear resistance, and corrosion resistance must be improved while its exceptional
CTE is maintained.
105
6000 265
Yi eld Strength, M Pa
5000
215
4000
3000 165
2000
115
1000
0 65
-100 100 300 500
Temperature, C
Laser processes have been used to enhance wear and corrosion resistance by
surface alloying with hard particles. TiC as filler material increases hardness
and wear resistance of the composite material [104-106]. TiC is most commonly
used because of its extremely high hardness (33% higher than that of WC at
room temperature), low specific gravity, and inertness. Smooth, rounded TiC
micrograins, as well as sharp-edged WC micrograins, are shown in Figure 7.6.
The TiC micrograins, exposed on the surface by suitable lapping and polishing
techniques, exhibit excellent wear and galling resistance with a very low coeffi-
cient of friction. Rounded, smooth and ultra-hard TiC grains can protect the
matrix from abrasion and adhesion while the matrix, in turn, securely supports
the carbides. The resistance of TiC to oxidation at elevated temperatures is well
established. The CTE of TiC, 7.4 mm/(m×K), is much lower than the CTEs of
conventional metals.
106
(a) TiC inclusion (b) WC inclusion
The TiC phase ensures high hardness, wear resistance, corrosion resistance,
and strength while INVAR provides exceptional CTE and toughness. Investigat-
ing the feasibility of INVAR and TiC metal matrix composites fabricated by lay-
ered deposition was therefore worthwhile.
The 2400W, CW Nd:YAG laser was used to fuse the mixing powders of TiC
and INVAR. The laser power was controlled automatically and the deposition
apparatus was moved across the surface of the substrate by a four-degrees-of-
freedom robotic manipulator. With Selective Powder Preplacing technique, the
composition of the deposit at any point on the surface could be controlled accu-
rately. All experimental samples were formed at the following controlled param-
eters: 20 mm/s velocity, 2000 W laser power, and 20 g/min powder feed rate. We
used a 2.5 mm spot size by controlling the laser beam focal point approximately
10mm beyond the substrate. During laser cladding, the area around the laser
spot was flooded with inert gas to shield the deposits from air and reduce oxida-
tion. Large flow rates of nitrogen (approx. 200 liters per minute) were used.
107
The powder size of INVAR (Heat 502625) from Starmet Corporation was
mesh-80 (-180 µ m). The chemical composition of INVAR powder is shown in
Table 7.1. The particle size of TiC from Atlantic Equipment Engineers was from
mesh-100 to +325 (45 ~150 µ m). The chemical composition of TiC powder is
shown in Table 7.2.
C Mn Si P S Cr Ni Co O Fe
0.028 0.40 0.24 0.09 0.003 0.12 35.30 0.12 0.005 bal
Table 7.1: Chemical elements in INVAR powder (weight %)
C Free C Fe O Ti
19.41 0.26 0.17 0.49 bal
Table 7.2: Chemical elements in TiC powder (weight %)
108
exceeded, solidification starts with the precipitation of carbides [104]. Figure
7.7b to Figure 7.7f indicate a significant increase in the amount of precipitated
carbides combined with a change in their morphology. In pure INVAR (Figure
7.7a) the grains are spherical and about 5mm in diameter. When 6.6 vol.% TiC is
added, the character of the microstructure changes dramatically, as shown in
Figure 7.7b. The resolidified TiC and INVAR leave a dendritic structure. When
the TiC content increases to 22 vol.%, the dendritic structure becomes even
more obvious. At 29 vol.% TiC, some spherical particles appear in addition to
the dendrites, as shown in Figure 7.7e. The non-dendritic grains become even
more evident at 52 vol.% TiC (Fig. 3.7f). These spherical or sharp-edged grains
are due to unmolten TiC particles, which remain in their original powder shape.
X-ray diffraction experiments showed only Fe-Ni and TiC structures in the
composites. This suggested that only a small amount, if any, of TiC dissolved
into the INVAR structure. The composites may have low CTEs since the matrix
INVAR may maintain its own microstructure, which is responsible for the
exceptionally low CTEs.
109
(a) 0% TiC (b) 6.6 vol. % TiC
110
7.3.4. Properties
The most interesting question was whether the CTE of the reinforced INVAR
remains low enough for there to be a significant reduction in thermal stresses
compared to conventional laser-assisted SDM materials like stainless or tool
steel. Hardness and tensile strength were also of interest. The hardness can give
some crude estimation of the mechanical strength of the material. To obtain
accurate mechanical strength values of the composites, tensile tests are neces-
sary.
For CTE tests, 6.35 mm x 6.35mm x 25.4mm coupons were deposited uni-
formly with TiC content that varied from 0 to about 20 vol.%, in an INVAR
matrix. The dilatometer method (ASTM E228) was used to measure the bulk
CTE of the coupons. The average CTEs of pure INVAR are about 1.88µ m/(m×K)
at the temperature range of 25-125 ºC and 3.6µ m/(m K) in the range of 25-220
ºC while that of pure TiC is 7.4 µ m/(m K). Average CTEs, as plotted in Figure
7.8, were measured in two temperature ranges: 25-125 ºC and 25-220 ºC. Figure
7.8 shows that the CTE of the composites stays well within the range given by
the CTE of INVAR and TiC. If there is some alloying between TiC and INVAR,
this does not seem to influence the low thermal expansion coefficient of the
matrix material. The achieved CTEs of all tested composites are well below the
CTEs of the materials used up to now (the CTE of stainless steel is about 17
mm/(m K)). The average CTEs for different TiC content are much lower than
those of tool steel, mild steel, stainless steel, and aluminum alloy.
The above CTE tests were performed at temperature below 220 ºC. We were
interested in obtaining the CTEs of the TiC+INVAR metal matrix composites at
temperatures up to 400 ºC. Coupons with 0%, 5.6%, and 24.8% TiC content were
tested. Figure 7.9 shows the curves of thermal expansion versus temperature.
The laser-deposited pure INVAR (0% TiC in Figure 7.9) yields a thermal expan-
111
sion in good agreement with the CTE data provided by Starmet Corporation. A
higher percentage of TiC content in INVAR results in a higher thermal expan-
sion at the same temperature.
24
Al 7075-T6 CTE,25~125 C
20
CTE, 1e-6 m/m k
CTE, 25~220 C
16 S.S 316L (cold finished)
12 ASTM-A572
H13
8
4
0
0 2 4 6 8 10 12 14 16 18 20
Vol.% of TiC in Invar
Fig.7.8. CTE of INVAR with different TiC content in comparison with Metals
6.0E-03
5.0E-03
Thermal Expansion
4.0E-03
3.0E-03
2.0E-03
5.6% TiC
1.0E-03 0% TiC
24.8% TiC
0.0E+00
0 100 200 300 400 500
Temperature, C
112
The average CTE is defined as the ratio of total expansion to total temperature
change from room temperature. Our results, as shown in Figure 7.10, match the
ones in Figure 7.8 in the range of 25 to 220 ºC. The coupon with 24.8% TiC
yields a much larger CTE than INVAR with 0% and 5.6% TiC at all temperatures
up to 400 ºC.
2.00E-05
Average CTE, (L-L0)/(T-T0)
1.50E-05
1.00E-05
5.6% TiC
5.00E-06
0% TiC
24.8% TiC
0.00E+00
0 100 200 300 400 500
Temperature, C
Further study in CTE (dL/dT) showed that the CTEs increased with tempera-
ture up to 300 ºC, beyond which they started to decrease, as presented in Figure
7.11. Further study needs to be done for a explanation of this phenomenon.
113
2.50E-05
2.00E-05
CTE (dL/dT)
1.50E-05
0.00E+00
0 50 100 150 200 250 300 350 400 450
Temperature, C
6.00E-03
5.00E-03
Thermal Expansion
4.00E-03
3.00E-03
2.00E-03
1.00E-03
0.00E+00
0 100 200 300 400 500
Temperature, C
114
7.3.6. Hardness and tensile tests
115
60
50
Hardness, RC
40
30
20
10
0
0 10 20 30 40 50 60
Vol. % of TiC in Invar
600
Yield Strength, MPa
550
H13 (annealed)
500
Al 7075-T6
450
400
350
300 ASTM-A572
250 S.S 316L (cold finished)
200
0 5 10 15 20
Vol.% of TiC in Invar
116
30
25
Elongation, %
20
15
10
5
0
0 5 10 15 20
Vol.% TiC in Invar
One of the most important issues limiting the quality of parts produced with
SDM is the accumulation of residual thermal stresses. These stresses, which
result from fusing and constraining material that has a temperature gradient
[36], can cause warpage, delamination and surface finish defects in the final
part. The application of the new materials in laser SDM is expected to reduce
thermal stresses. To test the feasibility of the technique, a series of rectangular
beams were fabricated to measure the deflections produced by laser depositions.
In each of the tests, a single 2mm-thick layer of material was deposited onto a
constrained steel substrate. The substrates were 150mm long, 25mm wide and
9.5mm thick low-carbon steel. The maximum deflections on the bottom surface
were measured. Composites of INVAR and TiC and stainless steel 316L were
used for this test. The deposition technique was a simple continuous scan of the
laser across the surface.
Figure 7.16 shows the measured deflections of the bottom surface of the steel
substrate for the different materials. If INVAR is used, the deflection is less by a
117
factor of 2 than if stainless steel is used. Using composites with 0~20 vol.% TiC
also results in less deflection.
0.8
SS316L
0.75
Deflection, mm 0.7
0.65
0.6
0.55
0.5
0.45
0.4
0 5 10 15 20
Vol.% TiC in Invar
118
gradually. While the distribution of the TiC content grows essentially linearly,
the hardness distribution is non-linear. However, the hardness variation is con-
tinuous. Thus, desired hardness can be obtained through accurate mixing of
those two powders. The results approximately match the hardness results (see
Figure 7.13) that are shown in the Rockwell C scale.
110
105
100
Hardness, RB
95
90
85
80
75
0 10 20 30 40 50
Distance, mm
119
7.4. TiC with S.S 316 and H13
INVAR with the addition of a TiC phase yields high hardness, wear resistance,
and strength while maintaining an exceptionally low CTE and reasonable tough-
ness. Other alloys with good properties for tools are SS316L and tool steel H13.
The question is whether they, too, can be improved by the addition of TiC,
which is expected to enhance the mechanical properties of the matrix material,
such as its hardness. Thus, we used laser deposition to mix powders of SS316L
and tool steel H13 with TiC to produce materials, and we tested their properties.
Figure 7.19 shows the hardness of laser-deposited SS316L with increasing TiC
addition. With about 40% TiC content, SS316L yields a hardness of 43 Rockwell
C. With the Selective Powder Preplacing technique, the desired hardness can be
well controlled by adjustment of the powder feed rates of SS316L and TiC.
50
40
30
HRC
20
10
0
0 10 20 30 40 50
Vol Percentage of TiC in SS316
Figure 7.20 presents the plot of hardness versus different TiC contents in H13.
Again, hardness increases with increasing vol.% of TiC. Laser-deposited H13
itself yields a hardness of 53 Rockwell C, and the hardness increases to 65
120
Rockwell C with a 38 vol.% of TiC. In the same figure, the hardness after
annealing is also presented. After 1 hour annealing at a temperature of 820 °C,
the hardness drops about 30 on the Rockwell C scale for H13. This annealing
process may improve the machinability of the composites without significantly
lowering their usefulness for tooling.
70
60
50
40
HRC
30
HRC, before
20 annealing
HRC, after
10 annealing
0
0 10 20 30 40
Vol Percentage of TiC in H13
7.5. Summary
This chapter discussed material issues in the laser-assisted SDM process.The
aim of the work presented here was to develop materials that can significantly
reduce residual stresses in laser-deposited parts while maintaining or even
improving all other properties (strength, toughness, corrosion and wear resis-
tance).
121
laser-assisted SDM. The experimental results show that the new materials yield
exceptionally low CTE, reasonable ductility, high hardness, and significantly
improved yield strength. The class of materials studied in this work can reduce
deformation caused by residual stresses and improve mechanical properties sig-
nificantly compared to other materials used in layered manufacturing.
Furthermore, TiC powders were mixed with SS316L and H13 for laser fabrica-
tion. The hardness of the materials thus obtained was significantly higher than
that of the original materials. With the Selective Powder Preplacing technique,
the desired hardness can be controlled by adjusting the powder feed rates of TiC
and SS316L/H13.
122
Chapter 8: Conclusions
Layered Manufacturing can be used to integrate fiber optic sensors during the
production of tooling and structural components. Sensors can be placed close to
points of interest prior to enclosure. Techniques for embedding fiber optic sen-
sors in metals, polymers, and ceramics have been developed. Embedding the
optical fibers into metals is especially challenging because a metal normally has
a high melting temperature. A sequence that involves cleaning optical fibers and
substrates, sputtering thin metallic film as a conductive layer on the fiber, elec-
troplating the fibers with a Ni layer as a protective layer, and then laser cladding
the treated fiber with a stainless steel layer was developed. The main manufac-
turing issues are found to be the quality of the bond between the nickel layer and
the substrate, the thermal stress induced by a large temperature gradient that
exists during layered deposition, and the fragile points where the optical fibers
enter and leave the metal part. The embedding quality was improved signifi-
cantly after we optimized the cleaning process and the parameters.
Fiber Bragg Grating (FBG) sensors were embedded in metallic, nickel and
stainless steel, polymers, and ceramic structures. The embedded FBG sensors
were characterized for temperature and strain measurements. The embedded
FBG sensors in nickel and stainless steel provided high sensitivity, good accu-
racy, and high temperature capacity for temperature measurements. An tempera-
ture sensitivity approximately 100% higher than that of bare FBGs was
demonstrated. A main cause for errors was identified as the temperature gradi-
ents in the metallic structures if the embedded sensors are under rapid thermal
cycles. For strain measurements, the sensors embedded in metal and polyure-
thane yielded high sensitivity, accuracy, and linearity. The sensitivity of the
123
embedded FBGs was in good agreement with that of bare FBGs. Moreover, a
decoupling technique for embedded FBG sensors was developed to separate
temperature and strain effects. However, sapphire fiber were not able to main-
tain their integrity in alumina, possibly due to the shrinkage and stresses during
the sintering process.
New techniques have been developed for temperature and strain measurements
of rotating components with FBG sensors embedded or attached to the surface.
Tunable laser diodes were incorporated into the sensing system for monitoring
the Bragg grating wavelength and thus the temperature or strain was determined.
Both low-speed and high-speed sensing methods were proposed, and experimen-
tal results verified the feasibility of the new methods. The non-contact sensing
system provides a new sensing tool that is superior to slip rings to optimize the
design for turbine blades, vehicle wheels, and other rotary tooling, etc. More-
over, with embedded sensor techniques, the new techniques provide a new
health monitoring methodology for rotating structures, especially those exposed
to hostile environments, such as blades in gas turbine engines.
124
ites, the Selective Powder Preplacing technique was also developed for the
laser-assisted SDM process. Furthermore, TiC powders were mixed with
SS316L and H13 for laser fabrication. The hardness of the materials thus
obtained was significantly higher than that of the original materials. With the
Selective Powder Preplacing technique, the desired hardness can be controlled
by adjusting the powder feed rates of TiC and SS316L/H13.
125
Bibliography
1) F. B. Prinz, chair, "Process Overviews" in “Rapid Prototyping in Europe
and Japan”, Vol. I, Analytical Chapters, International Technology Research
Institute, Loyola College in Maryland, March 1997
126
9) P.F. Jacobs, “Rapid Prototyping & Manufacturing: Fundamentals of Stere-
olithography”, 1st ed., Society of Manufacturing Engineers, Dearborn, MI, 1992
11) P.E. Reeves and R.C. Cobb, “The Finishing of Stereolithography Models
Using Resin Based Coatings”, Proceedings of the Solid Freeform Fabrication
Symposium, The University of Texas at Austin, Austin, TX, August 7-9, 1995,
pp. 96-106.
13) Stratasys, Inc., "Fast, Precise, Safe Prototypes with FDM", Proceedings of
the 1991 Solid Freeform Fabrication Symposium, The University of Texas at
Austin, Austin, Texas, August 1991, pp. 115-122
16) Safari, S.C. Danforth, V.R. Jamalabad, P.J. Whalen, R. Donaldson and J.
Pollinger, “Fused Deposition of Ceramics (FDC) for Structural Silicon Nitride
Components”, Proceedings of the Solid Freeform Fabrication Symposium, The
University of Texas at Austin, Austin, TX, August 12-14, 1996, pp. 335-343.
127
view”, Proceedings of the Solid Freeform Fabrication Symposium, The Univer-
sity of Texas at Austin, Austin, TX, August 12-14, 1996, pp. 385-392.
18) A. Bandyopadhyay, R.K. Panda, V.F. Janas, M.K. Agarwala, S.C. Danforth
and A. Safari, “Processing of Piezocomposites by Fused Deposition Technique”,
Journal of the American Ceramic Society, Vol. 80, No. 6, June 1997, pp. 1366-
1372
21) G. Zong, “Solid Freeform Fabrication Using Gas Phase Selective Area
Laser Deposition (Thick Films)”, Ph.D. dissertation, The University of Texas at
Austin, 1991
22) H.L. Marcus, “Positioning, Focusing, and Monitoring of Gas Phase Selec-
tive Beam Deposition”, US Patent #5,135,695 issued August 4, 1992
23) W. R. Thissell and H.L. Marcus, “Design of a Closed Loop Computer Con-
trolled System for Selective Area Laser Deposition. I. Laser Systems, Gasflow,
and Substrate Temperature Control”, Materials and Manufacturing Processes,
v.11, no.4, July 1996, pp. 673-701
24) W.R. Thissell and H.L. Marcus, “Design of a Closed Loop Computer Con-
trolled System for Selective Area Laser Deposition. II. Remote Plasma Source
and Pressure Control”, Materials and Manufacturing Processes, v.11, no.4, July
1996, pp. 703-725
25) K. F. Jakubenas, Y.L. Lee, M.S. Shaarawi, H.L. Marcus, and J.M. Sanchez,
“Selective Area Laser Deposition of Titanium Oxide”, Rapid Prototyping Jour-
nal, v.3, no.2, 1997, pp. 66-70
128
26) L. Sun, L.L. Shaw and H.L. Marcus, “Silicon Nitride Coatings Formed
Using the Selective Area Laser Deposition (SALD) Technique”, Materials
Research Society Symposium - Proceedings, v.555, 1999, pp.179-184
27) S. Harrison and H.L. Marcus, “Gas-phase Selective Area Laser Deposition
(SALD) Joining of SiC”, Materials and Design, v.20, no.2, 1999, pp.147-152
28) J.E. Crocker, S. Harrison, L. Sun, L.L. Shaw, and H.L. Marcus, “Using
SALDVI and SALD with Multi-material Structures”, JOM, v.50, no.12, Decem-
ber 1998, pp. 21-23
32) J.W. Kietzman, A.G. Cooper, L.E. Weiss, L. Schultz, J.L. Lombardi, and
F.B. Prinz, “Layered Manufacturing Material Issues for SDM of Polymers and
Ceramics”, Proceedings of the Solid Freeform Fabrication Symposium, The Uni-
versity of Texas at Austin, Austin, TX, August 11-13, 1997, pp. 133-140
33) J.W. Kietzman, “Rapid Prototyping Polymer Parts via Shape Deposition-
Manufacturing”, Ph.D. Thesis, Stanford University, Stanford, CA, 1999
129
Fabrication Symposium, The University of Texas at Austin, Austin, Texas,
August 1991, pp. 178-186
130
44) S.C. Danforth and A. Safari, “Solid Freeform Fabrication: Novel Manufac-
turing Opportunities for Electronic Ceramics”, IEEE International Symposium
on Applications of Ferroelectrics, v.1, 1996, pp.183-188
49) H.B. Denham, T.A. Anderson, E. Madenci, P.D. Calvert, “Embedded PVF2
Sensors for Smart Composites”, Proceedings of SPIE - The International Society
for Optical Engineering, v.3040, 1997, pp.138-147
51) L. Sun, K.J. Jakubenas, J.E. Crocker, S. Harrison, L.L. Shaw, and H.L.
Marcus, “In Situ Thermocouples in Macro-components Fabricated Using SALD
and SALDVI Techniques. I. Thermochemical Modeling”, Materials and Manu-
facturing Processes, v.13, no.6, November 1998, p.859-882
52) L. Sun, K.J. Jakubenas, J.E. Crocker, S. Harrison, L.L. Shaw, and H.L.
Marcus, “In Situ Thermocouples In Macro-components Fabricated Using SALD
131
and SALDVI Techniques. II. Evaluation of processing parameters”, Materials
and Manufacturing Processes; Nov 1998; v.13, no.6, p.883-907
53) L. Sun, K.J. Jakubenas, J.E. Crocker, S. Harrison, L.L. Shaw, and H.L.
Marcus, “In Situ Thermocouples in Macro-components Fabricated Using SALD
and SALDVI Techniques. III. Fabrication and Properties of the SiC/C Thermo-
couple Device”, Materials and Manufacturing Processes, v.13, no.6, November
1998, pp.909-919
54) L.Sun, and L.L. Shaw, “Solid Freeform Fabrication of In-situ SiC/C Ther-
mocouples in Macrocomponents”, Metallurgical and Materials Transactions A:
Physical Metallurgy and Materials Science, v.30, no.9, September 1999, p.2549-
2552
55) R.C. Luo, C.F. Lin, C.M. Chen, and Y.S. Chen, “Rapid Resin Mold with
Embedded Thin Film Pressure/Temperature Sensors”, IECON Proceedings
(Industrial Electronics Conference), v.3, 1999, pp.1301-1306
58) M. Fasching, F.B. Prinz, and L.E. Weiss, “'Smart' Coatings: a Technical
Note”, Journal of Thermal Spray Technology, v.4, no.2, June 1995, p.133
132
Case Study in Wearable Computer Design," Computer-Aided Design, vol. 28,
no. 5, 1996, pp. 393-404
65) J.G. Cham, B.L. Pruitt, M. Cutkosky, M. Binnard, L.E. Weiss, and G.
Neplotnik, "Layered Manufacturing with Embedded Components: Process Plan-
ning Considerations", Proceedings of DETC99: 1999 ASME Design Engineering
Technical Conference, Las Vegas, NV, September 12-15, 1999
66) S.A. Bailey, J.G. Cham, M. Cutkosky, and R. Full, "Biomimetic Robotic
Mechanisms via Shape Deposition Manufacturing", 9th International Sympsium
of Robotics Research, Snowbird, Utah, October 9-12, 1999, pp. 321-327
68) H. Du and B.E. Klamecki, "Force Sensors Embedded in Surfaces for Man-
ufacturing and Other Tribological Process Monitoring." Journal of Manufactur-
ing Science and Engineering, v. 121 no.4, November 1999, pp. 739-748
133
69) M.I. Friswell, D.J. Inman, “Sensor Validation for Smart Structures”, Jour-
nal of Intelligent Material Systems and Structures, v.10, no.12, December 2000,
pp.973-982
72) D.G. Krantz, J.H. Belk, “Remotely Queried Wireless Embedded Microsen-
sors in Composites”, Proceedings of SPIE - The International Society for Opti-
cal Engineering, v.3044, 1997, pp.219-226
75) C.M. Pereira, M.S. Mattice, and R. Testa, “Intelligent Sensing and Wire-
less Communications in Harsh Environments”, Proceedings of SPIE - The Inter-
national Society for Optical Engineering, v.3990, 2000, pp.194-203
76) E. Udd, “Fiber Optic Smart Structures”, Wiley (Interscience), New York,
1995
77) C.D. Butter and G.B. Hocker, “Fiber Optics Strain Gauge”, Applied
Optics, vol.17, no.18, September 1978, pp.2867-2869
134
78) R.O. Claus and E. Udd, “Fiber Optic Smart Structures and Skins IV”, Pro-
ceedings of SPIE - The International Society for Optical Engineering, v.1588,
1991
79) W.B. Spillman, “Fiber Optic Embedded Sensors”, IEEE LEOS Annual
Meeting - Proceedings, v.2, 1994, pp.230-231
80) C. Chang and J.S. James, “Metal-coated Optical Fiber Damage Sensors”,
Proceedings of SPIE - The International Society for Optical Engineering,
v.1918, 1993, pp.138-144
81) J.S. Sirkis, C. Chang, and B.T. Smith, "Low Velocity Impact of Optical
Fiber Embedded Laminated Graphite/Epoxy Panels", Journal of Composites
Materials, v. 28, no.14, 1994, pp. 1347-1370
82) C.M. Lawrence, “Embedded Fiber Optic Strain Sensors for Process Moni-
toring of Composites”, Ph.D thesis, Stanford University, 1997
85) X.D. Jin, J.S. Sirkis, J.K. Chung, V.S. Venkat, “Embedded In-line Fiber
etalon/Bragg Grating Hybrid Sensor to Measure Strain and Temperature in a
Composite Beam”, Journal of Intelligent Material Systems and Structures, v.9,
no.3, March 1998, pp.171-181
86) V.M. Murukeshan, P.Y. Chan, L.S. Ong, and L.K. Seah, “Cure Monitoring
of Smart Composites Using Fiber Bragg Grating Based Embedded Sensors”,
Sensors and Actuators, A: Physical, v.79, no.2, 2000, pp.153-161
135
87) V.M. Murukeshan, P.Y. Chan, and L.S. Ong, "Intracore Fiber Bragg Grat-
ings for Strain Measurement in Embedded Composite Structures", Applied
Optics, v. 40, no.1, Jan. 2001, pp. 145-149
88) K.O. Hill, Y. Fujii, D.C. Johnson, and B.S. Kawasaki, “Photosensitivity in
Optical Fiber Waveguides: Application to Reflection Filter Fabrication”,
Applied Physics Letters, vol.32, 1978, pp.647-649
89) B.S. Kawasaki, K.O. Hill, D.C. Johnson, and Y. Fuji, “ Narrow-band
Bragg Reflections in Optical Fibers”, Optics Letters, vol.3, 1978, pp.66-68
90) A. Othonos and K. Kalli, “Fiber Bragg Gratings: Fundamentals and Appli-
cations in Telecommunications and Sensing”, Artech House, Boston, 1999
91) A.D. Kersey, M.A. Davis, H.J. Patrick, M. LeBlanc, K.P. Koo, C. G.
Askins, M.A. Putnam, and E. J. Friebele, “Fiber grating sensors”, Journal of
Light Wave Technology, vol. 15, 1997, pp. 1442-1463
93) G. Lin, L. Wang, C.C Yang, M.C. Shih, and T.J. Chuang, "Thermal Perfor-
mance of Metal-clad Fiber Bragg Grating Sensors", IEEE Photonics Technology
Letter, vol.10, No.3, 1998, p. 406
94) R.B. Wagreich, W.A. Atia, H.Singh, and J.S. Sirkis, “Effects of Diametric
Load on Fiber Bragg Gratings Fabricated in Low Birefringent fiber”, Electronic
Letters, Vol.26, no. 6, 1996, pp. 1223-1224
96) R.J. Van Steenkiste, “Theory of Fiber Optic Strain and Temperature Sen-
sors”, Ph.D Dissertation, Stanford University, 1995
97) H.F. Taylor and C.E. Lee, “Sensing Applications of Fiber Fabry-Perot
Interferometers Embedded in Composites and in Metals”, Experiments in Smart
Materials and Structures, ASME, AMD-vol.181, 1993, p.47
136
98) C.E. Lee, J.J. Alcoz, W. Gibler, R.A. Atkins, and H.F. Taylor, “Method for
Embedding Optical Fibers and Optical Fiber Sensors in Metal Parts and Struc-
tures”, Fiber Optic Smart Structures and Skins IV, SPIE vol.1588, 1991, p.110
99) S.E. Baldini, D.J. Tubbs, and W.A. Stange, “Embedding Fiber Optic Sen-
sors in Titanium Matrix Composites”, Fiber Optic Smart Structures and Skins III
(SPIE vol.1370), 1990, p.162
101) A.K. Sinha, D.G. Bogard and M.E. Crawford, “Film-cooling Effective-
ness Downstream of a Single Row of Holes with Variable Density Ratio”, Jour-
nal of Turbomachinery, vol. 113, no.3, July 1991, pp. 442-449
102) G.W. Jumper, W.C. Elrod and R.B. Rivir, J. Turbomach., “Film Cooling
effectiveness in High-turbulence Flow”, Journal of Turbomachinery, v.113, no.3,
July 1991, pp.479-483
137