Reliability of Active Fiber Composites (AFC) by Mark Melnykowycz

DISS. ETH No. 17767

LONG TERM RELIABILTY OF ACTIVE FIBER COMPOSITES (AFC)

A dissertation submitted to ETH ZURICH for the degree of Doctor of Sciences

presented by Mark Myron Melnykowycz MSc. Materials Science Michigan State University (MSU) born March 22, 1978 citizen of The United States of America

accepted on the recommendation of Prof. Dr. P. Ermanni, examiner Dr. Michel Barbezt, co-examiner Prof. Dr. Edoardo Mazza, co-examiner 2008

Dedicated to Sophie Szajnar, . my grandmother, who lived to 103 years of age and never stopped accepting life as it cameâ&#x20AC;Ś

CONTENTS Abstract .................................................................................................................................... vii Zusammenfassung...................................................................................................................viii 1. Introduction .................................................................................................................... 1 1.1. Applications ............................................................................................................... 1 1.2. Lessons from Nature .................................................................................................. 2 1.3. Smart Materials .......................................................................................................... 4 1.3.1. SMA ....................................................................................................................... 4 1.3.2. EAP ........................................................................................................................ 5 1.3.3. Electro/magneto-rheological .................................................................................. 6 1.3.4. Piezoelectrics.......................................................................................................... 6 1.3.5. AFC State of the Art............................................................................................. 11 1.4. Passive Materials...................................................................................................... 16 1.4.1. Integration ............................................................................................................ 17 1.5. Objectives and Project Outline................................................................................. 19 2. AFC Reliability Characterization................................................................................. 22 2.1. Introduction .............................................................................................................. 22 2.2. Specimen Preparation............................................................................................... 24 2.2.1. Materials............................................................................................................... 24 2.2.2. Active laminate manufacture ............................................................................... 24 2.2.3. Surface Bonded Fragmentation Test (SBFT) Probes........................................... 28 2.2.4. Overview of Active Laminate Probes .................................................................. 28 2.3. Experimental ............................................................................................................ 28 2.3.1. AFC tensile test .................................................................................................... 28 2.3.2. Mechanical oscillation test ................................................................................... 29 2.3.3. Acoustic Emission Monitoring............................................................................. 31 2.3.4. Surface Bonded Fragmentation Test (SBFT)....................................................... 31 2.3.5. Fatigue Testing..................................................................................................... 31 2.4. Results/Discussion ................................................................................................... 34 2.4.1. AFC mechanical properties.................................................................................. 34 2.4.2. Influence of mechanical strain on AFC strain sensor performance ..................... 35 2.4.3. SBFT Observations and Damage Evaluation....................................................... 38 2.4.4. Rotated AFC Sensor Performance ....................................................................... 39 2.4.5. Acoustic Emission (AE) Monitoring.................................................................... 43 2.4.6. Integrated AFC Fatigue........................................................................................ 44 2.5. Conclusions .............................................................................................................. 47 3. AFC Damage Evaluation and Modeling ...................................................................... 48 3.1. Introduction .............................................................................................................. 48 3.1.1. AFC Damage Evolution ....................................................................................... 49 3.2. Microscopy Investigation......................................................................................... 50 3.2.1. AFC Microscopy Findings – 0º Loading ............................................................. 52 3.2.2. AFC Microscopy Findings – 45º Loading ........................................................... 55 3.2.3. AFC Microscopy Findings - 360º Coverage IDE ................................................ 56 3.3. Polarization............................................................................................................... 57 3.4. AFC Model Development ........................................................................................ 57 3.4.1. Analytical Formulation ........................................................................................ 58 3.4.2. Micromechanical Approach ................................................................................. 59 3.4.3. ANSYS Model ..................................................................................................... 64 3.4.4. ANSYS Model Results......................................................................................... 65 3.5. Electrode Design and AFC Integrity........................................................................ 66 i

3.6.

Conclusions .............................................................................................................. 69 Laminate Integrity Analysis ......................................................................................... 70 4.1. Materials................................................................................................................... 72 4.1.1. Dummy AFC (DAFC) manufacture..................................................................... 73 4.1.2. Integration of DAFC and AFC in GFRP laminates ............................................. 73 4.1.3. Integration of DAFC/AFC in CFRP laminates .................................................... 74 4.2. Laminate Manufacture ............................................................................................. 75 4.2.1. Tensile Testing ..................................................................................................... 75 4.3. Finite Element Model............................................................................................... 76 4.4. Laminate Tensile Results and Discussion................................................................ 77 4.5. Laminate Fracture Behavior..................................................................................... 82 4.5.1. GFRP Laminate Fracture Behavior...................................................................... 82 4.5.2. CFRP Laminate Fracture Behavior ...................................................................... 85 4.6. Laminate Integrity Characterization......................................................................... 87 4.7. Conclusions .............................................................................................................. 91 5. AFC Packaging Strategies............................................................................................ 92 5.1. Introduction .............................................................................................................. 92 5.2. Packaging Design Considerations............................................................................ 92 5.2.1. Actuator................................................................................................................ 92 5.2.2. Sensor ................................................................................................................... 93 5.2.3. Laminate Design .................................................................................................. 93 5.2.4. Interface................................................................................................................ 94 5.3. Current Work............................................................................................................ 94 5.4. Experimental ............................................................................................................ 95 5.4.1. Composite Laminates........................................................................................... 95 5.4.2. Modified interface AFC ....................................................................................... 96 5.4.3. Tensile Testing ..................................................................................................... 97 5.5. Results-Discussion ................................................................................................... 98 5.5.1. Influence of laminate configuration on AFC sensor performance....................... 98 5.5.2. Influence of AFC interface modification on sensor performance...................... 103 5.6. Conclusions ............................................................................................................ 107 6. Concluding Remarks .................................................................................................. 109 6.1. AFC Outlook .......................................................................................................... 110 Acknowledgements ................................................................................................................ 111 References .............................................................................................................................. 113 Appendix ................................................................................................................................ 117 Curriculum Vitae.................................................................................................................... 119 Publications ............................................................................................................................ 120 Oral Presentations .................................................................................................................. 120 4.

LIST OF FIGURES Figure 1.1 Electroactive polymer. .............................................................................................. 5 Figure 1.2 Dipole displacement above and below the Currie temperature. ............................... 6 Figure 1.3 Dipole arrangement during polarization process. ..................................................... 7 Figure 1.4 Traditional wafer (a) and stack (b) actuator configurations. .................................... 8 Figure 1.5 IDE pattern (a) with close-up view (b) of cross section. .......................................... 9 Figure 1.6 Piezoelectric composite configurations, particulate (a), rod (b), and longitudinal (c). ............................................................................................................................................ 10 Figure 1.7 AFC components (a) and close-up of fiber section (b) [26]. .................................. 11 Figure 1.8 Schematic of AFC with different regions of polarization....................................... 11 Figure 1.9 Crack propagation behavior of PZT in response to polarization direction [35]. .... 15 Figure 1.10 Classic integration approaches insertion (a) and cutout (b).................................. 17 Figure 1.11 Project map of the dissertation.............................................................................. 20 Figure 2.1 Outline of AFC Characterization strategy. ............................................................. 24 Figure 2.2 Different AFC types including long (a) and short (b). ........................................... 25 Figure 2.3 Figure Vacuum bagging setup. ............................................................................... 26 Figure 2.4 Cross-section of a laminate showing the placement positions of devices integrated via cutout insertion. .................................................................................................................. 27 Figure 2.5 Schematic of a laminate with AFC arranged and wires with Teflon/Mylar protection.................................................................................................................................. 27 Figure 2.6 Schematic side view of Teflon/Mylar wire protection. .......................................... 27 Figure 2.7 Experimental set-up for the monotonic cycle test of 30 x 5 cm tensile specimens.30 Figure 2.8 Applied strain program for monotonic cycle test. .................................................. 30 Figure 2.9 Schematic of strain ranges for fatigue experiments................................................ 33 Figure 2.10 Description of strain parameters for fatigue tests. ................................................ 33 Figure 2.11 Typical electrical-mechanical response of a AFC tensile specimen..................... 34 Figure 2.12 Results from the first oscillation test set, used to establish the sensitivity limits. 35 Figure 2.13 Amplitude response of the second oscillation test set, showing the transition in performance behavior via loading to intermediate strain levels. ............................................. 36 Figure 2.14 Absolute signal response of short AFC during the cycle test. .............................. 38 Figure 2.15 Description of mechanical loading of the AFC in relation to polarization direction when rotated 90°....................................................................................................................... 40 Figure 2.16 Classic scenarios of ferroelastic depolarization in PZT materials........................ 40 Figure 2.17 Forces on the 45° AFC in relation to laminate loading. ....................................... 40 Figure 2.18 Strain sensor performance of 45° specimens........................................................ 41 Figure 2.19 AFC charge output of 0° and 45° specimens, strain applied along fiber direction. .................................................................................................................................................. 42 Figure 2.20 Plot of AE event signals located along the AFC specimen during monotonic cyclic testing............................................................................................................................. 43 Figure 2.21 AFC fatigue behavior showing initial signal degradation. ................................... 45 Figure 2.22 AFC fatigue behavior below the tensile limit of the AFC. ................................... 46 Figure 2.23 AFC fatigue behavior near the tensile limit.......................................................... 46 Figure 3.1 Damage model of integrated AFC in relation to applied strain.............................. 50 Figure 3.2 Microscopy setup of sectioned AFC....................................................................... 51 Figure 3.3 Schematic of an AFC with regions of polarization and non-polarization, as related to the IDE pattern, and loading direction. ................................................................................ 52 Figure 3.4 Section of AFC taken with polarized and non-polarized regions........................... 53 Figure 3.5 Close-up section of AFC taken from the middle polarized region with cracking locations identified in relation to the IDE finger position........................................................ 53 iii

Figure 3.6 Micrograph of polarized AFC specimen. ............................................................... 55 Figure 3.7 Micrograph of 45º SBFT AFC................................................................................ 55 Figure 3.8 Micrograph of fracture in poly-IDE AFC............................................................... 56 Figure 3.9 Schematic view of a PZT fiber showing the non-uniformity of polarization......... 57 Figure 3.10 Figure Uniform Field Model (UFM) [26]............................................................. 58 Figure 3.11 Side oriented cross-sectional view of a fiber in the AFC. .................................... 58 Figure 3.12 Conventional material coordinate system definition. ........................................... 62 Figure 3.13 Rotated material coordinate system (RCS) definition. ......................................... 62 Figure 3.14 Analysis process for polarization of fiber material [62]. ...................................... 63 Figure 3.15 Comparison of different material property descriptions....................................... 64 Figure 3.16 Material property gradient resulting from polarization. ....................................... 65 Figure 3.17 First Principle strains in AFC polarized fiber model............................................ 66 Figure 3.18 Etched electrode idealization. ............................................................................... 67 Figure 3.19 Integrated through-thickness electrode................................................................. 67 Figure 3.20 Material property distribution of etched IDE. ...................................................... 68 Figure 3.21 Critical stress distribution of etched IDE.............................................................. 68 Figure 4.1 DAFC showing Kapton surface and GFRP core. ................................................... 73 Figure 4.2 Position of the AFC/DAFC for the insertion (a) and cutout insertion (b) integration procedure in the GFRP laminates............................................................................................. 74 Figure 4.3 Integration of elements in cross-ply CFRP laminates. ........................................... 75 Figure 4.4 Schematic view of the laminate test specimens...................................................... 75 Figure 4.5 Ultimate tensile strength results of GFRP laminates. ............................................. 77 Figure 4.6 Young’s modulus of the GFRP laminates. ............................................................. 78 Figure 4.7 Ultimate tensile strength of the CFRP laminates.................................................... 79 Figure 4.8 Figure 3.4 Young’s modulus of the CFRP laminates. ............................................ 79 Figure 4.9 Load response of insertion integration GFRP laminates. ....................................... 80 Figure 4.10 Load response of cutout integration GFRP laminates. ......................................... 81 Figure 4.11 Loading curves of CFRP inserted laminates......................................................... 82 Figure 4.12 Loading curves of CFRP interlaced laminates. .................................................... 82 Figure 4.13 Far off-center symmetric cutout insertion specimens with DAFC. ...................... 83 Figure 4.14 Center placement symmetric cutout insertion specimens with AFC. ................... 83 Figure 4.15 Center (a) and off-center (b) placement DAFC post failure................................. 84 Figure 4.16 AFC specimen integrated in GFRP post failure. .................................................. 85 Figure 4.17 CFRP Far Off Center AFC laminates with inserted (a) and interlaced (b) integration................................................................................................................................. 86 Figure 4.18 Fracture behavior of AFC-CFRP laminates using the insertion (a) and interlacing (b) integration techniques......................................................................................................... 87 Figure 4.19 TWSI plot of integration region in a woven GFRP laminate with center inserted integration................................................................................................................................. 88 Figure 4.20 TWSI plot of integration region in a woven GFRP laminate with far off center inserted integration................................................................................................................... 88 Figure 4.21 TWSI plot of the top ply of a CFRP interlaced far off center composite. ............ 89 Figure 4.22 FPF plot of ANSYS results and actuation of GFRP laminates............................. 90 Figure 4.23 FPF plot of ANSYS results and actuation of CFRP laminates............................. 91 Figure 5.1 AFC encapsulated in silicon rubber........................................................................ 96 Figure 5.2 AFC specimen laminated with CFRP material....................................................... 96 Figure 5.3 Pre-tensioning concept and applied compressive force on the AFC. ..................... 97 Figure 5.4 CFRP pre-tensioning curing apparatus. .................................................................. 97 Figure 5.5 Stain sensor performance of AFC integrated in CFRP........................................... 99 Figure 5.6 Performance curves for AFC integrated into GFRP and CFRP. ............................ 99

Figure 5.7 Schematic view of force distribution per ply in a laminate. ................................. 100 Figure 5.8 Comparison of forces transferred through a laminate ply for woven GFRP as compared with cross-ply CRFP for comparable strains......................................................... 102 Figure 5.9 Sensor performance curve of AFC with CFRP interface. .................................... 103 Figure 5.10 AFC CFRP interface specimen which failed at 0.80%....................................... 105 Figure 5.11 AFC strain cycle performance curve with pre-tensioned CFRP interface and no interface.................................................................................................................................. 105 Figure 5.12 Comparison between normal AFC and prestressed CFRP interface. ................. 107

LIST OF TABLES Table 2.1 Dimensions of AFC for tensile and integration analysis. ........................................ 25 Table 2.2 Specimens for quasi-static strain sensor performance evaluation............................ 28 Table 2.3 Specimens for fatigue strain sensor performance evaluation................................... 28 Table 2.4 AFC fatigue test specimens including strain level and fatigue cycles. .................... 33 Table 2.5 Performance recovery of integrated AFC at 0.05% strain after loading to each strain level. ......................................................................................................................................... 37 Table 3.1 Description of the AFC Damage Evaluation and Modeling investigation. ............. 49 Table 3.2 Description of AFC in microscopy investigation..................................................... 51 Table 4.1 Comparison of AFC and DAFC mechanical properties. ......................................... 73 Table 4.2 Summary of integration configurations tested. All laminates had specimen sets with integrated DAFC, those that also were tested with AFC are indicated............................ 76 Table 5.1 Description of laminate types used for tensile testing. ............................................ 95 Table 5.2 Description of probes and modified interface AFC test specimens. ........................ 95 Table 5.3 Material properties for GFRP and CFRP laminates............................................... 101

Abstract Active Fiber Composite (AFC) materials are composed of piezoelectric Lead Zirconate Titanate (PZT) fibers encased in an epoxy matrix. The combined mechanical and electrical properties of PZT make AFC ideal as sensor and actuation devices, which can be integrated into composite laminates. This enables the manufacture of structural composite laminates with active properties. Although already studied as an actuation system, the sensor function and reliability of AFC has not been investigated. Furthermore, the fiber failure and fragmentation behavior has not been fully studied, and methods for improving the structural properties of AFC have not been proposed. The current work uses mechanical characterization techniques to characterize AFC strain sensor behavior including fragmentation in the PZT fibers, the influence of electrode placement on fiber failure and how AFC integration affects the integrity of glass and carbon fiber laminates is also presented. While AFC integration affected composite laminate integrity, this impact could be mitigated by integrating AFC into cross-ply as opposed to woven laminates. Active laminates with AFC were shown to exhibit excellent fatigue reliability so long as the integrity of the PZT fibers was not compromised. Fiber cracking occurs near electrode edges due to the modification of fiber material properties resulting from electrical polarization of the fibers. The use of a pre-stressed CFRP interface was used to effectively place PZT fibers in compression and thereby extend the strain region of the AFC.

vii

Zusammenfassung Active Fiber Composites (AFC) bestehen aus den Fasern des Lead Zirconate Titanate (PZT) welche in eine Epoxidmatrix eingebettet werden. Die Kombination von mechanischen und elektrischen Eigenschaften von PZT eignen sich ideal für Sensor und Betätigung Vorrichtungen und können in Kunststoffe integriert werden. Dieses ermöglicht die Herstellung der AFC mit aktiven Eigenschaften. Obwohl die Eigenschaften des AFC als Aktor-System zwar studiert wurden, fehlen Untersuchungen zur Sensor-Funktion und Zuverlässigkeit von AFCs. Außerdem wurde der Faserausfall und das Zerteilungsverhalten nicht vollständig erfasst und Methoden für das Verbessern der strukturellen Eigenschaften von AFC wurden nicht vorgeschlagen. Die vorliegende Arbeit verwendet mechanische Charakterisierungs-Techniken um AFCs auf Belastung, Sensor-Verhalten einschließlich der Zerteilung in den PZT Fasern zu charakterisieren. Außerdem wird der Einfluss der Elektroden Platzierung auf Faserausfall und die Beeinflussung von AFC auf Glas und Carbon Faser Kunststoffe untersucht. Obwohl die Einbindung von AFC in Kunststoff- Verbundstoffe die Intaktheit beeinflusst hat, könnte eine AFC Einbindung in die cross-ply anstelle der woven laminate mildern. Aktive, AFC versehen Kunststoffe haben eine ausgezeichnete ErschöpfungsBeständigkeit aufgewiesen, solange die Intaktheit der PZT Fasern nicht gefährdet waren. Faserbrüche nahe den Enden der Elektroden gehen auf die Modifikation der Faser Materialien aufgrund der elektrischen Faserpolarisation zurück. Der Einsatz eines vorgespannten CFRP Anschlusses machte es möglich, die PZT Fasern effektiv komprimiert zu platzieren und somit die Spannungs-Region des AFC auszuweiten.

viii

1. Introduction Engineering fields have traditionally focused on using uni-functional materials with set mechanical properties to design and build structures and products which fulfill certain design intents. Materials such as wood, concrete, and metals have served vast uses throughout history; enabling the advancement of industries and societies. The industrial introduction of metals such as titanium or aluminum alloys and fiber-based composite materials have allowed the production of lightweight structures such as airplane fuselages and numerous inventions. While great strides have been made in the field of materials, modern science and engineering does not tell how to design and manufacture multifunctional materials, which can be designed for multiple design intents such as actuation and sensing in addition to fulfilling load-bearing requirements. Traditional materials such as wood, concrete and metals are generally highly engineered and optimized to certain design requirements. Systems in Nature however, use complex materials, which are genetically optimized through natural selection to accommodate to the changing conditions of a certain operating environment. In many ways current material design concepts are being rethought with an eye towards systems, which integrate active components into traditional materials to create material systems which interact with their environments. One method of accomplishing this goal is the integration of sensors and actuators into laminate materials. This allows the engineering of multi-functional products, which can sense and respond to changes in their environment and thereby increase the functionality of the original design. The field of smart materials seeks to fill the gap between traditional uni-functional and controllable multifunction materials. The goal of smart materials research is to enable the design of materials that can function beyond the traditional design interests such as strength, stiffness, thermal conductivity, etc. An active or smart material system can generally be thought of consisting of three essential elements, the active material, the passive material, and the control system. The active material is defined as the actuator or sensor element. As a sensor, the device exists to receive information from the operating environment (such as vibration or mechanical deformation). As an actuator the device enables interaction with the environment via shape or vibration control. The term â&#x20AC;&#x153;passive materialâ&#x20AC;? or "host structure" denotes the material that the device is integrated into. Finally, the control system refers to the mechanism or program which collects information from the sensor elements and controls actuation of the material. Modeling of various control systems has been addressed in numerous works, but research into the coupling between device and structure is infrequently addressed. Ideally, the same device should be able to act as a sensor and actuator, thereby lending greater flexibility to fulfilling the specific design requirements of any given application.

1.1.

Applications

The ultimate goal of smart materials research is the development of new materials systems which can be used to benefit society via the design of more intelligent products, systems, and inventions. Various applications exist for smart materials, both in theory and in industry. A number of smart material based products have been developed including tennis rackets [1], adaptive skis [2], and rotor blades [3]. Some of the most promising applications for smart materials includes, active shape control of aeroelastic structures, active damping of structures, and structural health monitoring (SHM) of aeronautic and civil structures [4, 5]. The control of aeroelastic structures encompasses a large part of smart materials research activities. The Active Wing Project [6], administered by DARPA in the United States, has shown that the integration of active control surfaces can have a beneficial effect on the efficiency of aeroelastic structures. At the heart of any given smart materials system lays the sensor or actuator material component. First, one must consider how a smart material should be designed. The development of smart materials should not imply that traditional materials such as wood, ceramics, and metals are archaic or old-fashioned, rather that it is not yet possible to design material systems to fulfill a specific application requirement whereby the material can act as a designable component in a sensing and actuation capacity. Such materials do exist in the Natural world. There are numerous examples of biological materials, which are optimized to work with their environments, whereas within the constructs of modern engineering, materials and their applications do not generally follow coincident development pathways. A car frame can be designed to certain strength requirements, but can not be designed to actively damp out road vibrations, while airplane wings can be tailored to certain aeroelastic loading conditions, they can not currently be designed for automatic flutter control or shape adaptive requirements in response to fluctuating flow conditions. These applications both use materials such as aluminum, metal alloys, and polymers, but these materials were designed separately from the application. Birds employ wing designs which actively change in response to fluctuating aerodynamic loading conditions, while modern aircraft use an essentially static wing design philosophy. In fact, as noted by [7] the general design of aircraft has changes little in the last 50 years, and, to a certain extent, has stagnated, whereas the wing of birds change frequently from each region of the world. The use of materials in modern society differs greatly to the way materials are developed and applied in Nature.

1.2.

Lessons from Nature

In Nature, materials are in a sense, created and developed along with their intended application. For example, trees and plants are optimized to succeed in a given environment through genetic evolution over time and as a reaction to environmental factors such as the availability of nutrients and the harshness of cold and wind in which they must adapt to survive in. This enables the design of materials which are extremely reliable for their given purposes. However, engineering problems such as car bodies and airplane wings are generally designed according to certain static design

constraints, and can not respond to variations in their operating environment by modifying their properties to optimize their material response. One question arises, if looking to the natural world for hints on smart materials design, how should examples in Nature be interpreted? What materials design philosophies should be taken from the Natural world and how should they be applied to current topics in science and engineering? The Stingray is a classic example from the marine environment, used to show how fully active skins might be used to design nautical and aeronautical crafts with fully deformable and controllable active surfaces instead of jet or propeller engines. The Stingray has a fully deformable wing structure that allows it to move with flow currents, whereas boats and submarines employ static hulls, which react against changing flow conditions (as opposed to accommodating them). The example of the Stingray might be used to make a case for the development of fully active materials, which enable structural actuation as well as offering load-bearing support. However, there are few if any materials which can be used to create actuation and are robust enough to be used as the main load bearing component in a smart material system (unless very low forces are required). Many material-structural systems in Nature retain active combined with passive-structural and elastic components, as opposed to one complete active-structural material. A classic example in biomemetics is that of the Venus Fly Trap. The Venus Fly Trap is often used as a reference to a biological system with a very high strain ability and functionality, and is used as an example of a biological system that might be mimicked to develop fully active smart materials. However, here it is not a question of one material creating actuation and at the same time supporting the structure of the plant. The Venus Fly Trap structure exploits the three aforementioned components of a smart material system: a passive material combined with an active material and a control system. These components are optimized to fulfill a certain function, in a reliable, reproducible way over an extended lifetime, where a part of the structure is active, and a part offers passive structural support. The Venus Fly Trap includes two states, open and closed. When the plant is open, and a fly lands, hairs detect movement, and a signal is sent to close the trap. Tension along the active surface builds in the lip until a certain threshold is reached, and then snaps shut. The large deflections are achieved via a snap-through effect as initiated by the active surface, not due to the action of one active-structural material [8]. The snap-through design philosophy has been employed in a number of studies, where an actuation controls an unstable or bi-stable laminate to create large actuation [9-11]. The skeletal-muscular system in animals and humans is another example of an active material system with separate passive and active components coupled with a control system, which work together to fulfill a certain function. Muscles are also used as examples of material concepts which could be exploited to design active structures such as morphing wings. In this example, bones give structural shape and support to the weight of the body, while muscles mainly serve only to actuate when needed. For example, when the body stands erect, very little muscular force is required to support

the body since the skeletal structure has been optimized and aligned along the center of gravity of the body. Although some muscles are continually active such as the heart and diaphragm, the majority of muscles are in a resting state until activated for short periods of motion to fulfill the certain specific functions for which they were "designed" such as the biceps contracting to raise the forearm in one direction, while the deltoids and shoulder muscle system enables rotation of the arm via independent motion. Movement in the muscular-skeletal system is achieved via the use of optimized actuators (muscles) working in conjunction with passive support structures (bones), and controlled via the central nervous system. When investigating the topic of smart materials, it is apparent from the genetically optimized examples in Nature, that systems with separate passive and active materials can be intelligently used together to design reliable active structures. The question then becomes, of the numerous active and passive materials currently available, which combination should be investigated to advance the general body of knowledge of smart materials in the scientific community?

1.3.

Smart Materials

Many different materials are in one way or another referred to as smart or intelligent materials, but are generally categorized according to their function and base material type. Therefore different smart materials generally fall into one of the following categories, shape memory alloy (SMA), electro-active polymer (EAP), magneto and electrical rheological, or piezoelectric.

1.3.1. SMA The SMA category is made up of metallic alloys with shape memory effects, which rely on a temperature driven phase transformation to realize shape change in the material. Essentially, a material can be deformed, and returned to its original shape by heating. Materials that heal themselves, actuators, numerous medical devices, and many other applications exist for SMAs. The most common SMA is Nitinol (Ti3Ni4), and research into it dates back to research by the United States Navy, hence it's name, Nickel Titanium Navy Ordinance Labs (Nitinol). The morphing affect of SMAs relies on the temperature dependent phase transformation between austenite and martensite [12]. The shape memory effect is possible through reversibility of the martensite transformation via self-accommodation of martensitic plates. Shape memory alloys rely on a transformation sequence between martensite and austenite. A shape memory material may be mechanically deformed while exhibiting a martensitic phase, and return to its original shape when heated above its transformation temperature. When heated, the material goes through a reverse transformation, the phase changing from martensite to austenite. Upon cooling from the austenite phase, the material crystallography returns to martensite. The

transformation is rather versatile, enabling the manufacture of three different types of shape memory effects: one-way, two-way, and mechanical shape memory. The narrow composition tolerance of Nitinol, and cost of the raw materials makes it difficult to manufacture in large bulk. Advances are being made in the area of iron and copper based SMAs, which offer a lower material cost. Super elastic tendons and texture-changing surfaces are common applications of Nitinol-based smart material system designs. However, the phase transformation dependence on temperature renders a poor actuation response time unless the material can be heated and cooled at high rates. This implies that the surface area to volume ratio of a Nitinol actuator should be very high, such as occurs with a Nitinol coating as opposed to a Nitinol beam actuator. Despite a great deal of research expenditure, Nitinol based smart material designs have seen limited success. One drawback of Nitinol, as with all metals, is that the accumulation of dislocations in the crystal lattice can lead to plastic hardening of the material, which can limit the service life and long-term reliability of Nitinol actuators. Over millions of fatigue cycles degradation in shape recovery can be seen.

1.3.2. EAP Electroactive polymers (EAP) represent a relatively new (as compared with Nitinol) class of smart materials. EAPs can exhibit very high deformations coupled with low forces and are classified in two forms, dielectric and ionic based. In the dielectric form polymers are layered with electrodes and a voltage is applied, the electrostatic forces then lead to deformation of the material. Dielectric EAPs therefore rely on the electrostatic forces between electrodes to induce actuation by expansion of the polymer layer as shown in Figure 1.1. When a high voltage is applied electrostatic attraction between the electrodes leads to an expansion in the plane of the actuator. Ionic EAPs function via the displacement of ions in the polymer [13]. The required driving voltages are generally on the order of a few volts, less than that required for dielectric EAPs. However, a larger driving current is required.

Figure 1.1 Electroactive polymer. Projected applications for EAPs include artificial muscles, body sensors, and peristaltic pump designs. Due to the viscoelastic nature of the base polymer material, EAP actuators generally exhibit low response times. At the present time, the commercialization of EAP materials is not sufficiently advanced to provide a 5

characterized material source to be considered for a smart materials characterization study. The long-term reliability of EAP actuators is also a concern, dielectric actuators need to be pre-stressed but degradation of the pre-stress stiffness can occur. The advent of new polymers will no doubt lead to improvement in EAP design and reliability.

1.3.3. Electro/magneto-rheological Electro-rheological materials include fluids or gels which change their viscosity in response to the application of an electric field. Essentially a non-conducting fluid is combined with fine conductive particles, which are in a state of suspension. The conductive particles react to the application of an electric field, and the viscosity of the fluid can then be modified by the applied electric field strength. If magnetic particles are used, a magnetic field is used to control the material and they are termed magnetorheological.

1.3.4. Piezoelectrics In piezoelectric materials deformation of the material leads to the development of a potential at the material surface. Conversely, the application of an electrical potential leads to deformation of the material. This phenomenon was discovered by Jacques and Pierre Curie in 1880. A number of different materials exhibit piezoelectric properties (including bone). Some of the most common and well researched for smart material applications are the Lead Zirconate Titanate (PZT)-based piezoelectrics. The mechanics of the piezoelectric phenomenon are tied to an electrical dipole charge imbalance at the crystallographic level. 1.3.4.1. Polarization

Figure 1.2 Dipole displacement above and below the Currie temperature. At the crystallographic level, upon cooling below the Currie temperature, the crystal lattice is deformed and the central atom (Ti or Zr) is displaced from its equilibrium position as shown in Figure 1.2. This position change results in a charge imbalance,

which is the heart of use of the material as a sensor or actuator. Given the isotropic nature of polycrystalline PZT, there will initially be no global preference for the direction of the charge imbalance throughout the material. On the macroscopic level, an unpolarized material will have electrical domains with random orientations throughout the material. Polarization entails the alignment of the individual electrical domains along a preferential direction. During polarization a voltage is applied, which leads to an alignment of the domains. After the polarization voltage is removed, the electrical domains retain a preferred global orientation along the direction of the polarization field. This polarization process is illustrated in Figure 1.3. Application of a voltage in the direction of polarization will lead to a deformation of the material, and allows used of PZT as an actuator. The application of a stress along the polarization direction leads to the development of a voltage, which allows use of the PZT as a sensor. It is apparent then, that the application of the polarizing electric filed is of great importance in designing a sensor or actuator, and can be defined by the electrode pattern on the material surface.

Figure 1.3 Dipole arrangement during polarization process. 1.3.4.2. Depolarization Depolarization is the process by which the electrical dipole moments lose their preferential orientation along a prescribed direction. Often, the reliability of PZT actuators and sensors is tied to the ability of the material to retain polarization along the desired direction. Various factors can lead to depolarization of the piezoelectric effect and therefore, performance degradation of PZT-based devices. This may occur due to mechanical (ferroelastic), electrical (ferroelectric), or temperature loading. If a voltage is applied in a direction other than the polarization direction, the material will expand or contract according to the polarity and direction of the applied field and the domains will remain oriented along the original direction of polarization. If a strong electrical field is applied however, the domains will reorient to the new 7

polarization direction. An analogous behavior is seen for mechanical loading. If, upon unloading, the domains reorient to the original polarization direction, then the depolarization is termed reversible. If however, the domains remain oriented in the new direction, irreversible depolarization has occurred, and the material must be repolarized (along the original polarization direction) to reach the original state of polarization. Temperature also has a large effect on polarization. If polarized PZT is heated above the Currie temperature, then the preferential polarization will be lost and a homogeneous polarization state will exist. A PZT actuator or sensor will show a decrease in performance if an increasing temperature load is applied. However, so long as the maximum temperature is below the Currie temperature, upon cooling, performance of the actuator or sensor will largely attain the original value [14]. 1.3.4.3. PZT Material Properties PZT material properties are characterized according to mechanical as well as electrical-mechanical properties. The deformation per volt is described by the dxy coefficient, where x represents the direction of the driving voltage, and y represents the direction of deformation. PZT can be produced as a soft or hard piezoelectric. Many studies have addressed the ferroelastic and ferroelectric properties of bulk PZT [15-21] which lends insight into the physical principles governing PZT sensor and actuator devices. Soft PZT retains a higher d33 coefficient, but can depolarize easier than hard PZT. PZT can be processed into different shapes and forms, allowing the development of tailored sensors and actuators. In addition, PZT can be produced as a powder and mixed with a binder can be extruded to produce fibers with various diameters and cross-sectional shapes. PZT-based devices generally take the form of a wafer, stacked actuator, or composite. 1.3.4.4. PZT Actuators and Sensors

a b

Figure 1.4 Traditional wafer (a) and stack (b) actuator configurations. The most basic PZT device configuration is the planar wafer design (Figure 1.4a). In this configuration a PZT wafer is sandwiched between opposing polarity electrodes and polarized through the wafer thickness in the d33 direction. A driving voltage is applied, which produces expansion parallel to the wafer plane and is defined by the d31

coefficient. A variant of the basic wafer design leads to the Rainbow [22] and Thunder [23] actuators. The Rainbow/Thunder design employs a thin steel plate, which gives a slight curvature to the assembled device. In recent years the LIPCA [24] actuator has been developed to improve upon the Rainbow/Thunder configuration by using composite layers. In the LIPCA design the wafer is sandwiched between a layer of carbon fiber composite on the top, and glass fiber composite on the bottom. The PZT and fiber composite layers are processed together, and upon cooling the difference in thermal expansion properties between the carbon and glass composites leads to a slight out of the plane curvature. The wafer-based designs are the easiest to manufacture, but suffer from low damage tolerance and actuation performance. Furthermore, the performance of the basic wafer electrode design is dominated by the d31 coefficient, whereas it would be ideal to utilize the d33 as the primary coefficient (which is twice as large in magnitude). By stacking PZT wafers on top of one another, the stack configuration is realized (Figure 1.4b). The stack configuration offers improved actuation along the polarization direction by utilizing the higher d33 coefficient. Higher forces can be developed without increasing the applied voltage. However, a stack actuator is much more difficult to incorporate into a smart material system given itâ&#x20AC;&#x2122;s out of plane bulk. For planar geometries, the electrode configuration can be modified to offer improved inplane performance.

a b

Figure 1.5 IDE pattern (a) with close-up view (b) of cross section. The electrode geometry can be modified to create an interdigitated pattern arrangement (Figure 1.5a) which utilizes the d33 coefficient to create actuation in the plane of the actuator. The interdigitated electrode (IDE) pattern defines fingers along a planar surface with alternating positive and negative polarity (Figure 1.5b). When used in place of the sandwich electrode design the IDE pattern allows polarization and actuation parallel to the wafer plane. This is similar in design intent as the stack configuration, enabling the use of the d33 coefficient. One drawback of the wafer and stack configurations is that they offer little resistance to crack propagation, as the bulk material (PZT) is essentially a brittle ceramic wafer. This drawback can be improved upon by employing a composite design, which has been enabled with the advent of PZT fibers.

c a

Figure 1.6 Piezoelectric composite configurations, particulate (a), rod (b), and longitudinal (c). The construction of PZT-based composites generally follows one of three forms, particulate (Figure 1.6a), rod (Figure 1.6b), or longitudinal (Figure 1.6c). When considering a composite piezoelectric sensor or actuator design, the electrode pattern must be configurable such that good connectivity can be assured between the electrode and piezoelectric surface. In the case of a particulate composite this is difficult since the piezoelectric material is embedded and therefore insulated by the matrix material. Then expansion of the piezoelectric particles will be accommodated by strain in the softer matrix material, and will result in low actuation ability. The rod configuration (Figure 1.6b) offers better surface connectivity, but useful piezoelectric expansion can only be realized in the thickness direction, as in-plane expansion will be insignificant given the strain of the softer matrix material. In this case the rod configuration would not offer much, if any versatile advantage over the wafer design, unless perhaps when used in a stack actuator. Even in this use, the wafer stack will offer better performance given the larger volume of active material, unless a certain conformity or threedimensional shape is required. The longitudinal configuration however, offers continuous material in-plane, and can be combined with an IDE pattern to offer high actuation along the rod direction while retaining the advantages of a composite material design such as, lower weight, directional material properties, and better flexibility. The combination of a longitudinal configuration and IDE pattern has been realized using PZT fibers and Kapton速 printed IDE. The general term for this design is Piezoelectric Fiber Composite (PFC) and is further divided into two main types: Macro Fiber Composite (MFC) and Active Fiber Composite (AFC). MFC devices were developed by NASA [25] and are produced with rectangular cross-sectioned fibers. The fibers are cut from a PZT wafer and processed with a polymer matrix. Finally the fiber layer is sandwiched between two IDE layers. AFC were developed at the Massachusetts Institute of Technology [26, 27] and are manufactured using circular cross-sectioned fibers and are generally processed together with the electrodes. An important difference between the MFC and AFC design is the contact between the IDE fingers and fiber surface. With the MFC, the fiber surface is flat, and the IDE layer is simply affixed on top of the fiber layer. The AFC manufacturing process requires that the IDE layer be pressed into the fiber layer, to ensure that the flat IDE layer accommodates to the non-uniform AFC surface topography. Furthermore, variation in the quality and straightness of circular PZT fibers means that time might be required to properly sort poor fibers from good fibers. While better electrode coverage can be attained with circular fibers, more attention of the manufacturing process is required to ensure good coverage of the electrode layer.

1.3.5. AFC State of the Art

Figure 1.7 AFC components (a) and close-up of fiber section (b) [26]. Improvement on the limitations of wafer (low actuation strain) and stack (poor surface conformability) designs can be found in the work of Bent and Hagood [27] at the Massachusetts Institute of Technology in the original development of the AFC actuator configuration. AFC are composed of PZT fibers embedded in an epoxy matrix and sandwiched between two interdigitated electrodes (IDE) as shown in Figure 1.7. The AFC design surpasses the traditional PZT wafer in many areas. One advantage of AFC over their wafer counterpart is the nature of crack propagation through the device. Once a crack initiates, it can quickly propagate through a brittle PZT wafer and invoke severe damage and performance degradation in the device. As with all fiber/matrix based composites, AFC, when compared with the bulk material, reduce the effects of catastrophic crack propagation by transferring load from the fibers to the matrix. In addition the matrix component might fill surface void defects on the PZT fiber surface, and thereby reduce the onset of cracks due to surface flaws.

Figure 1.8 Schematic of AFC with different regions of polarization. The electrode pattern in an AFC can include two regions of polarization as shown in Figure 1.8. Due to the overlapping nature of the finger design, in the traditional design, the edges of the AFC can not be polarized [28], whereas the middle region displays sufficient overlap to be fully polarized. The edges then act as a block to actuation since AFC are generally manufactured with fibers in those non-polarized regions. Since the initial introduction, the AFC concept has been given a lot of attention in the research community. Actuation studies have been carried out to assess the long-term actuation reliability of AFC actuators [3, 28]. It was shown that over millions of

actuation cycles, no degradation was seen in actuation performance. Furthermore, the only observable damage was in the form of small burns at some of the IDE fingers. Fatigue characterization has so far included tensile-tensile fatigue at a prescribed strain level, followed by actuation testing in an unstrained state [29]. Further study by Hagood and Wickramasinghe showed that actuation performance degrades at higher applied strains when AFC actuators are laminated with GFRP and subjected to a global tensile strain (along the fiber direction). This degradation was attributed to crack opening and closure in the PZT fiber layer [28]. Despite the body of work surrounding AFC, questions pertaining to their reliability and characterization still persist. The AFC was universally used as an actuator and sensor performance was neglected. As an actuator, the piezoelectric domains will be continually realigned to the direction of polarization during operation. For SHM applications, it is important to investigate the sensor properties of the AFC, without the benefit of continued repolarization. When used as a strain sensor (for example) do fatigue loading conditions lead to irreversible ferroelastic depolarization? Furthermore, the mechanisms leading to degradation in AFC performance have not been fully addressed. It was proposed by Hagood and Wickramasinghe that fiber cracking leads to actuation degradation at higher strains [28], but a description of that cracking and how it might be addressed to improve higher strain performance have not been presented. The AFC was developed for applications such as rotor blade actuation. However, while some work concerning the integration of wafer actuators into composite laminates exists, a comprehensive investigation into how AFC and similar devices might degrade the reliability of laminate materials including GFRP and CFRP laminates has not been presented. Does the interaction between laminate lay-up and design have an effect on the actuation or sensing performance of the AFC? Can the AFC be packaged in a certain way to improve its performance? 1.3.5.1. AFC Design When assessing the reliability of a system, it is important to evaluate the individual components to determine the limiting factors, and define which are critical, which are redundant, and those that can be modified to improve system performance. Despite the inherent electrical-mechanical complexity of the AFC, it is composed of a few key components: fiber, electrodes, insulating layer, and matrix. Any modifications of the AFC must consider these basic components. 1.3.5.1.1. Fiber The fiber defines the base component of the AFC, as well as the limiting factor concerning actuation and electrical mechanical abilities. Given its ceramic nature, the fiber also is the limiting factor concerning the tensile material limit and crack propagation behavior. Improving upon the actuation ability requires changing the fiber to one with a higher d33 coefficient. This would be ideally accomplished with the use of single crystal fibers [30]. Bulk PZT single crystals exhibit a higher d33 coefficient than polycrystalline PZT; however, the commercial manufacture of single crystal fibers has not yet been realized to rival the extent of polycrystalline PZT.

Therefore, the use of polycrystalline PZT fibers is logical for research and application purposes. In tensile loading, the PZT fiber can be characterized according to the applied stress and the pre-stress. The fiber is initially in a state of compression due to the nature of the material. As an external tensile stress is applied, the external stress eventually balances with the pre-stress state, beyond this limit the fiber will fail [31]. One method of improving the tensile strain of the AFC is to increase the initial pre-stress loading state on the fibers. However, this strategy has various implications for AFC performance. Increasing the pre-stress increases the strain which the fibers can experience before catastrophic failure. While modifying the mechanical pre-stress improves the tensile limit, the effect on the electrical-mechanical properties must be considered as well. Increasing the pre-stress also acts to increase the blocking force on the fibers. When used as an actuator, the AFC develops a certain force in relation to the applied voltage. If a pre-compression exists on the fibers, this force must be overcome before a global actuation can be communicated to the surrounding environment. On the other hand, research into PZT has in fact shown that a precompression on the material will enable an increase in actuation efficiency. This is one design philosophy behind the LIPCA, and Rainbow/Thunder actuator designs. 1.3.5.1.2. Electrode Modification of the IDE implies changing the material or geometry of the electrode fingers. The finger geometry and spacing can be modified by enlarging the finger width, or changing the finger spacing. Increasing the finger spacing increases the uniformity of the applied electric field between positive and negative electrode fingers. However, higher electric fields are then required to produce actuation due to the larger distance between electrodes. Increasing the width of the electrode fingers decreases the inhomogenity of the electric field below electrode fingers, but at the same time increases the dead-zone which reduces the actuation ability of the AFC. Decreasing the finger width and increasing the finger spacing can lead to an increase in actuation ability. However, this strategy also requires increasing the actuation voltage to accommodate the larger span between electrodes. Decreasing the finger spacing allows a lower actuation voltage, but also increases non-uniformity along the fiber length and the number of dead zones. As finger spacing decreases, the risk of shortcuts between electrodes increases. In stack and wafer configurations, the electrodes are separated by the dielectric piezoelectric material, and can be made as thin as possible. In the AFC, the finger spacing constitutes the shortest path for the electrical field to travel. In the AFC, the electric field is forced to take a longer path through the fiber. This path is taken only so long as sufficient insulation exists to prevent a direct path between fingers of opposite polarity. Given a large enough applied voltage, a break-down will occur and a short-cut will be created, leading to failure of the actuator. In this sense, the current IDE geometry with a finger width of 200 Îźm and a spacing of 900 Îźm represents an ideal design.

Modification of the electrode layer also can mean moving away from the sandwich construction, where the electrodes cover the top and bottom of the fiber layer, to using electrodes which completely encircle the fibers. The actuation ability of an AFC rests in the distribution and strength of the applied electric field. The distribution of the electric field is a function of the electrode finger geometry. In principle the traditional IDE design only utilizes part of the active fiber volume. Improving upon the contact angle of the IDE fingers should then lead to an increase in actuation ability. Work by Pini [32] realized an IDE pattern with 360° coverage of the fibers, and represents the extreme of IDE geometry modification. In this case the electrodes were grown around the fibers, not only offering complete coverage, but also becoming a part of the fiber surface, whereas in the traditional IDE design, the fibers are pressed over the fiber mat surface, but the materials remain separate. 1.3.5.1.3. Interface Layer The insulating layer defines the external interface between the fiber layer and the environment. Kapton® has been the material of choice as the insulating layer in MFC and AFC designs. Kapton® displays good flexibility coupled with high mechanical stiffness and good strain properties as well as excellent electrical insulating abilities, retaining a dielectric strength of 303-154 kV/mm, depending on the thickness of the film. In addition, Kapton® offers excellent resistance to temperature and chemical attack. One drawback is its slight resistance to surface printing, but this can be accounted for with a proper printing technique. Despite this, Kapton® offers the best material choice as the IDE substrate/insulating layer. 1.3.5.1.4. Matrix The matrix serves to bind the piezoelectric fibers and electrodes together. For general considerations, this is the least critical component of the AFC design. A main requirement is that during processing the matrix display a viscosity low enough to enable good flow and fiber wetting. A thermosetting resin is used, as is often used with fiber-based composites. Once cured, the matrix will define the upper temperature operational limit that the AFC can endure before thermally breaking down. In one sense it would seem crucial to pick a matrix with a higher thermal resistance. However, it is known from research into piezoelectric and PZT in particular, that depolarization will occur and lead to a degradation in the piezoelectric effect with a temperature increase. In its current state, the AFC is limited to lower temperature applications. From this standpoint then, a general thermosetting resin with low viscosity is ideal for use with the AFC. The composition of the matrix can be modified. Ideally the matrix would not only provide binding between the fibers and electrodes, but also contribute electrical-mechanical properties. 1.3.5.2. Reliability of PZT-based Devices

Reliability in smart materials entails the ability of the material to maintain its active properties despite the influence of different environmental factors such as mechanical, electrical, and thermal loading; as might be expected in an application environment. Reliability studies of PZT often focus on actuation ability. PZT actuators and sensors have seen success in industry due to their reliability and robustness. Unlike SMAs, PZT-based devices will not accumulate dislocations, and although micro-crack accumulation might lead to degradation in sensor performance, continued actuation in the d33 direction should ensure realignment of disrupted electrical domains. AFC characterization work has been performed for actuation performance and the specific application of angle of twist actuation of a Boeing active material rotor [3] and characterization of PZT wafer based elements has been performed for the Thunder/Rainbow [22] and LIPCA [33] actuators. However, few studies have addressed the reliability and performance parameters of AFC or PZT wafer devices as a component in a smart structure and the synergy between the active element and the surrounding host structure. Micro-cracking has been shown to reduce the piezoelectric performance of bulk PZT by inducing a change in electric field at the crack tip and leaving a â&#x20AC;&#x153;wakeâ&#x20AC;? of depolarization during crack propagation [34]. The accumulation of cracks can then lead to performance degradation over the lifespan of a sensor or actuator.

Figure 1.9 Crack propagation behavior of PZT in response to polarization direction [35]. The electrode pattern defines the polarization character and is used for actuation or sensing, but it can also have an effect on the properties of the bulk material. Research by Fang and Yang [35] has shown via micro indentation of polarized PZT wafers, that the polarization direction greatly influences crack propagation behavior as shown in Figure 1.9. In the anisotropic polarization plane, the crack length in the direction perpendicular to the polarization direction was found to be much larger than in the polarization direction. Conversely, in the isotropic polarization plane, the cracks had similar lengths. Furthermore, Tanimoto found that the tensile strength of PZT is lower when polarized along the loading direction [21]. Research by Ru et al. have investigated the influence of electrodes on crack propagation in PZT actuators [36]. It 15

has been shown that the region near the electrodes is more susceptible to cracking than the rest of the material [37]. It has been postulated that this is due to anisotropic expansion in the regions near the electrodes. The dependence of material properties on polarization character is a very important topic from a reliability standpoint, since it implies that different electrode configurations would impart different localized material properties and could lead to localized stress concentrations. In summation, the use of PZT in the AFC or MFC forms retains a robust collection of favorable properties and, overall is the best choice for use as the active element in the development of a smart materials system. The question then becomes, of the numerous materials available, which family of materials should be used for the passive component?

1.4.

Passive Materials

As stated previously, a goal of smart materials research is to enable the design of material systems with specific designable characteristics. The ability to optimize or assign material properties such as stiffness and strength in a particular direction increases the possible material configurations. If examples from Nature are again considered, the passive component of a smart materials system should retain a great deal of adaptability and design flexibility. Many optimized structures in Nature, such as leaves, tree trunks and bird wings are composite in design. From this viewpoint, laminate composite materials are ideal candidates for use as the passive component. In modern engineering, Fiber Reinforced Plastic (FRP) materials offer the ideal passive component. FRP materials can be designed for a multitude of operating environments as the correct choice of matrix and fiber combination allows for optimization of strength-to-weight requirements and a great deal of flexibility in assigning directiondependent stiffness and strength values. Many of the afore mentioned smart materials use electrical excitation as a driving mechanism, and therefore, Glass Fiber Reinforced Plastic (GFRP) offers a good base material given its electrical insulating properties coupled with high strength and stiffness. Carbon Fiber Reinforced Plastic (CFRP) offers the ability to conduct current (which the active material can be insulated against) as well as excellent thermal and structural characteristics. With easily varied mechanical properties via fiber arrangement and ply orientation, FRP materials can be formed into a number of two and three dimensional shapes in various temperature and pressure processing conditions. From a characterization perspective, a great deal of research has been conduced into the mechanical and fatigue properties of GFRP and CFRP laminates, and they are widely available commercially. This knowledge base is important since it reduces the amount of characterization required in smart materials development using FRP products. The combination of material design freedom and manufacturing flexibility make FRP materials, in particular GFRP and CFRP the logical passive material choice in the study of smart materials systems, and no doubt this flexibility is also why current smart material applications already employ a smart material system

design based on PZT and FRP materials (tennis rackets [1], adaptive skis [2], and rotor blades [38]).

1.4.1. Integration With the active and passive material components chosen, one must now combine those elements in a robust and reliable way such that the smart material system can be manufactured and characterized. For a laminate sensor/actuator-laminate system, the most pressing question is: “How should the active and passive materials be combined with one another?” Additionally, what problems might arise from the combination of such dissimilar materials, and how will integration affect both materials, especially their performance and reliability. Integration may be defined as any technique used to incorporate an active element into a structure or material. In a laminate composite the two most basic integration techniques include “insertion” and “cutout insertion” or “embedding.” Insertion (Figure 1.10a) entails simply placing the device between the laminate layers, essentially sandwiching it in place. Embedding or cutout (Figure 1.10b) insertion involves removing laminate material, and then placing the device into the void so that the device becomes continuous with the surrounding structure or material. Combining the two techniques leads to “interlacing.” The interlacing technique involves distributing the cutout integration region through the thickness of the laminate.

Figure 1.10 Classic integration approaches insertion (a) and cutout (b). The integration of smart devices is expected to have an effect on the mechanical performance and integrity of a structure. In particular, degradation of the ultimate strength properties is a great concern. If the integration of an active element into a structure severely degrades the ability of the structure to satisfy the original design requirement, then the overall design becomes unusable. For some applications, the integration of smart elements may degrade the mechanical performance of a host structure to unsafe levels. This is of particular concern to critical weight applications such as rotor blades or airplane wings where efficient structural designs with high strength-to-weight ratios are required. It is less critical for structures with high safety factors built into their design such as bridges and similar civil structures or 17

applications which might see low mechanical loads such as acoustic actuators or vibration sensors. The relationship between smart materials integration and the structural integrity of laminate materials has been addressed in the literature in various forms. Past studies on smart materials integration have focused on the integration of PZT wafer devices in composite laminates with the active element integrated in the 0º or 90º plies with respect to the loading direction. In a balanced [0,90] cross ply laminate composed of unidirectional (UD) material, device integration in the 90º plies will naturally reduce the impact of integration on laminate integrity as compared with integrating the devices into the load bearing 0º plies. The stiffness of the 90º plies is much less than the 0º plies and will therefore carry less load in response to an applied global strain. Integrating smart devices into the low load bearing plies of a laminate is ideal since this method will have the smallest impact on the mechanical integrity of the laminate. However, in practical applications it may prove difficult to accomplish since each ply would be expected will carry load depending on changing loading conditions during its service life. Crawley and De Luis [39] investigated the integration of PZT wafer actuators into woven composites using analytical and experimental methods and proposed design criteria for PZT actuators such as a high modulus of elasticity to allow stress transfer with the surrounding laminate. In static tension, the integration of PZT devices via cutout insertion was found to have little effect on the elastic modulus, but a 20% decrease in the ultimate tensile strength of the integrated GFRP laminates was observed. Mall and Coleman [40] investigated the affect of PZT wafer integration on CFRP laminates under monotonic tensile and fatigue conditions and reported that with quasi-isotropic CFRP lay ups the integration technique does not greatly affect laminate strength, reporting a difference of only 4% between the mechanical properties of composite laminates without integrated devices and those that employed the simple and cutout insertion integration techniques with PZT elements integrated in or adjacent to the 90º low load bearing plies. Paget and Levin [41] performed similar work with PZT wafers inserted in the center of CFRP laminates and only saw a 3% difference in the ultimate tensile properties between laminates with integrated PZT wafers and the reference laminates. Interlacing the laminate plies and distributing the integration region throughout the laminate thickness has been found to influence laminate or device reliability. In a first study [42] an optimal interlacing design was proposed using finite element techniques. A second study [43] tested the design by tensile testing UD GFRP laminates with integrated glass slides to simulate the presence of active elements. Vizzini reported that increasing the taper length between the laminate and integration region reduces load transfer but also increases the resin rich zone at the integration region. Reducing the taper length improves failure at the device interface but also degrades laminate integrity. In certain applications it may be possible to maximize the ratio of passive to active plies in a design, and thereby reduce the effect of device integration on the mechanical integrity of the laminate. Adding many layers to a laminate and testing the mechanical

integrity of the system may show little affect if the number of plies is very large as compared with the number of integrated devices in the structure [44, 45]. An understanding of the failure of smart laminates is needed to design structures that fulfill the adaptive as well as the structural design requirements of a specific application. In particular, methods for determining the affect of smart materials integration on laminate integrity can be used to better characterize the reliability of smart material systems and to assess the trade off between adding actuation and sensing capabilities while influencing the mechanical integrity of the structural design. The current work focuses on the mechanical evaluation of smart device integration in CFRP and GFRP laminates via experimental methods, investigating the case where the ratio of active to passive plies is low and integration occurs in load bearing as well as non-load bearing plies. Focusing on device integration in the load bearing plies, the critical impact of device integration on laminate integrity can be investigated. Tensile testing of integrated laminates was used to establish the performance of different material and laminate designs. This included the use of Active Fiber Composites (AFC) as well as Dummy AFC (DAFC) elements. DAFC were developed as a physical model of the AFC, mimicking the longitudinal material stiffness of the AFC, and provided a cost effective method of evaluating different integration methods.

1.5.

Objectives and Project Outline

A great deal of research has been focused on the characterization of AFC, MFC, PZT wafer devices, and other smart materials. However, these studies have generally focused on the properties of the base materials, and not focused on a characterization of the smart material as component in a material system. In general, reliability studies are carried out, which look at the sensor or actuator fatigue reliability of the active element. Studies into sensor signal reproducibility and actuator performance over millions of fatigue cycles have shown PZT-based materials to be very reliable and robust under certain operating conditions. The characterization of the active element is of course crucial. A sensor must out last the structure which is has been embedded into. If the sensing component of a smart material system fails prior to the end of the service life of a product, then the dead sensor element is simply a liability to the function of the application design. However, when considering a smart materials system, one is presented with the combination of an active element and a passive support structure. The question now arises, not simply how reliable the active element is, but how the inclusion of the active element impinges on the reliability of the passive material, and how the two materials act together. The goal of the current work is the characterization of AFC as a component in a smart materials system. This encompasses the question of AFC reliability, but also concerns the reliability of the passive material, once combined with the active AFC element. The primary considerations thus, are the robustness of the AFC under different loading and integration configurations, and the impact of AFC integration on laminate reliability. The AFC can be used in a number of different ways, as an actuator, or as a sensor. A

great deal of research has already been conducted in the actuation properties of AFC. To expand the uniqueness of this work, many experiments here consider the sensor properties of the AFC.

Figure 1.11 Project map of the dissertation. The goal of the current research project is focused on characterizing and improving the reliability of AFC when used as a component in a smart material system. The essential project map is shown in Figure 1.11, and includes two primary goals: characterizing AFC performance, as well as that of the passive structural material with integrated AFC. AFC characterization is accomplished by assessing the mechanical properties of the AFC as a stand-alone element, and then as a component in different composite laminates. The strain sensor performance of AFC is used as a primary health monitoring characteristic of the AFC, since the electrical signal derived from the AFC can be used as a performance marker as well as a damage indicator, providing evidence for when the AFC undergoes damage in PZT fiber layer. The investigation methods are heavily biased towards experimental testing, and generating hard data for characterization purposes. Numerical models are employed when needed to provide a design tool for the engineering of smart materials. By employing a routine which connects material properties and polarization of the PZT fibers, different electrode designs can be assessed. To gain a better understanding for how AFC integration affects laminate materials, GFRP and CFRP laminates with different ply types are used as passive materials in the creation of active laminates. Tensile test methods will 20

be presented, showing how AFC integration affects laminate integrity. Finally, methods will be presented to improve the reliability and performance of AFC via different packaging methods, which modify the mechanical boundary conditions of the AFC in relation to the laminate material. • AFC Reliability Characterization focuses on the AFC properties as component in a smart materials system. Important topics includes: limits of AFC performance ability, damage mechanisms, and assessment of AFC fatigue reliability. • AFC Damage Evolution and Modeling seeks to quantify in the virtual environment the results observed experimentally and the use of those findings to aide in the design smart material systems with an eye towards assuring reliability of the active material. • Laminate Integrity Analysis focuses on the passive structural component once integrated with AFC, which is very important from an application engineering perspective. This chapter explores the impact of AFC integration on laminate tensile properties and how to manage that impact. • AFC Packaging Strategies investigates how to integrate the AFC with the passive material with the goal of modifying the local loading environment to the AFC to improve the robustness of the element and modify AFC characteristics while improving long tern reliability.

2. AFC Reliability Characterization 2.1. Introduction The characterization of smart materials does not currently have any specific guidelines or requirements, such as those which exist for the mechanical testing of materials (ASTM standards). Due to the multi-functional aspect of smart materials, characterization methods are in general developed for each type of smart material by researchers focused in those fields. Given the combined structural and electrical properties of the AFC, a characterization method should include or be able to differentiate between characteristics which are required for a particular application (such as actuation) and basic mechanical properties (such as tensile strength). AFC characterization work has been conducted by researchers, but so far has mainly focused on the actuation capabilities of AFC. The current characterization investigation seeks to pinpoint strain sensor performance parameters, and to identify the critical limits of the AFC during use. The functioning of the AFC in conjunction with the passive structural material must also be considered. Therefore, the AFC needs to be evaluated as a stand-alone device, as well as in conjunction with passive structural materials. Since past investigations have focused on the actuation properties, the current AFC characterization study focuses on establishing the performance and limits of the AFC as a strain sensor, which is beneficial to establishing AFC reliability for SHM applications. Specifically this study will show how the AFC performs as a strain sensor under quasi-static tensile and fatigue loading conditions. Previous work by Bent and Wickramsinge [29] have investigated the actuation properties of AFC in short and long-term timeframes, both integrated in GFRP and as stand-alone devices. In characterizing the actuation performance at different global strains, Wichramsinge used AFC laminated with layers of GFRP, strained the AFC to increasing strains and actuated the AFC in a quasi-static manner. This method reproduced the structural constraints of a helicopter blade, which was the intended application of the study. The decrease in actuation ability as a function of applied global strain was seen to decrease beyond strains of 0.20%, a deep decrease between 0.25% and 0.45% was seen, beyond which actuation ability reached a minimum value and became less strain dependent. When the global strain was reduced, AFC actuation ability dramatically improved. The decrease in actuation ability was attributed to crack formation in the PZT fiber layer. However, the nature of cracking or methods for addressing this problem has not been presented. Mechanical fatigue studies showed that if GFRP-laminated AFC are fatigued at strains above the failure level of the AFC, and then actuated in an unloaded state, AFC actuation ability is nearly identical to an undamaged AFC. This characterization work serves actuation applications, but does not address the sensor function or its reliability. When assessing the reliability of a PZT-based smart material depolarization should be accounted for. 22

For AFC, the direction of actuation coincides with the polarization direction. When an AFC is actuated, the piezoelectric domains are continually aligned along the preferred direction, so that even if depolarization occurs during use, the effect will not be prominent. For sensor applications, actuation characterization does not ensure the same performance and reliability, and previous AFC studies have not focused on sensor performance and reliability. SHM applications might require an AFC be integrated into an airplane wing, or used in wireless systems where many sensors are distributed over a structure, and the high voltages required for repolarization may not be available. Still, given the relationship which exists for piezoelectric materials, behavior as an actuation device can be very comparable to the sensor function. Therefore, the current study focuses on using the sensor function of AFC as a performance marker, while using testing methods which are comparable to those used by Wickramasinghe and Hagood in designing the AFC Characterization investigation. This will be accomplished using mechanical testing methods including quasi-static tensile and fatigue loading conditions. An outline of the current AFC Characterization study is presented in Figure 2.1 which seeks to evaluate the mechanical and electrical properties of the AFC in a smart material system. The AFC is tested as a strand-alone device, and also as part of an active laminate with the AFC integrated in a GFRP composite laminate. In these tests, mechanical integrity can be assessed using force and strain measurements, while the health and performance of the AFC are monitored via the electrical signal from the AFC. Initial testing looks at the AFC using traditional ASTM mechanical tensile test methods to evaluate the Youngâ&#x20AC;&#x2122;s Modulus and the maximum stress and strain of the AFC. The subsequent evaluation focuses on the AFC as a component in a smart materials system, where AFC are integrated into or bonded onto GFRP laminates, and subjected to strain-controlled cyclic loading. A standard test specimen was developed, based on the ASTM standard for the testing of Fiber Reinforced Plastic (FRP) materials, which established the standard specimen geometry for testing and evaluation purposes. Long-term reliability was assessed using this standard active laminate with a tensile-tensile based fatigue loading regime. Surface Bonded Fragmentation Test (SBFT) specimens include AFC bonded to GFRP laminates, which could be observed with a stereoscope during testing to monitor cracking in the PZT fiber layer. In addition, acoustic emission monitoring is used to track damage evolution during tensile loading. The active laminates were predominantly tested via cyclic strain loading, which entailed straining the active laminate to prescribed global strains, and then mechanically cycling at low strain amplitudes. This low amplitude cycling continually activates the electrical domains of the PZT fibers, and allows multiple data points to be taken for each strain level, offering good accuracy for each strain point. Based on these results, fatigue testing was conducted at levels above and below the maximum strain of the AFC to investigate long-term sensor reliability. In the different tests, the electrical sensor signal from the AFC could be acquired and analyzed to assess AFC performance and reliability as a function of the mechanical loading environment. In principle, this testing strategy is similar to the methods used by Wickramasinghe and Hagood, but focuses on sensor, as opposed to actuator performance. The fatigue tests are also established in such a way to be comparable

with the studies by Mall [40, 46] concerning the fatigue of PZT modules in FRP materials.

Figure 2.1 Outline of AFC Characterization strategy.

2.2.

Specimen Preparation

2.2.1. Materials Glass fiber epoxy laminates were produced with a vacuum bagging technique using Isopreg HR 320P-40 plain weave pre-impregnated (pre-preg) composite plies supplied by Isolvolta (www.isovolta.com). AFC were produced using 250 μm diameter PZT5A fibers produced by Smart Material Corp., Osprey Fl, USA (www.smartmaterial.com). The fibers were laminated with an epoxy system (Aradur 2954 hardener with Araldite LY564 resin) supplied by Vantico AG, Basel, Switzerland (www.vantico.com) and sandwiched between two sheets of 25 μm thick Dupont Kapton® 100 HN film (www.dupont.com) with screen-printed IDE. Electrical connections for the AFC were made using 250 μm POLYSOL-180 wires supplied by Elektro-Feindraht AG.

2.2.2. Active laminate manufacture 2.2.2.1. AFC AFC were manufactured by hand using 250μm diameter piezoelectric fibers and epoxy resin. The fibers were arranged, secured with tape and then the fiber layer was transferred to an aluminum tool plate with a Kapton® film printed IDE. The epoxy 24

was prepared with a ratio mass composition of 3:1 epoxy to hardener. Epoxy was applied to the fiber layer and the tool plate was degassed at 60°C to remove air from the matrix. Wires were connected to the electrodes using a conductive silver epoxy. A second Kapton®/IDE layer was laid on top and a pressure was applied for 1 hour at 120°C. The AFC were post cured for 8 hours at 160°C to ensure full cure. AFC with two different shapes (long and short) were manufactured as shown in Figure 2.2. The short AFC is the standard shape used for general applications, but is not long enough to be used for tensile testing. The long AFC was used specifically for tensile testing. AFC with three different configurations were manufactured for tensile testing (dimensions 2 x 15 cm). The first set included three AFC composed of only Kapton® and PZT fibers with the epoxy matrix. The second set included three AFC with screen-printed IDE. Finally the last set consisted of three AFC with IDE, and were poled for 20 minutes with an applied voltage of 2.5 kV at 80°C. Eight AFC (dimension 3.3 x 4.0 cm) were produced for integration into the GFRP laminates. The essential differences between the two types are listed in Table 2.1. Both AFC types were manufactured by hand using 250 μm diameter piezoelectric fibers, IDE screen printed Kapton® (200 μm finger width, 900 μm finger spacing) and epoxy resin.

(b)

(a)

Figure 2.2 Different AFC types including long (a) and short (b). Table 2.1 Dimensions of AFC for tensile and integration analysis. AFC Type Short Long

Active Area (mm2) 620 1420

Length (mm) 33 150

Width (mm) 40 20

Thickness (mm) 0.315 0.320

2.2.2.2. Composite Laminate manufacture 25

Laminates composed of plain weave fiberglass plies arranged in an orthotropic stacking sequence were manufactured in a hot press using a vacuum bagging and pressing process. Plies were prepared by cutting sections of the pre-preg strips to dimensions of 30 x 30 cm. An aluminum tool plate was pre-heated to 80°C and placed on a waist level cantilever stand for proper arrangement of the plies. Pre-preg plies were arranged on the hot tool plate in a symmetric orthotropic manner. The plate was placed inside a vacuum bag and then placed in a press (pre-heated to 80°C) and vacuum was applied via an external pump. The press was closed and the temperature was raised to 120°C. When the temperature of the press reached 120° C a pressure of 5 bar was applied to the tool plate. A schematic of the laminate and vacuum bagging materials is shown in Figure 2.3. According to manufacturer specifications the resin system would fully cure after six minutes at 120°C. The plate was removed from the press and the vacuum bag after 30 minutes to ensure a full cure cycle of the laminate.

Figure 2.3 Figure Vacuum bagging setup. 2.2.2.3. AFC integration Short AFC were integrated inside GRFP laminates (thickness 2.2 mm) using cutout insertion as shown in Figure 2.4. Cutout insertion was used, offering conformity between the AFC and the surrounding laminate. The integration position was as far from the laminate symmetric axis as possible. This position represented the ideal location from a smart materials applications standpoint. From classical beam theory it is known that the maximum stress in a beam due to bending occurs at the beam surface, the maximum distance from the neutral axis. Accordingly, this position offers optimal positioning for actuation of the material as well as sensing of vibrations. Figure 2.3 shows the numbered sequence used to define ply positions in the 9-ply laminates. The elements could be placed either “center,” “off-center,” or “far offcenter” with respect to the middle of the laminate cross-section. Two integrated laminate sets were manufactured, each consisting of four test specimens. One with AFC placed in the 5th ply position and the other with AFC in the 2nd ply as shown in Figure 2.4.

Figure 2.4 Cross-section of a laminate showing the placement positions of devices integrated via cutout insertion.

Figure 2.5 Schematic of a laminate with AFC arranged and wires with Teflon/Mylar protection. An outline of element placement was mapped out on a ply using a permanent pen and drafting tools. Using the cutout insertion technique, sections corresponding to the size of the AFC were cutout and removed from the pre-preg ply. The latter ply was laid down and AFC were carefully placed in the cutout cavity. A schematic of AFC and the layout of plies and wire connections is shown in Figure 2.5. Protective layers of Teflon sheet and Mylar film with double-sided tape was placed on the edge of the integration ply to protect the wire connections from resin flow during processing as shown in Figure 2.6.

Figure 2.6 Schematic side view of Teflon/Mylar wire protection. 27

2.2.3. Surface Bonded Fragmentation Test (SBFT) Probes The SBFT specimens were manufactured by bonding AFC to a 4-ply GFRP laminate, with dimensions identical to the integrated specimens described previously. The goal of the SBFT test was to observe damage in the PZT fiber layer as it occurred. For this reason, only four plies of GFRP were used, so that light could be filtered through the laminate if desired to illuminate the PZT fiber geometry and IDE pattern. SBFT specimens were prepared with AFC oriented along (0°) and at 45° to the laminate loading direction.

2.2.4. Overview of Active Laminate Probes Active laminates with integrated AFC were tested in a variety of configurations in different loading environments including quasi-static as well as fatigue testing. Table 2.2 summarized the quasi-static experimental probes, while the fatigue specimens are summarized in Table 2.3 (AFC integrated in the middle of the laminate thickness). Table 2.2 Specimens for quasi-static strain sensor performance evaluation. Set Configuration Position AFC Orientation 1 Integrated Center 0° 2 Integrated Far Off Center 0° 3 Surface Bonded 0° 4 Surface Bonded 45° Table 2.3 Specimens for fatigue strain sensor performance evaluation. Set AFC Orientation Median Strain Strain Amplitude F1 0° 0.05% 0.01% F2 0° 0.06% 0.05% F3 0° 0.20% 0.02% F4 0° 0.20% 0.05%

2.3.

Experimental

2.3.1. AFC tensile test Long AFC (Figure 2.2a) were tested to failure using a servo-hydraulic machine (Instron 1251) at an extension rate of 0.3 mm/min. Specimens were prepared with dimensions of 2 x 15 cm with tabs of 2 x 2.5 cm (resulting gage length of 10 cm) glued to the ends with Aerolite epoxy glue. In addition, HBM strain gages were affixed to

the AFC surface. Youngâ&#x20AC;&#x2122;s modulus and the ultimate mechanical properties were calculated from the resulting stress-strain curves.

2.3.2. Mechanical oscillation test Strain sensor performance of the integrated short AFC specimens was measured via a monotonic cyclic load test. Test specimens had dimensions as described in Figure 2.7 and a servo-hydraulic test machine (Instron 1251) with 5 cm grips and 200 kN load cell were used. Unidirectional HBM 350ÎŠ strain gages calibrated to 1% strain were affixed to the specimens and a strain controlled test procedure was used to oscillate the specimens around a prescribed strain level at 1 Hz for ten seconds with strain oscillation amplitude of 0.01% as shown in Figure 2.8. Specimens were loaded twice to the desired maximum strain for each test in strain steps, which were defined as an increase of 0.05%. Each specimen was loaded through the strain steps, up to the ultimate level and then taken back to zero strain before again loading to the same level. As the specimens were cycled, the signal from the AFC was recorded. The sensitivity (SAFC) was defined as the amplitude of the resulting electrical signal as shown in Figure 2.8. Three specimen sets were tested. The first two sets consisted of AFC with the PZT fiber oriented along the direction of laminate loading. Four specimens with AFC integrated in the center position and were loaded from 0.05% to 0.70%. This established the maximum and minimum performance behavior. A second set of four specimens was then tested with AFC integrated in the far off-center position with upper strains of 0.20%, 0.25%, 0.30%, 0.40%, and 0.50%. This second set investigated the transition behavior of performance identified in the first test set. The second set also included acoustic emission monitoring to investigate damage evolution in the test specimen. During testing the resulting AFC signal and strain data were collected via a data acquisition PC card and LabVIEW software as shown in Figure 2.6. The data acquisition card had a range of -5 to +5 volts, a range which would be exceeded by the output voltage of the AFC. For this reason a capacitor with a capacity of 1277 nF was connected in parallel to the AFC during the test to reduce the voltage on the PC card. Given the electro-mechanical coupling inherent in PZT performance, characteristics observed using the sensing function of AFC also gives great insight into the actuation performance given the same mechanical environment.

Figure 2.7 Experimental set-up for the monotonic cycle test of 30 x 5 cm tensile specimens.

Figure 2.8 Applied strain program for monotonic cycle test.

2.3.3. Acoustic Emission Monitoring Acoustic Emission (AE) measurements were performed during the cycle test on integrated AFC to lend insight into damage evolution at increasing applied strain levels. The placement of AE sensors on the test specimens is shown in Figure 2.7. Standard 150 kHz resonant sensors (type SE-150M from Dunegan Engineering Inc., DECI, diameter 20.5 mm, height 14 mm, total weight 12 g including case) were used. AE signal parameter acquisition and analysis was performed with commercial AE equipment (AMS-3 and software from Vallen Systeme GmbH). Detection thresholds were set at 50 dBAE with a gain of 34 dB and a rearm time of 3.28 ms. A silicon-free vacuum grease was used as coupling agent between the sensors and specimen and sensors were mounted by wrapping with duct tape. Hsu-Nielsen sources (pencil lead breaks) provided simulated AE signals for verification of the sensor coupling and of the location accuracy. AE source location was determined from linear interpolation based on arrival time differences using an average signal propagation speed. Simulated AE on the laminate top and bottom edges of the AFC provided reference points for the determination of the average signal propagation speed.

2.3.4. Surface Bonded Fragmentation Test (SBFT) A major drawback of the integrated AFC specimens was that direct visual observation of damage characteristics was not possible due to the poor transparency of the laminate. For this reason additional specimens were fabricated with a short AFC bonded to the surface of a 4-ply GFRP laminate substrate (with dimensions identical to those of Figure 2.7) using AralditeÂŽ epoxy glue and subjected to the mechanical oscillation test regime. SBFT specimens were prepared with AFC oriented at 0Â° and 45Â° deg to the laminate loading direction. This allowed identification of different fracture patterns in the fiber layer in relation to AFC orientation and loading direction. To visually observe the fragmentation behavior a stereoscopic microscope was aligned perpendicular to the AFC surface. Thus, during the mechanical oscillation test damage evolution could be observed at each strain level.

2.3.5. Fatigue Testing In general, composite material fatigue tests are conduced to investigate the mechanical property reliability over a long-term time frame. An active laminate is defined by its mechanical and electro-active properties. For this reason the current fatigue investigation is designed around the limits of the AFC as determined by the tensile and cyclic tests. A primary motivation of the fatigue investigation was to determine if microcracks and depolarization occur in a fatigue state below the fiber fragmentation onset strain and affect AFC performance. Additionally, if failure has occurred in the PZT fiber layer, can the AFC continue functioning over long time periods? Or, can sensor performance be expected to decrease? 31

In previous active laminate fatigue investigations Mall has shown that the functioning of wafer PZT actuators will decrease over the course of fatigue loading [46]. Given the brittle ceramic nature of the PZT fibers, fatigue loading could initiate microcracks, which might accumulate over time and possibly lead to ferroelastic depolarization, which would also be seen in the AFC electrical signal. A strain-based fatigue test was developed using the ASTM D3479 standard for fatigue testing of composite materials. The first consideration was to either design around a stress or strain methodology. Based on the results of the tensile tests as well as the cyclic testing of AFC active laminates, it was decided to use a strain-based test definition, centered around the failure strain of the AFC. The AFC tensile test and integrated GFRP cyclic testing show that fiber failure and fragmentation begins at approximately 0.20%-0.25% strain. GFRP laminates with center integrated AFC were subjected to tensile-tensile fatigue loading. The median strain level and applied strain amplitude were established in relation to the failure limit of the AFC as observed from the tensile measurements (refer to Results section). Quasi-static fatigue tests were conducted with strains either above or below 0.20% strain. The strain ranges are depicted in Figure 2.9. The maximum and median strain parameters and number of cycles (N) are listed in Table 2.4 and graphically depicted in Figure 2.10. The quasistatic fatigue measurements were strain controlled with a frequency of 5 or 6 Hz and the signal from the AFC was recorded by the Instron controller. The tests included two separate stages; initially the specimens were fatigued with a strain-signal recorded for each strain cycle. The resulting signal was monitored until any observable damage had saturated as evidenced by a plateau in the AFC signal. The test was then paused and continued with data collection occurring every one thousand cycles until test termination and lasted for between 3 and 9 million cycles. Initially specimens were fatigued for longer time periods to determine critical changes in the AFC strain-signal behavior.

Figure 2.9 Schematic of strain ranges for fatigue experiments.

Figure 2.10 Description of strain parameters for fatigue tests. Table 2.4 AFC fatigue test specimens including strain level and fatigue cycles. Specimen Strain Range N R εr (%) εa (%) εm (%) 6 AFC297 0.05%/0.01% 9x10 0.67 0.02 0.01 0.05 6 AFC299 0.06%/0.05% 9x10 0.11 0.10 0.05 0.06 AFC335 0.20%/0.05% 3x106 0.25 0.10 0.05 0.20 AFC340 0.20%/0.02% 10x106 0.22 0.04 0.02 0.20 AFC341 0.20%/0.05% 6x106 0.25 0.10 0.05 0.20 6 AFC342 0.06%/0.05% 1 x10 0.11 0.10 0.05 0.06 33

2.4.

Results/Discussion

2.4.1. AFC mechanical properties The characteristic electrical-mechanical AFC tensile response is shown in Figure 2.11. The mechanical response was nonlinear, consistent with the observed tensile behavior of bulk PZT [19, 47] as well as that of PZT fibers [48] from the literature. The elastic modulus was calculated from the slope of the initial linear response between 0.01%0.03%. From a sample pool of 6 specimens, the average stress at failure (Ď&#x192;MAX) was 47 MPa with standard deviation of 8.2%. A Youngâ&#x20AC;&#x2122;s Modulus (E) of 29 GPa was found with a standard deviation of 4.7% and an ultimate strain (ÎľMAX) of 0.25% with a high scatter of 16.8% standard deviation. These results correlated with tensile test results from the study of Wickramasinghe and Hagood [28], which used AFC specimens with a similar design (IDE pattern, fiber size, AFC thickness) as in this study. AFC in the present study had a linear density of 3.33 fibers per millimeter as compared with 3.6 fibers per millimeter for the AFC used in [28]. Findings from bulk PZT tensile testing [19] suggest that polarized samples exhibit greater compressive strength (when loaded parallel to the polarization direction) than non-polarized samples. However, no statistical difference was seen between polarized and nonpolarized AFC specimens in this study. This may have been due to the non-uniform polarization nature of PZT fibers in AFC. While fiber sections in between opposing polarity IDE fingers are polarized along the loading direction, the material sandwiched in between parallel electrodes is not [26].

Stress (MPa)

8 20

0 0.00

0.05

0.10

Stress Response

Signal Response

0.15

0.20

AFC Signal (C)

10 30

0 0.25

Strain (%)

Figure 2.11 Typical electrical-mechanical response of a AFC tensile specimen. 34

The AFC response shows a slight nonlinearity beyond 0.15%, but is predominantly linear in character. Compressive depolarization behavior of PZT [18] is well documented in the literature. Under compression bulk PZT exhibits a linear charge output until domain switching begins due to an increase in applied strain resulting in a nonlinear profile. In tension loading the extent of ferroelastic switching is generally not investigated since failure occurs at very low strains and it has been shown that ferroelastic switching is not significant prior to tensile failure [47]. In either case, depolarization would be evidenced by a noticeable slope change in the electrical response due to the switching of electrical domains. While some depolarization of the AFC appears prior to failure (evidenced by the slight nonlinearity), it is evident that tensile failure of the specimens poses a larger threat to material reliability.

2.4.2. Influence of mechanical strain on AFC strain sensor performance

Normalized Response Amplitude

1 0.70% 1st Run

0.8

0.70% 2nd Run

0.6 0.4 0.2 0 0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

Strain (%)

Figure 2.12 Results from the first oscillation test set, used to establish the sensitivity limits.

0.20% 1st Run 0.25% 1st Run 0.30% 1st Run 0.40% 1st Run 0.50% 1st Run

Normalized Amplitude

0.8

0.20% 2nd Run 0.25% 2nd Run 0.30% 2nd Run 0.40% 2nd Run 0.50% 2nd Run

0.6 0.4 0.2 0 0.00

0.10

0.20

0.30 Strain (%)

0.40

0.50

0.60

Figure 2.13 Amplitude response of the second oscillation test set, showing the transition in performance behavior via loading to intermediate strain levels. AFC strain sensitivity performance in response to the quasi-static 1 Hz cyclic strain test is shown in Figure 2.12 and 2.13. Both curves were produced by averaging the results of each test set. Figure 2.12 shows the results normalized against the maximum attained sensitivity, where sensitivity is defined as the amplitude of the AFC signal response during oscillation. Sensitivity, as measured from the AFC output signal ranged from 0.016 ÎźC to 0.14 ÎźC. The first oscillation test data set is shown in Figure 2.12 where samples were loaded from 0.05% to 0.70%, unloaded, and then reloaded to 0.70%. After the essential behavior was established the second test set, with results shown in Figure 2.13 established the intermediate strain performance with upper strain limits reaching 0.20%, 0.25%, 0.30%, 0.40%, and 0.50%. The intermediate load curves (Figure 2.13) show that the 2nd cycle of each strain step exhibits quite similar behavior of the 1st cycle of each higher subsequent strain step. For the intermediate strain curves a slope change is seen at the end of each curve. The most prominent slope changes are seen at 0.20% and 0.45% with a near-linear signal degradation path existing between 0.20% and 0.45% strain. The strain signal amplitude drops off severely beyond 0.20% and approaches a minimum at 0.50%. This response is similar in character to that that seen by Mall [49] in the monotonic loading of PZT wafers embedded in carbon laminates as well as by Wickramasinghe and Hagood in similar experiments [28] for actuation of AFC while under tensile loading. In both studies PZT response dropped off significantly beyond strains of 0.20%. One of the most interesting findings however, is that performance recovery occurred at lower strain levels. The performance at 0.05% after loading to each strain level is listed in Table 2.5. Despite sensitivity degradation at higher strains, when reduced to 0.05% specimens exhibited performance recovery. Recovery was lower when loaded to 0.70%, exhibiting only 70% of the original sensitivity. However, at the lower strains (0.20%-0.40%) performance recovery was nearly 100%. Fiber breakage is a likely explanation for the performance degradation. A disruption in the

electrical field paths in the PZT fibers due to cracking would hinder performance and explain the drop in sensitivity. Table 2.5 Performance recovery of integrated AFC at 0.05% strain after loading to each strain level. 0.25 0.30 0.40 0.50 0.70 ε (%) 0.20 State 100% 98% 93% 93% 85% 70% Figure 2.14 shows the absolute AFC signal obtained during oscillation for each strain step. The absolute signal, defined as the AFC output generated at the maximum strain during oscillation of each strain step, ranged from 1.16 μC to 4.98 μC. During the test, each sample was loaded twice to the desired maximum strain level in a stepwise fashion before unloading and then moving on to the next strain step. In Figure 2.14 only data from the second loadings are depicted so as to give a clear representation of the behavior. Here the AFC signal at the maximum strain for each oscillation level was normalized against the maximum value obtained during the 0.05% strain step. The absolute signal character can be useful in characterizing the ferroelastic behavior of bulk PZT with applied strain [50]. The AFC signal is linear within the 0.05%-0.20% range, which is consistent with the tensile test results. At strains above 0.25% the signal exhibited increasingly nonlinear behavior and the initial linear slope decreased. This coincides with the ultimate AFC strain (εMAX = 0.25%), pointing towards cracking or material damage to explain the nonlinearity. Here the nonlinearity is quite evident as strain increases, pointing to fiber damage, thereby reducing the ability of the AFC to release charge in response to increased strain levels. Straining of piezoelectric domains is the heart of PZT sensor operation. Crack opening would create free surfaces in the PZT layer, and as global strain increases the gaps due to crack formation would increase. Consequentially, the fibers would not strain as much in response to the applied strain on the specimen, giving rise to nonlinearity in the signal response. When loaded from 0.05% up to 0.70% AFC sensitivity shows stability in the strain range of 0.05%-0.20%. A slope change is seen when the applied strain reaches 0.25%. Beyond that level sensitivity decreased in a near-linear manner until a second slope change at 0.45%. Beyond 0.45% sensitivity continues to decrease, but at a much reduced rate (near-plateau). When loaded a second time to 0.70% sensitivity dropped off dramatically in the range of 0.05%-0.10%. It then approached the near-plateau character exhibited in the first loading beyond 0.45%.

Normalized AFC Signal (C)

1 0.00

0.20

0.20%

0.25%

0.30%

0.40%

0.50%

0.70%

0.40 Strain (%)

0.60

0.80

Figure 2.14 Absolute signal response of short AFC during the cycle test.

2.4.3. SBFT Observations and Damage Evaluation In the SBFT investigation an AFC was bonded to a 4-ply GFRP laminate and strained with the cyclic strain step loading test, while observed with a stereoscope to monitor damage evolution in the fiber layer. The goal of performing the SBFT was to visually identify the fragmentation behavior in the PZT fiber layer of AFC during tensile loading. In the literature Single Fiber Fragmentation Testing (SFFT) has been widely used to assess adhesion properties for a fiber embedded in a continuous matrix [51]. Via SFFT it is established that fibers embedded in a matrix that are strained at a low rate will begin to fracture and eventually reach fragmentation saturation [52]. At this point the fiber would be divided into fragments of uniform length, the length dependent on the interfacial strength at the fiber/matrix interface. Due to fiber arrangement and matrix volume fraction fragmentation behavior in SFFT can not be directly correlated to damage in an integrated AFC. However, similar behavior is to be expected in that as strain increases the PZT fibers are expected to fragment, and at some point fragmentation will reach saturation. Visual observations with the stereoscope from the SBFT showed crack formation at 0.35% strain at the IDE finger edges. Cracks were identified as horizontal lines running perpendicular to the fiber direction, severing the fiber completely. Cracks were sparse at first, propagating through two or three fibers at a time. At 0.40% the number of cracks increased throughout the fiber layer. These experimental 38

observations of fragmentation saturation correlate with the actuation degradation (due to strain) model presented by Hagood and Pizzochero [31]. At 0.50% cracks were widespread over the AFC, but always resided at the IDE fingers. When the strain was reduced to zero no cracks were visible. Since the IDE are not transparent visual crack identification was impossible at lower strains. When samples were loaded again to 0.50% separation between fiber free surfaces (termed the “fragmentation gap”) was clearly evident. Fragmentation gaps increased with strain throughout the fiber layer. At low strains the fragmentation gaps closed and no damage was visible below 0.30%. This crack closure behavior explains why AFC sensitivity could recover so well at 0.05% strain. Closure of the cracks would re-enable electric field paths in the fiber layer and lead to a recovery in performance. While it was possible to visually identify damage at 0.35%, material failure occurs earlier as evidenced by the drop in performance observed in the AFC signal at 0.25%. In the 45° SBFT specimens, the AFC was oriented 45° to the direction of laminate loading and subjected to laminate strains up to 0.70%. During loading of the 45° specimens, cracks initiated at the IDE fingers in a way similar to the 0° specimens, but at a 45° orientation to the fiber axis. Cracks occurred near the IDE fingers at laminate strains above 0.45%, indicative of failure in the fiber. However, cracks oriented at 45° to the PZT fiber direction were also seen to appear between the IDE fingers, with an orientation of 45° to the fiber axis.

2.4.4. Rotated AFC Sensor Performance Rotation of AFC elements according to the direction of loading was investigated with the AFC oriented at 45° to the laminate loading direction. Comparison between different applied loading directions is more complex than the previous investigation given the indirect relationship between applied load and polarization direction of the PZT fibers when the elements are rotated with respect to the laminate load direction. When loaded along the fiber direction, the elastic component 90° to the fiber direction is very small, as is the force component, and can be neglected in the corresponding data analysis. When rotated however, the strain along the laminate loading direction will be the same, but the force on the AFC becomes a combination of loading along and perpendicular to the fiber direction as shown in Figure 2.15. Since a significant force component is now present perpendicular to the polarization direction in the fiber, the possibility of ferroelastic depolarization must be considered in evaluating the AFC sensor signal. Ferroelastic depolarization occurs when a stress is applied perpendicular to the polarization direction. At the crystallographic level the compression of the perovskite crystal disrupts the geometric arrangement of the charge imbalance related to the central ion, and encourages switching to a new location. Hence, the PZT material becomes ferroelastically depolarized.

Figure 2.15 Description of mechanical loading of the AFC in relation to polarization direction when rotated 90째. F

Figure 2.16 Classic scenarios of ferroelastic depolarization in PZT materials. Figure 2.16 illustrates the effect of tensile stress perpendicular to polarization and compressive stress applied opposite the polarization direction. These loading conditions will induce ferroelastic depolarization if they exceed the coercive stress of the PZT material, since stresses which exceed the coercive stress of the PZT leads to electrical domain switching at the crystallographic level. F2

Figure 2.17 Forces on the 45째 AFC in relation to laminate loading. Figure 2.17 shows a rotated element and the associated forces. In this case, if F2 is significant, then depolarization could occur due to tensile loading of the piezo-active

fibers. It is useful to explore the difference of damage evolution in the fiber layer and analyze the resulting strain-signal behavior since many rotor blade designs consider the use of AFC aligned at 45° to the blade axis, and it is therefore important to investigate the reliability of AFC in this configuration.

Figure 2.18 Strain sensor performance of 45° specimens. The 45° strain sensitivity curve is displayed in Figure 2.18 along with the characteristic response from the 0° specimens, where the strain from the AFC is plotted against AFC signal amplitude. An initial increase in signal sensitivity (as seen in the 0° AFC) was observed, followed by a sharp decrease, and then a continuing decrease in signal sensitivity. The slight initial increase is related to loading on the PZT fibers, which perform better when slightly elongated [53]. The recovery signal path was initially coincident with the loading path, but diverged when AFC strains exceeded 0.20%, indicative of increasing damage in the fiber layer. Degradation in performance started around a strain of 0.15% in the AFC. This value is slightly lower than that observed for AFC loaded along the fiber direction, where degradation is seen around 0.20%. This difference can be attributed to the mixed loading case which the AFC is under, leading to earlier damage initiation. The applied strain amplitude along the 45° AFC fibers was 0.001%, whereas the 0° AFC were subjected to an amplitude of 0.01%, this could account for the large difference in the sensitivity values displayed in Figure 2.18.

Figure 2.19 AFC charge output of 0° and 45° specimens, strain applied along fiber direction. Figure 2.19 depicts the absolute charge signal from the 45° AFC at each strain step. The output for the 0° AFC at a comparable strain is also included for comparison. In the lower strain levels the charge out is linear and then becomes slightly nonlinear near 0.12%, and then after peaking near 0.15%, begins to drop off. This implies that the charge output is linear and stable until damage occurs, which is in agreement with the previous findings. The initial nonlinearity corresponds to that observed in the AFC tensile test and Figure 2.11, where slight depolarization is observed prior to failure of the PZT fibers. The slope of the charge-strain curve is below that observed for the 0° AFC, this suggests that at least some ferroelastic depolarization is occurring, resulting from the perpendicular force on the PZT fibers. This depolarization however, appears to be reversible, which coincides with findings from tensile experiments on PZT from the literature [47].

Sensor 2

AFC

Sensor 1

Distance From Sensor 1 to Sensor 2 (cm)

0 0.20

0.25

0.30%

0.40%

Maximum Applied Strain Level

0.50%

1st Loading 2nd Loading

Figure 2.20 Plot of AE event signals located along the AFC specimen during monotonic cyclic testing.

2.4.5. Acoustic Emission (AE) Monitoring AE monitoring was also used to characterize damage evolution in relation to applied strain. The location of AE events for each strain step is plotted in Figure 2.20. In the vicinity of the short AFC, multiple damage mechanisms might produce AE signals. Crack formation or crack propagation, as well as fiber-matrix debonding and fiber breakage could occur in the GFRP laminate, as well as in the PZT fiber layer, including fiber breaks, or delamination between the Kapton® film and the PZT fibers or at the Kapton®/GFRP interface. The Kapton® film represents a soft layer (εmax = 75%) as compared with the PZT and GFRP components. Therefore, it is able to withstand a great deal more deformation than the PZT or GFRP. For this reason, delamination or failure at the Kapton® interface is not to be expected at these strains. As described previously, specimens were loaded twice to each prescribed strain level. During the monotonic tests, the first strain step yielded AE events throughout the laminate. During the second loading to the same strain however, the number of AE events located in the GFRP is greatly reduced due to the Felicity effect [54]. Upon step-wise loading to higher strains, the first loading always yielded AE signals located both in the GFRP and AFC. However, for strains of 0.25% and higher, the majority of located AE events originated in the region of the integrated AFC. In the second loadings, AE signals were almost exclusively located in the area of the integrated AFC. While for the strain step to 0.25% the AE signal amplitudes reached values up to 75 dBAE in the first loading, they were considerably lower (up to 65 dBAE except for

two events) in the second loading. For all other strain steps the AE signal amplitudes did reach values of 80 dBAE or more during both, first and second loading to the respective strain. The majority of the located AE events with amplitudes above 75 dBAE originated from the area of the AFC and very few from the remaining GFRP area. Even though an AE amplitude analysis does not yield unambiguous identification of the source mechanisms, the large amplitude AE signals (> 75 dBAE) are consistent with the formation of larger damage zones (such as delamination) or with fiber breaks [55]. Below 0.25% the distribution of located AE sources was fairly uniform along the specimen length, where no distinction seems to exist between AE signals from the GFRP or the AFC. From the event plot (Figure 2.20) it is evident that some AE activity occurs in the AFC vicinity at 0.20%. However, 0.25% shows a much greater cluster of AE events originating from the area of the AFC, pointing to the onset of PZT fiber damage. This point corresponds to the slope change and nonlinearity seen in the AFC signal amplitude (Figure 2.12, 2.13) and absolute response curves (Figure 2.14). The clustering of AE events at the AFC region increases dramatically at 0.40% and 0.50%. The large number of AE events occurring at 0.40% and 0.50% corresponds to the slope change and plateau performance regions of the AFC sensitivity and absolute response curves. The current AE monitoring data can not fully characterize damage in the AFC due to the various phenomena (delamination, GFRP/PZT failure) that may produce AE signals. However, when interpreted in conjunction with visual and AFC output observations the AE results strongly suggest that material damage in the AFC initiates between 0.20% and 0.25% strain, and reaches a saturation between 0.40% and 0.50%.

2.4.6. Integrated AFC Fatigue AFC were integrated into GFRP laminates and fatigue loading was applied at different strain configurations in a tensile-tensile fatigue environment. In general, the AFC signal showed an initial decrease in performance in the first hundred cycles, after which the AFC signal retained a uniform character. When subjected to fatigue strains below the AFC tensile limit, very good signal reproducibility and stability were observed. This is depicted in Figure 2.21, where the most significant change in AFC signal occurs in the first 100 cycles. The signal then plateaus, and remains constant for the duration of each test.

Figure 2.21 AFC fatigue behavior showing initial signal degradation. When fatigued below (AFC 342) the tensile limit the AFC displays a higher signal than when fatigue above the tensile limit (AFC 335,340,341). The longest fatigue test (AFC 340) was run for 10 million cycles, but aside from the initial signal degradation the signal remained constant. This data shows that even in the damaged state, the AFC can retain its ability to reliably produce a strain sensor signal. This is a great improvement when contrasted with the findings of Mall [40] concerning the fatigue of PZT wafers. Mall investigated PZT wafers integrated into CFRP laminates, where failure of the PZT element was defined as a 40% drop in initial signal strength. PZT patches were only reliable up to their tensile strain limit of 0.10%. By comparison the signal from AFC 341, which was subjected to a maximum strain of 0.25% (above the tensile limit) dropped by only 13% before achieving a plateau character, which was maintained for six million cycles. These findings show that AFC exhibit superior reliability as compared with PZT wafer devices. The primary difference between AFC fatigue above and below the tensile limit is seen in the development of a hysteresis in the strain-signal loop and the performance decrease as shown in Figure 2.21. Figure 2.22 shows the strain signal loops for AFC 299 (0.06%/0.05%) where the strain-signal behavior is nearly linear and constant over millions of cycles, to the extent that the 1000th and 3 millionth are nearly identical. This behavior is similar to that seen in the AFC tensile test (Figure 2.9), where a slight non-linearity in the sensor signal was reversible just prior to failure. This non-linearity was taken to be evidence of reversible ferroelastic depolarization. Since the nonlinearity is reproduced and the signal becomes linear in the stain range near the middle strain value, this implies recovery of any ferroelastic depolarization. This is significant, since it shows that the depolarization is reversible, and no noticeable signal degradation is apparent due to the fatigue loading. A continued decrease in the AFC sensor signal would be indicative of irreversible depolarization, due to depolarization 45

at the crystallographic level, or due to the accumulation of micro-cracks in the fibers, which would be evidenced by a change in slope of the fatigue curve.

AFC 299 Fatigue 0.06%/0.05% 2.E-06 1000 1 million 3 million

AFC Signal (C)

1.E-06 5.E-07

5000 2 million

0.E+00 -5.E-07

0.02

0.04

0.06

0.08

0.1

0.12

-1.E-06 -2.E-06

Strain (%)

Figure 2.22 AFC fatigue behavior below the tensile limit of the AFC.

AFC 340 Fatigue 0.20%/0.02% 3.E-07 1e3 3e6 6e6 9e6

AFC Signal (C)

2.E-07 1.E-07 0.E+00 0.17 -1.E-07

0.18

0.19

0.2

0.21

1e6 4e6 7e6 1e7

0.22

-2.E-07 -3.E-07 Strain (%)

Figure 2.23 AFC fatigue behavior near the tensile limit.

2e6 5e6 8e6

0.23

In the median strain range (Figure 2.23) the AFC is strained above the tensile limit, and a hysteresis has developed. Again, the signal does not change over ten million cycles, but any cracks in the fiber layer would be opened. However, since the signal is not changing, it can be assumed that no new damage or microcracks or depolarization is occurring. The fatigue data shows very good reliability of the AFC as a strain sensor, and is in accordance with the AFC actuator fatigue research carried out by Wickramasinghe and Hagood [29]. The sensor performance was found to be quite reliable, just as the actuation behavior has been shown to be uniform and reproducible over millions of actuation cycles, so long as the integrity of the PZT fibers is not compromised. The AFC fatigue results show a decrease in sensor signal performance is tied to the applied strain level and fiber damage, and not tied to ferroelastic depolarization of the PZT fibers. As noted in the tensile investigation, the PZT fibers fail and fragment before depolarization can occur. The intact fiber pieces retain their level of polarization even though the material has physically failed. Although good signal reliability was seen after damage saturation, the main factor in improving AFC fatigue reliability lies in increasing the critical strain level of the AFC. This implies that the fiber fragmentation onset and rate should be reduced, to improve AFC reliability.

2.5.

Conclusions

The performance of AFC elements integrated in GFRP laminates in a tensile strain environment was investigated, and used as a method to characterize the reliability of AFC. AFC sensor performance was found to be stable up to a strain of 0.20%. A degradation behavior is seen between 0.20%-0.40% and is attributed to fragmentation in the PZT fiber layer, which saturates beyond 0.45% strain. Crack formation occurred at IDE fingers and is attributed to the inhomogeneous polarization of PZT fibers, which is characteristic of AFC design. Despite damage to the fiber layer, sensor performance recovers at lower strains as cracks close. Due to the electromechanical coupling of PZT, AFC performance characterization as a sensor can also be used to describe actuation performance under similar loading conditions. The investigation into actuation performance under increasing strain by Wickramasinghe [28] reported nearly identical performance degradation at similar strain levels. Fatigue behavior showed that the AFC displays excellent reliability over millions of loading cycles, the only problem being, the maximum applied strain level, which was shown to have the largest impart on signal strength degradation. The current study has provided information concerning the performance limits of AFC in a mechanical environment, and identified the need to focus on reducing fiber failure and fragmentation as a primary means of increasing the long-term reliability of AFC devices.

3. AFC Damage Evaluation and Modeling 3.1. Introduction The AFC Characterization investigation showed how the strain sensor performance of AFC integrated in GFRP laminates decreases with applied strain, and highlighted that recovery of sensor ability occurred when the laminate strain was reduced. This behavior coincided with the findings of Wickramasinghe and Hagood [28] in their investigation of the actuation performance of AFC under similar testing conditions. Although the relationship between AFC actuation ability and applied strain was attributed to the formation of, and opening and closing of cracks in the PZT fiber layer, studies by Wickramasinghe and Hagood did not explain the nature or evolution of this damage. Furthermore, methods of reducing or modifying the onset of damage in AFC have not been presented. The AFC Characterization investigation findings suggest that cracking occurs near the IDE electrode fingers, one aim of the current work is to conduct a microscopy investigation of specimens from the characterization study and determine the relationship between electrode position and fiber breakage in relation to the polarization state of AFC. The current investigation will evaluate the evolution of damage in AFC, present a damage model for AFC and to show how the polarization of PZT fibers affects fiber failure. Secondly, a modeling procedure for the material properties within the PZT fibers of AFC will be used to evaluate different electrode designs and determine if AFC can be improved using different electrode geometries. An outline of these different work packages is presented in Table 3.1. The development of AFC technology has largely focused on the actuation ability and modeling of the AFC from a uniform materials property standpoint [26, 27, 56]. However, research at the material level of PZT has revealed that both the crack propagation and mechanical property behavior are largely tied to and defined by the polarization character existing within the material [19]. This has important implications for AFC, which exhibit an alternating polarization character along the fiber length. This then implies that investigation into material properties as a function of polarization would play an important role in characterizing the reliability of material systems based on PZT. From a reliability point of view, this is important for AFC design and applications as well as for other piezoelectric sensor and actuator systems. Understanding the AFC as an element is important in characterizing AFC actuation ability. At the micromechanics level modeling can help to characterize the affect of polarization on fiber material properties and to identify the weak points of the AFC design. Assessing the influence of IDE geometry on AFC reliability requires an assessment of polarization and material properties so as to identity how failure in the AFC occurs and if improvements can be made in the design. The modeling of AFC and piezoelectrics based on PZT is available in many forms. Basic piezoelectric theory for actuation and sensing includes using the permittivity and piezoelectric coefficients to describe strain as related to applied voltage loads as well as the 48

electrical signal resulting from applied mechanical boundary conditions. In general, this treatment correlates very well to experimental findings for low voltage linear and high voltage non-linear [56] models. The current work examines the analytical formulation of the AFC, the limitation of such a modeling approach, and how modeling the AFC at the micromechanical level illuminates the experimental findings related to AFC failure and reliability. Experimental finding regarding crack development in AFC is presented, along with micromechanical modeling characterization of the material property description within the AFC.

Damage Model

Present how the AFC fails and how fiber fragmentation evolves when active laminates are strained beyond the tensile limit of the AFC, based on findings from the AFC Characterization investigation.

Microscopy

Use of microscopy techniques to evaluate damage incurred by the AFC tested in the characterization investigation and present findings concerning electrical polarization and crack position in relation to electrode placement.

AFC Modeling

Table 3.1 Description of the AFC Damage Evaluation and Modeling investigation.

Use of a pre-developed ANSYS model and routine to evaluate the relationship between polarization and the development of stress concentrations in PZT fibers. Assess the ability to improve AFC integrity by modifying electrode designs.

3.1.1. AFC Damage Evolution From the AFC sensitivity and output curves combined with visual observation of crack behavior and AE monitoring, a model of damage evolution can be put forth. Damage evolution occurs in three stages: crack formation, extension, and saturation with a corresponding graphical representation of performance and applied strain illustrated in Figure 3.1.

Figure 3.1 Damage model of integrated AFC in relation to applied strain. In Figure 3.1 AFC damage evolution is described using three curves and five numbered slope changes, which mark critical points. The “Initial” curve represents the loading of a sample from a strain of 0.05% to 0.50%. The “Intermediate” curve depicts behavior for a sample loaded within the strain range of 0.05%-0.40%, unloaded, and then reloaded to the same level. The “Saturation” curve represents a sample that has been loaded and then reloaded from 0.05% to 0.50%. When loaded along the Initial curve, AFC performance is stable up to 0.20% where a slope change occurs at Marker 1. Cracking and subsequent fiber fragmentation occurs in the PZT fiber layer until the second slope change at Marker 2. Marker 2 represents the fragmentation saturation point, beyond which little new damage occurs in the fiber layer. In the Intermediate curve, AFC performance drops in a concave or inverse fashion until Marker 3. In this region, cracks which were introduced in earlier loadings are opened up due to material strain. The fiber layer remains stable however, due to gap extension between crack free surfaces until the slope change at Marker 4. At Marker 4 the fiber layer has exceeded the maximum strain from the previous loading and new cracking occurs, the behavior following that of the Initial curve. The Saturation and Intermediate curves are similar in the initial loading, with crack opening occurring until the slope change at Marker 5. Since fragmentation has already reached the saturation point, only gap extension is present and follows a plateau region, which converges with the Initial curve beyond 0.50%.

3.2.

Microscopy Investigation

The microscopy investigation of the AFC specimens was used to reveal the nature of fiber failure in the AFC under different mechanical loading and electrical polarization conditions. AFC microscopy specimens were prepared from the mechanical testing

specimens used in the AFC Reliability Characterization study. The specimen pool included AFC oriented at 0º and 45º to the mechanical loading direction, either bonded to a GFRP laminate (via the SBFT tensile test), or integrated into a GFRP laminate. Specimens were also investigated with AFC which had been polarized, but not mechanically loaded in any way. Additionally, specimens were prepared from AFC manufactured with conductive polymer electrodes [32] which were integrated into a GFRP laminate and mechanically strained according to the standard test procedure described previously in the AFC Reliability Characterization chapter.

Figure 3.2 Microscopy setup of sectioned AFC. Specimens were sectioned using a diamond cutting wheel and mounted in epoxy for grinding and polishing. AFC specimens were ground on metallographic grinding wheels to a fine grit of 2500. The AFC specimens were ground down on one side to reveal a flat surface, and at the same time retain the electrode pattern on the opposite surface as a reference for the electrode finger position. Fiber-optic light sources were used to illuminate the mounted specimen from two sides and direct the light at an angle with respect to the plane of the mounted AFC (Figure 3.2). A microscope was then used to observe the AFC surface and shadow patterns in the PZT fiber layer to reveal the location of cracks and fragments of the fibers. Table 3.2 Description of AFC in microscopy investigation. Mechanically Loaded 0º 45º 90º * Polarized X X X Not Polarized X X Not Mechanically loaded Polarized X

This configuration included AFC with traditional IDE (60º electrode contact angle) as well as AFC with conductive polymer IDE (360º electrode contact angle). 51

The microscopy findings were compiled to determine how damage evolved in the PZT fiber layer with different electrode configurations and loading conditions. Table 3.2 lists the variation between applied load direction and polarization of each AFC investigated.

3.2.1. AFC Microscopy Findings â&#x20AC;&#x201C; 0Âş Loading The AFC used in the SFBT test was manufactured with PZT fibers extending into the non-overlapping section of the IDE pattern. Therefore, the IDE pattern in this AFC defined two regions of polarization in the PZT fiber mat as shown in Figure 3.3. In the middle region of the AFC, the fibers were polarized along the fiber direction. However, in the side regions the positive and negative electrode patterns did not overlap one another. Consequentially, the fibers in these side regions remained unpolarized. However, due to the fact that the AFC was bonded to a GFRP substrate and subjected to tensile strains, both polarization regions of the AFC were subjected to the same mechanical loading conditions. The fragmented SBFT specimen is shown in Figure 3.4, where the dark horizontal region is the bottom IDE finger, which establishes the position of the IDE finger relative to the fibers. Figure 3.5 shows a close-up view of the polarized region. These two figures represent the damage patterns which existed under tensile loading of a bonded AFC as a result of different polarization states.

Figure 3.3 Schematic of an AFC with regions of polarization and non-polarization, as related to the IDE pattern, and loading direction.

Figure 3.4 Section of AFC taken with polarized and nonpolarized regions.

Figure 3.5 Close-up section of AFC taken from the middle polarized region with cracking locations identified in relation to the IDE finger position. In Figure 3.4 and 3.5 sharp transitions are seen near the edge of the IDE finger. These transitions correspond to free surfaces in the fibers, indicative of cracks running through the thickness of the fiber. In each case, cracks reside at the edge of the IDE finger, and in many instances, cracks are seen to exist on both edges of the fingers. The unpolarized region differs considerably, while a few cracks can be identified, the region is largely devoid of damage. As can be seen on the right side of Figure 3.4, there are few sharp transitions at the IDE fingers, and therefore little evidence of cracking in the unpolarized fiber section. In Figure 3.4 and 3.5, a correlation is seen between IDE finger position and the location of cracks in the polarized PZT fibers. This indirectly implies that the material properties of the fibers are dependent on the polarization of the fiber, and ultimately lead to critical stress concentrations near the 53

IDE edge during mechanical loading. In theory, cracking in the fibers might also be related to the mechanical interaction between the IDE material and the fiber surface. However, since the right side of Figure 3.4 displays no crack evidence near the IDE fingers, it can be established that the cracks in the polarized region (Figure 3.4) are not the result of mechanical load transfer issues between the electrodes and the fibers. Rather, polarization of the fibers influences the material properties of the fibers and due to the sharp transition at the IDE edge, critical stresses reside at the IDE finger edge during mechanical loading. These assumptions are further supported by the unpolarized AFC findings. The unpolarized specimen displayed cracking near to, as well as away from the IDE finger edges. Cracks were seen to run thought adjacent fibers, extending through one or many fibers perpendicular to the loading direction. The polarized SBFT specimen was seen to exhibit crack saturation at 0.50%, similar to the findings of the integrated AFC. However, the unpolarized SBFT specimen showed only sparse cracking at 0.50%, and did not display wide-spread cracking until taken to higher strains. This observation is in agreement with the polarized specimen, where a clear distinction is seen between the polarized and non-polarized regions (Figure 3.4). The most important observation being that IDE position did not dominate or coincide to the position of cracks. This strong reinforces the theory that polarization of the fibers imposes non-uniformity in the material properties along the fiber length. In principle, it is possible that cracks near the IDE edge were the result of strain mismatches in those regions which might occur during polarization and expansion of the fiber. A logical possibility is that the strain mismatch due to fiber elongation might produce cracks prior to mechanical testing, as opposed to actually manipulating the material properties of the fibers. To address this possibility, a specimen was prepared which had gone through the normal manufacturing and polarization process, but was not mechanically loaded. Investigation of this specimen (Figure 3.6) revealed that while a crack might be identified, there existed no evidence of wide-spread crack formation in the region of the IDE fingers (as compared with Figure 3.4 and 3.5). This implies that fiber crack location is indeed linked to the polarization character of the fibers and resulting stress concentrations, which evolve during tensile loading. The only reason would be related to a change in material properties at those locations, resulting from polarization of the fibers.

Figure 3.6 Micrograph of polarized AFC specimen.

3.2.2. AFC Microscopy Findings – 45º Loading Investigation of 45º SBFT specimens with polarized and unpolarized specimens mirrored the results of the 0º findings.

Figure 3.7 Micrograph of 45º SBFT AFC. The 45º SBFT specimen showed predominantly torsional failure (Figure 3.7), as evidenced by the cracks propagating through the fibers at a 45º angle, which is indicative of torsion failure in ceramic materials. Clear crack indication can be seen at the IDE finger edges. Some evidence of 45º cracks can be identified in the region of uniform fiber polarization between IDE fingers. However, the majority of cracks exist at the IDE edges. This shows the strong effect of IDE position on cracking behavior despite the variation in loading direction. The regions of changing polarization are far more susceptible to failure than the uniformly polarized sections. The non-polarized 55

45º specimen showed damage evidence near and away from the IDE fingers, again supporting the theory that polarization of the fibers influences the material property description.

3.2.3. AFC Microscopy Findings - 360º Coverage IDE The previous specimens included AFC manufactured with traditional IDE designs. The evidence shows that cracking originated at the electrode fiber tip on AFC with an electrode contact angle of 60º. Increasing the contact angle to full 360º coverage of the fiber would logically reduce stress concentrations near the electrode and possibly influence the nature of crack propagation. In addition to testing traditional AFC, it was possible to evaluate the fracture patterns of AFC with conductive polymer electrodes. AFC were prepared with IDE, which were grown through the fiber layer [32]. This resulted in IDE with a contact angle of 360º, fully covering the surface of each fiber. This design differed considerably from the conventional AFC, which exhibited a contact angle of 60º. Such a contact angel would reduce the nonuniformity in polarization character near the IDE edge around the fiber. Additionally, the conductive polymer could offer a more gradual transition in electrical field character or mechanical strains than the silver-printed IDE pattern. The poly-IDE AFC were tested according to the established technique of integration into a GFRP laminate and cycle testing to increasing strain levels as described in the AFC Reliability Characterization chapter and then prepared for microscopy. Microscopy of the fragmented poly-IDE AFC showed crack evidence perpendicular to the fiber length, situated near the transition region between IDE finger and fiber as shown in Figure 3.8.

Figure 3.8 Micrograph of fracture in poly-IDE AFC. Despite the extreme IDE modification of achieving a 360º contact angle around the fiber, the transition region between IDE finger and fiber was again the predominant feature in defining the location of cracks in the fiber layer. The accumulation of

experimental evidence shows that cracking in AFC result from stress concentrations related to fiber material properties as induced by electrical polarization.

3.3. Polarization A plausible explanation for the IDE-related crack behavior concerns the nonhomogeneous polarization character present in the PZT fibers of AFC. During polarization electrical field lines flow between opposite sign electrodes. Figure 3.9 shows a schematic of a PZT fiber section in an AFC, with the electrical field lines marked. Electrical domain orientation will be determined by these field lines, resulting in preferred polarization alignment along those lines. In AFC this results in areas that do not exhibit uniform polarization when compared with one another. Multiple studies have demonstrated the susceptibility of cracking at the PZT/electrode interface of piezoelectric actuators due to a singularity in the electrical field at that location and subsequent mismatch in strain during actuation [37, 57]. In the case of AFC, the fiber section between opposing (opposite sign) electrode fingers has a more uniform character than the section between parallel (like sign) electrodes. The section between opposing electrodes is polarized along the direction of tensile loading while the section between parallel electrodes is not. This is significant since research has shown that polarization direction will affect the tensile strength of PZT and lead to ductile failure at the grain boundaries whereas non-polarized PZT will exhibit brittle intergranular failure [19]. Due to the inhomogeneous direction of electrical field lines inherent in the AFC design tensile failure is most likely to occur between parallel IDE electrodes as opposed to the region between IDE fingers as was found in the SBFT observations. Modeling the connection between PZT fiber material properties and electrical polarization would then lend incite into how electrode design affects AFC integrity and reliability.

Figure 3.9 Schematic view of a PZT fiber showing the non-uniformity of polarization.

3.4.

AFC Model Development 57

3.4.1. Analytical Formulation Use of the AFC in a CAD/CAE environment requires characterization of the active behavior, and the integration of that knowledge into a useable design construct. Linear piezoelectric theory can be used to calculate the theoretical extension of an AFC under free and constrained actuation conditions.

Figure 3.10 Figure Uniform Field Model (UFM) [26]. The AFC analytical modeling methodology follows that of the Uniform Field Model (UFM) depicted in Figure 3.10, as applied to AFC structures by Bent [26]. The UFM considers the AFC from a Rule of Mixtures perspective, the active fiber and passive matrix and electrode materials represented as homogenous, perfectly bonded entities. From the piezoelectric stand point, polarization in the fibers is considered uniform in the longitudinal direction, extending as a vector between opposite-sign electrodes as shown in Figure 3.11. The area in between like-sign electrodes is considered in-active or dead, and does not contribute any electrical effects to actuation or sensing properties.

Figure 3.11 Side oriented cross-sectional view of a fiber in the AFC. Îľ FS = F=

d 33U Applied s

d33 AAFC E ( s) s33

U Applied

(Equation 3.1) (Equation 3.2)

The theoretical free strain in the AFC can be formulated by applying linear piezoelectric theory to the section in between opposing IDE fingers. The theoretical 58

free strain within the active region, ÎľFS is calculated according to Equation 3.1, where s is the IDE finger spacing, d33 is the piezoelectric coefficient of the fiber and UApplied is the voltage applied across the fingers. The area between like-sign IDE fingers is considered to be negligible in this formulation, based on the asymptotic character of the polarization field that exists between like-sign electrode fingers. The theoretical piezoelectric-based force developed by the AFC is calculated according to the classic block force equation for PZT wafer actuators as described in Equation 3.2 where AAFC is the active area of the AFC and SE33 is the PZT fiber compliance term. While the analytical formulation allows integration of actuation behavior into a modeling platform such as ANSYS, such an approach does not shed light onto the experimental results of the AFC Reliability Characterization chapter, where fiber cracking was shown to relate to IDE finger position. A micromechanical modeling approach is needed to provide a link between materials properties and polarization. The location of critical stresses in the model could then be coupled with the experimental findings.

3.4.2. Micromechanical Approach The experimental analysis of fiber fragmentation behavior in AFC has shown the pronounced influence of IDE finger position in relation to crack development in the PZT fiber layer. It has been established in the literature that the mechanical properties and crack propagation in PZT are tied to the electrical polarization character of the material [21]. The experimental findings strongly suggest that a gradient in material properties exist at the IDE finger locations, leading to stress concentrations and fiber failure during mechanical loading. Since a polarization gradient exists near the IDE fingers in the AFC, the material properties should also change in these regions, leading to the development of stress concentrations near the IDE edge during mechanical loading. To gain a more accurate picture of these findings, the relationship between material properties and polarization character of the PZT fiber should be considered and characterized in a modeling environment. Characterizing this behavior in a modeling environment would lend insight into new actuator and sensor designs and better characterize the reliability of AFC at the micromechanics level. Such an approach is investigated starting from the UFM and extended to a micromechanical model. As shown previously, the UFM offers a fairly accurate representation of actuation behavior, since the active and inactive polarization regions can be treated differently. The area along the fiber length is considered to be uniformly polarized, while the area sandwiched between IDE fingers is taken to have no polarization. In reality, there is a changing electrical field, and therefore, a polarization gradient exists near the IDE fingers. The UFM only considers two uniform polarization states, and was not designed to represent the relationship between polarization and material properties. Therefore, the UFM is insufficient to represent the affect of polarization on the material properties in the region of the IDE fingers in the PZT fiber, and is therefore

not suitable for evaluating new and complex electrode designs. Although micromechanical modeling of the AFC has been previously investigated [58] a link between polarization character and the material property description in the PZT fibers of the AFC was not considered. A micromechanical modeling approach to AFC must directly or indirectly consider the modeling of the PZT fiber and the effect of electrical domain switching on the global material properties of the fiber. Previous investigations into the micromechanical modeling of ferroelectric materials have yielded a focused body of knowledge characterizing domain switching in ferroelectric ceramics. Considering the problem from a crystallography-based viewpoint, at the micron level piezoelectric materials are composed of crystallographic grains, which can be represented as cubic volumes of a fixed size, with a random domain orientation. A grain is considered as a single crystal, consisting of a uniform alignment of electrical dipoles in the individual crystals. In this way, switching is considered at the grain level, and not via evaluation of each specific dipole orientation. External mechanical and electrical loads can be defined according to a global coordinate system. Each grain domain includes a local coordinate system, which is used to define the polarization direction in response to ferroelectric and ferroelastic influences. Domain orientation is then realized as a rotation of the local coordinate system for each grain. A volume average of the grains then defines the macroscopic polarization state of the ceramic. A micromechanics modeling approach generally starts by considering a ferroelectric ceramic as unpolarized with a random orientation assigned to each grain. The influence of mechanical stress and electrical loading must be tied to the domain orientation at the crystallographic, grain, and finally the global material level. A critical point in the modeling process is to define a dipole switching criterion at the crystallographic level. Once the polarization of each grain can be defined, the total number of grains can then be summed together to calculate a uniform polarization state at the global material level [59]. Domain switching is based on the energy required to rotate a dipole orientation, which can occur due to an applied mechanical stress parallel to the polarization direction or an applied electric field in a direction different from the polarization direction [60]. Ferroelastic switching occurs when the coercive electric field is exceeded, enabling a displacement of the central ion of the crystal lattice to a new position, which is in alignment with the applied electrical field direction. Ferroelastic switching requires that a stress deform the crystal and force it towards a cubic geometry. This displaces the dipole equilibrium in the crystal and the central ion is displaced to a new location (90ยบ from the original position), aligned perpendicular to the applied stress. Defining a set of criteria for domain switching is then needed as a basis for defining the polarization state of a piezoelectric due to mechanical and electrical loading conditions. An energy criterion, based on the work required to switch a dipole can be used to define the driving force required to switch the polarization direction of the material [59]. The electrical (Welectrical) and mechanical (Wmechanical) work required to switch a domain are described in Equation 3.3 and 3.4. E is the applied electrical field,

sigma σ the applied stress. The dε and dD are the respective change in strain electrical displacement. Welectric = ∫ Ei dDi

Wmechancial = ∫ σ ij deij

(Equation 3.3) (Equation 3.4)

The separate work terms can be summed together for the total work (Equation 3.5), where Ec is the critical electrical field and Dc is the magnitude of spontaneous polarization. Wtotal = σ ij Δε ij Ei ΔDi = 2 E c D c

W − ΔU 1 ≥ 2 D0 E 0

(Equation 3.5) (Equation 3.6)

The threshold for switching is then defined in Equation 3.6, where dU is the change in interaction energy. D0 is the magnitude of spontaneous polarization. The switching criteria can be evaluated for each grain, and the local domain coordinate system rotated in reference to the global system, dependent on the new polarization direction of the grain. This modeling approach has been evaluated and validated against experimental results for different ferroelectric materials including 8/65/35 PLZT [61] and PMN-32%PT [59]. While the previous approach has been shown to accurately predict the polarization state of ferroelectric materials, it does not lend incite into material property change due to electrical polarization. An alternative approach to modeling the exact polarization state of a ferroelectric material is to forego the micromechanical description of individual piezoelectric domains, and instead to focus on the relationship between applied electrical field strength and direction so that the polarization direction can be represented as a change in material properties [62]. Such a technique, which relates material properties to field strength, offers a view of the material property description which reflects the experimental findings, and which can serve as a tool for engineers interested in the impact of different electrode configurations on actuator and sensor reliability. The goal of the current strategy is to model the change in material properties of the PZT fibers in the AFC near the IDE fingers, and to expose stress concentration areas in the electrode design, which can be correlated to the experimental findings. Basic ferroelastic theory and experimental evidence show that dipole moments and polarization direction are determined by the electrical field lines. From the strength of materials standpoint, a modeling methodology at the micromechanical level could primarily consider the link between electrical field direction and polarization direction, and describe this relationship as related to the material properties of the fiber near the IDE fingers in an AFC. The question of how to define polarization direction in response to field direction then becomes a concern, which is not directly addressed in computer aided engineering and analysis programs such as ANSYS or the UFM.

Figure 3.12 Conventional material coordinate system definition. When modeling a piezoelectric material in a finite element analysis program such as ANSYS, the material properties and polarization direction are considered to lie coincident with the local coordinate system of the elements. This concept is illustrated in Figure 3.12 where a local coordinate system governs the material property description of the entire fiber, while electrical field lines retain the characteristic asymptotic character. In this conventional description, no link exists between polarization direction and the mechanical properties of the fiber. Therefore, when considering the AFC, this does not allow for an accurate representation of the material properties in the vicinity of the IDE fingers. Since the asymptotic character of the field line is similar to that of a cylindrical coordinate system, one could be used in the region between the IDE fingers to offer a better material property definition. However, for different IDE designs or material configurations, the cylindrical coordinate system would not necessarily accurately model the field line orientations. For material properties to be aligned along the field lines, a routine can be used, which models polarization of the PZT and applies material property definitions along the field lines. This technique is illustrated in Figure 3.13, where the material coordinate systems describing the mechanical properties of the fiber are oriented along the electrical field lines.

Figure 3.13 Rotated material coordinate system (RCS) definition. 62

As the field line direction changes, this then implies that various sets of material properties, corresponding to the magnitude of polarization would be required. The number of graduations in materials could be tied to the grain structure to better approximate the case in reality. However, the goal in defining the number of material graduations is to allow a sufficient number of materials such that a changing material field can be established. Defining models with increasing numbers of materials would establish a convergence of material levels, beyond which no change in material description would be seen. The assignment of material directions is described in Figure 3.14 where the electric field description of the model enables modification of material direction in accordance with the electric field strength and direction. A subsequent electric field application is compared with the first, and if no change is found, the stress analysis is conducted. This methodology was programmed into an ANSYS routine [63]. A number of material property definitions could be defined, ranging from partial to full polarization. However, since a general trend is desired, the difference between modeling a larger or small number of states would not result in a large difference in the material property definition. The mechanical properties are then seen to vary in accordance with the curvature of the electrical field lines, exhibiting a transition from asymptotic to uniform polarization regions. A voltage threshold of 1500 V/mm is used as the switching criteria. This coincides with the poling procedure used in manufacturing AFCs as described in section 2.2.2.1.

Figure 3.14 Analysis process for polarization of fiber material [62]. The extent of dipole reorientation and the resulting remnant dipole direction after polarization would require integration of switching criteria into the model, and would be extremely complex to fully validate with an AFC configuration. Spontaneous 63

polarization would naturally occur due to external forces, which is not captured in this modeling methodology. However, from a reliability standpoint, spontaneous switching would not necessarily need to be captured in order to define stress concentrations in the fiber. Spontaneous polarization due to mechanical loading in the AFC would not occur unless the fibers were compressed. Furthermore, due to the polarization gradient near the IDE fiber, it is unlikely that any significant domain switching due to mechanical loading would adversely affect the stress concentration point near the IDE edge. For the purpose of characterizing AFC reliability, it is more important to define the general trend in material properties as defined by polarization, than to model individual electrical domains or the polarization character of individual PZT grains which would also be highly costly from a computational view. In the current methodology, electric field strength and direction are used as a criterion for domain switching. Since the change in material properties is sought, the electric field strength can be used to define the material property gradient. The domains are represented as the individual finite elements. Initially, homogeneous material properties are assigned, which is analogous to defining an unpolarized material area or volume. Thus, the current micromechanical modeling strategy considers how material properties would be ideally directed long electric field lines, but can not be considered a definitive definition of exactly how material property variation exits within the fiber with respect to a specific definition of polarization. Inclusion of polarization state can be included in ANSYS via a programmed routine. Critical strains as predicted by the model can then be compared with the previously presented microcopy evaluation.

3.4.3. ANSYS Model The model in ANSYS included the main components of the AFC, and was realized as a single fiber section with positive and negative electrodes. Electrical potential boundary conditions can then be applied to polarize the fiber section. A stress analysis could then be conducted to assess critical stresses in the design. A two-dimensional model was created with Plane 223 elements, which allowed for mechanical and piezoelectric capabilities. The electrodes, insulating layer, and fiber material were modeled as separate components.

Figure 3.15 Comparison of different material property

descriptions.

Figure 3.16 Material property gradient resulting from polarization.

3.4.4. ANSYS Model Results Figure 3.15 presents the three different material property descriptions for a fiber in the AFC. The UFM shows two uniform regions while the 2 and 5 polarization regions display nonlinear material profiles, with the 5 step region offering a larger transition region (Figure 3.16). The first principle strains developed within the fiber section are presented in Figure 3.17. Stress concentrations exist at the material property boundaries, but are most pronounced at the IDE finger edge. This is also the area of greatest change in the electrical field orientation. Here there exists a large stress concentration near the fiber surface, situated very close to the edge of the IDE finger. The modeling results suggest that stress concentrations and therefore, crack nucleation centers are likely to originate at the tip to the IDE finger region. This is in direct agreement with the microscopy investigation results, which showed quite prominently that fiber failure is localized at the IDE edge.

Figure 3.17 First Principle strains in AFC polarized fiber model.

3.5. Electrode Design and AFC Integrity Micromechanical modeling of the AFC has shown that critical strains in the AFC correspond to the experimental findings concerning crack growth in PZT fibers. The ANSYS modeling methodology can then be used to investigate how modifying the electrode design might improve AFC reliability by reducing stress concentrations due to material property inconsistencies resulting from electrical polarization. Reducing the stress concentration point may be possible by embedding the electrode finger inside the fiber, as opposed to positioning it on top of the fiber. Two methods are considered, embedded and fully integrated. Modifying the current electrode geometry to an etched or a fully integrated design would improve the uniformity of the electrical field lines, and hence produce a more uniform material property description. However, this would also introduce inconsistencies in the fiber. Two possibilities are presented, an etched (Figure 3.18) and a fully integrated (Figure 3.19) electrode design.

Figure 3.18 Etched electrode idealization. In the etched case, a design is considered where surface micromachining techniques could be used to etch away material from the fiber, and then deposit a suitable metallic layer for the electrode. This is depicted in Figure 3.18, where isotropic etching is assumed, and a curved depression is created in the fiber surface. Conceptually the electrical field lines would retain a more gradual transition, and therefore the material property description would also be more gradual. This would modify the location or intensity of the stress concentration point.

Figure 3.19 Integrated through-thickness electrode. A second case is the fully embedded design, where an electrode would be embedded through the diameter of the fiber. This method would eliminate the asymptotic material region, and offer a very uniform field, essentially creating a stack design. The fully embedded design poses the most drawbacks. From an application standpoint, if the embedded electrode were composed of a softer material than the PZT fiber, the finger would provide an accommodation of actuation force during electrical activation. This would reduce the effectiveness of the AFC as an actuator. Alternatively, a thin metallic layer would result in a design similar to a classical stack actuator. This would dramatically reduce the tensile strength of the device, which would be limited by the bonding strength of the fiber-electrode interface. Additionally, manufacturing of the fully embedded design would be difficult to realize on the laboratory or industrial scale. The etched design was modeled and evaluated, to ascertain if a more favorable material property description could be obtained as compared with the traditional IDE 67

design. An etch depth and profile was modeled based on the assumption that isotropic etching would produce a curved profile, similar to an ellipse. A deep etch depth was assigned, for the purpose of creating a geometry between that obtained for a traditional and fully integrated electrode. Five material property gradients were assigned, as depicted in Figure 3.20. The etched profile produced a material property description slightly more uniform than the traditional design, but not significantly different.

Figure 3.20 Material property distribution of etched IDE.

Figure 3.21 Critical stress distribution of etched IDE. Evaluation of the critical strains showed that the concentration point again is located at the transition point between the fiber and electrode materials, which is also the point of the shortest distance between electrode fingers (Figure 3.21). This would offer little advantage as compared with the sandwich IDE configuration. The first principle stress showed a reduction due to the etched configuration. The conventional IDE developed

a maximum stress of 27.11 MPa, while the etched IDE maximum stress was only 17.17 MPa. Although the critical stresses are better distributed over the etched IDE region, failure would still occur near the IDE edge. The etched design also requires that material be removed from the fiber. This would reduce the cross-sectional area of the fiber, and therefore reduce the maximum force which the fiber could carry. The manufacturing complexity of the AFC would also increase, as would the cost of production. These cases represent the extremes of IDE integration but would not result in improving the reliability of the AFC design. The etched IDE would result in a reduction in the cross-sectional area of the fiber, and logically lead to crack nucleation. The thickness of the fiber would be reduced, allowing failure at lower forces. Both solutions might offer some incremental advantage, but include too many drawbacks, including increased manufacturing complexity.

3.6. Conclusions Within the framework of AFC reliability, it is apparent that the current IDE design offers the best combination of performance and manufacturability. The limitation being, that during polarization non-uniformity develops in the material property description of the fibers near the IDE fingers. This has been illustrated via modeling as well as observed experimentally. While this phenomenon can be characterized and understood, improvements to AFC reliability can not be attained by manipulation of the IDE geometry. The current printed IDE design offers the best compromise between performance and manufacturing complexity, and offers the most reliable design solution. If the current AFC design can be considered optimal, then the means of improving AFC integrity and reliability would logically need to exist in other areas. As a component in a smart material system, the AFC design only encompasses part of the overall system. Exploring the interaction between the active element and supporting passive structure can be assessed and optimized to improve AFC reliability. Packaging strategies will be explored, which seek to modify the force and strain transferred to the AFC and thereby influence the fragmentation onset strain to extend the usable strain range of the AFC. Aside from the performance characterization of AFC, the investigation of AFC as a component in a smart material system requires an assessment of the AFC as well as studying the impact of AFC integration on the integrity of the structural composite which the AFC is integrated into.

4. Laminate Integrity Analysis Current smart materials design concepts generally only consider the question of actuation ability, maximizing actuator performance without presenting a clear picture of how structural reliability is impacted. In some cases the active layer can be a continuous component of the structural design. Generally the active component is taken to be a material inclusion in the passive structural material. If an active rotor can be completely covered in an actuator layer, then it becomes an integral layer of the laminate design. However, the desire in smart materials design is generally to optimize actuator and sensor placement on a surface [64] so that the actuators are placed for optimal transfer of force to the structure or record the maximum signal from the environment. This design technique reduces material cost and any added weight penalties associated with the active material. This also reduces the cost associated with the active materials. Ideally a smart material should serve as a structural and active component, so that the entire structure can be controlled as desired. Currently smart materials are composed of two distinctly dissimilar materials, the structural and active components. Creating material systems of two dissimilar materials invariably will lead to problems with the distribution of one material with respect to the other, which will lead to problems with structural reliability. The integration of smart devices is expected to have an effect on the mechanical performance and integrity of a structure. In particular, degradation of the ultimate strength properties is a great concern. If the integration of an active element into a structure severely degrades the ability of the structure to satisfy the original design requirement, then the overall design becomes unusable. For some applications, the integration of smart elements may degrade the mechanical performance of a host structure to unsafe levels. This is of particular concern to critical weight applications such as rotor blades or airplane wings where efficient structural designs with high strength to weight ratios are required. It is less critical for structures with high safety factors built into their design such as bridges and similar civil structures or applications which might see low mechanical loads such as acoustic actuators or vibration sensors. The relationship between smart materials integration and the structural integrity of laminate materials has been addressed in the literature in various forms. Past studies on smart materials integration have focused on the integration of PZT wafer devices in composite laminates with the active element integrated in the 0º or 90º plies with respect to the loading direction. In a balanced [0,90] cross ply laminate composed of unidirectional (UD) material, device integration in the 90º plies will naturally reduce the impact of integration on laminate integrity as compared with integrating the devices into the load bearing 0º plies. The stiffness of the 90º plies is much less than the 0º plies and will therefore carry less load in response to an applied global strain. Integrating smart devices into the low load bearing plies of a laminate is ideal since this method will have the smallest impact on the mechanical integrity of the laminate. However, in practical applications it may prove difficult to accomplish since each ply 70

would be expected will carry load depending on changing loading conditions during its service life. Crawley and De Luis [39] investigated the integration of PZT wafer actuators into woven composites using analytical and experimental methods and proposed design criteria for PZT actuators such as a high modulus of elasticity to allow stress transfer with the surrounding laminate. In static tension, the integration of PZT devices via cutout insertion was found to have little effect on the elastic modulus, but a 20% decrease in the ultimate tensile strength of the integrated GFRP laminates was observed. Mall and Coleman [40] investigated the affect of PZT wafer integration on CFRP laminates under monotonic tensile and fatigue conditions and reported that with quasi-isotropic CFRP lay ups the integration technique does not greatly affect laminate strength, reporting a difference of only 4% between the mechanical properties of composite laminates without integrated devices and those that employed the simple and cutout insertion integration techniques with PZT elements integrated in or adjacent to the 90ยบ low load bearing plies. Paget and Levin [41] performed similar work with PZT wafers inserted in the center of CFRP laminates and only saw a 3% difference in the ultimate tensile properties between laminates with integrated PZT wafers and the reference laminates. Interlacing the laminate plies and distributing the integration region throughout the laminate thickness has been found to influence laminate or device reliability. In a first study [42] an optimal interlacing design was proposed using finite element techniques. A second study [43] tested the design by tensile testing UD GFRP laminates with integrated glass slides to simulate the presence of active elements. Vizzini reported that increasing the taper length between the laminate and integration region reduces load transfer but also increases the resin rich zone at the integration region. Reducing the taper length improves failure at the device interface but also degrades laminate integrity. In certain applications it may be possible to maximize the ratio of passive to active plies in a design, and thereby reduce the effect of device integration on the mechanical integrity of the laminate. Adding many layers to a laminate and testing the mechanical integrity of the system may show little affect if the number of plies is very large as compared with the number of integrated devices in the structure [44, 45]. An understanding of the failure of smart laminates is needed to design structures that fulfill the functional as well as the structural mechanical design requirements of a given application. In particular, tools and methods for determining the effect of smart materials integration on laminate integrity can be used to better characterize the integrity of smart material systems and to assess the trade off between adding actuation and sensing capabilities while influencing the mechanical integrity of the structural design. Many smart material applications require service life of sensors and actuators in fatigue and cyclic environments. Separately both PZT and composite materials have shown robustness over millions of fatigue cycles. A critical point with integrating the materials together into a system is the ability to characterize damage evolution due to the mismatch of material properties and how this impacts the initiation of damage in the laminate system. Characterizing the reliability of the

material system at the quasi-static loading state is pertinent to the fatigue loading condition. In the context of this study reliability entails a resistance to mechanical strength degradation. Numerically characterizing laminate integrity can be accomplished by employing failure criterion predictions for different laminate configurations with integrated smart elements. A number of failure theories are available for composite materials, and are generally characterized as limit, interactive, or partially interactive [65]. The World-Wide Failure Exercise [66, 67] was conducted to assess the effectiveness of various failure criterions for predicting the behavior of composite laminates. The Tsai-Wu quadratic failure criterion [68] was shown to perform well in predicting laminate response and ultimate failure [67]. The First Ply Failure (FPF) prediction is also available, and has been found to agree with experimental results. Limitations were found for post FPF and Last Ply Failure (LPF) values [69], however it has been found to be a robust tool for engineering with composite materials. Tsai-Wu is limited in that differentiation is not established for matrix versus fiber failure in a laminate. This distinction is covered in a partially interactive criterion such as Puck [70]. However, the easy implementation and performance of Tsai-Wu makes it an ideal choice for evaluating composite smart laminate reliability. The interactive Tsai-Wu criterion is relatively easy to integrate into a finite element program such as ANSYS. The current work focuses on the mechanical evaluation of smart device integration in CFRP and GFRP laminates via experimental methods, investigating the case where the ratio of active to passive plies is low and integration occurs in load bearing as well as non-load bearing plies. Focusing on device integration in the load bearing plies, the critical impact of device integration on laminate integrity can be investigated. Tensile testing of integrated laminates was used to establish the performance of different material and laminate designs. This included the use of Active Fiber Composites (AFC) as well as Dummy AFC (DAFC) elements. DAFC were developed as a physical model of the AFC, mimicking the longitudinal material stiffness of the AFC, and provided a cost effective method of evaluating different integration methods. Numerically, ANSYS modeling was employed to model the smart structures with failure criterion implemented to investigate how plies fail in the laminate with respect to different integration techniques. Modeling in ANSYS also enabled an investigation into the smart functionality of the laminate design by including the actuation and sensing ability of the laminates with PZT actuators/sensors.

4.1.

Materials

Glass fiber epoxy laminates were produced with a vacuum bagging technique using Isopreg HR 320P-40 plain weave pre-impregnated (pre-preg) composite plies supplied by Isolvolta (www.isovolta.com). CFRP laminates were manufactured using ELITREX pre-impregnated (pre-preg) CFRP composite plies (EHKF 420-UD24K-40) supplied by Stesalit (www.stesalit.com). AFC were produced according to the process detailed in the AFC Characterization work.

4.1.1. Dummy AFC (DAFC) manufacture In order to optimize testing procedures, reduce manufacturing time and predict the effect AFC integration would have on the integrity of the laminates, DAFC were produced as shown in Figure 4.1. In order to simulate the mechanical properties of the AFC, DAFC were manufactured using Kapton® film and a core of UD glass pre-preg plies. The UD plies were arranged with a +/- 20º orientation, to effectively match the longitudinal modulus of the AFC. Kapton® film was used in the DAFC to simulate the bonding characteristics with the laminate structure. To manufacture, a sheet of Kapton® foil was placed on an aluminum tool plate followed by the UD GFRP plies and a top sheet of Kapton®. The tool plate was placed in a press with a pressure of 7 bar and the temperature was raised to 120ºC and maintained for 90 minutes in accordance with the recommended curing cycle. Standard tensile test specimens were cut from the DAFC sheet and tensile tested to determine the tensile mechanical properties. A comparison between the AFC and DAFC tensile properties is presented in Table 4.1, where it is seen that the E modulus of the AFC and DAFC differ by only a few percent. For integration testing DAFC were cut from the GFRP- Kapton® plates with dimensions of 4 x 4 cm and a thickness of 0.42 mm.

Figure 4.1 DAFC showing Kapton surface and GFRP core. Table 4.1 Comparison of AFC and DAFC mechanical properties. AFC

DAFC

σMAX (MPa)

E (GPa)

εMAX (%)

σMAX (MPa)

E (GPa)

εMAX(%)

Avg.

0.25

537

31.38

1.86

StDev

3.49

1.36

0.04

1.10

0.11

%Dev

8.15

4.71

16.82

9.71

3.52

6.05

4.1.2. Integration of DAFC and AFC in GFRP laminates

The AFC and DAFC elements were integrated into the laminates using different integration techniques and placement positions with respect to the laminate thickness. The element position through the laminate thickness varied with respect to the midline of the laminate. Three positions were defined: center, off center, and far off center (Figure 4.2). This relative position was kept constant in the different materials systems even though the plies were of different thicknesses. DAFC were integrated into the woven GFRP laminates using the insertion (Figure 4.2a) and cutout (Figure 4.2b) methods as well as into GFRP laminates in the center and far off-center positions using the cutout method.

a Figure 4.2 Position of the AFC/DAFC for the insertion (a) and cutout insertion (b) integration procedure in the GFRP laminates.

4.1.3. Integration of DAFC/AFC in CFRP laminates The insertion and interlacing integration techniques were used to manufacture CFRP laminates with DAFC (DAFC-CFRP). DAFC were integrated at three different positions; center, off-center and far off-center, in relation to the mid-plane of the laminate, as displayed in Figure 4.3. In Figure 4.3 the light layers correspond to 0° plies while the dark 90° plies, the inclusion position is easily identified in between the continuous layers. The resin rich transition areas are indicated by the triangular space between the ply and inclusion regions. The CFRP laminates with AFC (AFC-CFRP) were prepared with the far off-center position using insertion (Figure 4.3c) and interlacing (Figure 4.3f) methods. Each laminate set consisted of five specimens. Integration of AFC required attention to electrical insulation of the AFC from the conductive CFRP laminate. The AFC and attached wires were laminated with Kapton® tape to ensure that no conduction paths could exist between the AFC and the conductive carbon fibers of the laminates. The wires extending from the AFC were protected by silicon tubing at the end of the specimen, and a small recess in the frame ensured wires did not move during processing. In addition, wax was applied to the wires at the recess point to protect them from encasement in resin flow during processing.

(a) Center insertion

(d) Center interlaced

(b) Off-center insertion

(e) Off-center interlaced 1

(f) Off-center interlaced 2 (c) Far off-center insertion Figure 4.3 Integration of elements in cross-ply CFRP laminates.

4.2.

Laminate Manufacture

Orthotropic plain weave GFRP laminates were manufactured according to the process described in the Reliability Characterization chapter. CFRP cross-ply laminates composed of 12 plies of UD pre-preg were manufactured with a [0,90]3S configuration. CFRP plies were cut to exact dimensions using a computer controlled Z체nd M-1600 cutter. This allowed precise dimensioning of the cut out area for integration of DAFC and AFC. Ply arrangement was performed by hand on an aluminum tooling plate and an aluminum frame was used to contain the plies. The CFRP laminates were processed in an autoclave under 4 bar of pressure and cured at 120째C for 90 minutes. Following processing, the reference (containing no AFC/DAFC) and integrated (containing AFC/DAFC) laminate plates were prepared in accordance with ASTM Standard D 3039/D 3039M-00. The specimen geometry and dimensions is shown in Figure 4.4. Aerolite epoxy glue was used to secure tab material to the ends of each specimen.

Figure 4.4 Schematic view of the laminate test specimens.

4.2.1. Tensile Testing 75

Tensile testing was performed in accordance with ASTM Standard D 3039/D 3039M00 using an Instron 1251 test machine with a 200 kN load cell and an extensiometer (50 mm gage length) was used to determine strain during the test. A loading rate of 2 mm/min was used. Specific specimen configurations are summarized in Table 4.2. Table 4.2 Summary of integration configurations tested. All laminates had specimen sets with integrated DAFC, those that also were tested with AFC are indicated. Material Integration Device Probes DAFC 5 Center AFC 4 Off Center DAFC 5 Cutout DAFC 5 Far Off Center GFRP AFC 4 Symmetric DAFC 5 Center DAFC 5 Insertion Off Center DAFC 5 Far Off Center DAFC 5 Center DAFC 5 Off Center DAFC 5 Insertion DAFC 5 Far Off Center CFRP AFC 5 Center DAFC 5 Interlaced Off Center DAFC 5 Far Off Center DAFC 5

4.3. Finite Element Model The laminate geometry and different integration methods were modeled in ANSYS to assess their mechanical reliability and actuation capabilities. The layered SHELL99 element was employed, which allowed each ply angle to be easily defined. The influence of ply compaction on the ply mechanical properties was taken into account. To assess mechanical reliability, Tsai-Wu Strength Index (TWSI) values (Last Ply Failure) were computed by imposing the failure loads recorded experimentally. Experimental results and TWSI values were compared for validation of the model. FPF loads (i.e. TWSI equal 1) were then determined for the different laminate configurations. To assess the sensing and actuation capability of the different configurations, a custom shell element with piezoelectric capabilities [71] was used to model the AFC modules. The laminate model was cantilevered at one end and actuated to produce tip deflection at the free end of the laminate. Conversely, a deflection imposed at the free end yielded a sensing voltage at the AFC module. With this approach the strain deformation energy, respectively the sensing voltage generated by the integrated AFC module could be compared with the mechanical reliability study and an evaluation of damage in the integration region.

4.4.

Laminate Tensile Results and Discussion

The failure load and Young's modulus results from the GFRP laminates are graphically represented in Figure 4.5 and 4.6 while the CFRP laminate results are shown in Figure 4.7 and 4.8 and are included in the Appendix (Table A.1 and A.2). As can be seen in Figure 4.5 the maximum force of the GFRP laminates decreased with the integration of DAFC in the laminate (center placement) and decreased further as the DAFC were placed far from the laminate mid-line. Center placement decreased the maximum force value by about 6-9% while the off center and far off center placements showed a drop in strength of 13-18% and 24-25% respectively. Although a quantifiable difference is seen due to integration, considering the standard deviation the type of integration technique (i.e. cutout vs. insertion) appeared to have an equivalent effect on the strength of the GFRP laminates. The higher overall maximum force exhibited by the cutout configuration is due to the additional fabric ply as opposed to the insertion configuration. In contrast to the GFRP findings, the influence of DAFC integration on the strength of the CFRP laminates Figure 4.7 was statistically insignificant for center and off center placement, since the decrease in maximum force was on the same order of magnitude as the standard deviation. However, the far off center placement showed a reduction in strength of 6-7%, regardless of the integration method. The maximum force values for the CFRP laminates with AFC showed a difference from between 5.5% to 11% as compared with the reference laminate. GFRP Laminate Strength

Maximum Force (kN)

70 60

Reference Center Off-Center Far Off-Center AFC Center AFC Off-Center

50 40 30 20 10 0 Simple Insertion

Cutout Insertion

Figure 4.5 Ultimate tensile strength results of GFRP laminates.

GFRP Laminate Modulus

E Modulus (GPa)

30 25 Reference Center Off-Center Far Off-Center AFC Center AFC Far Off-Center

20 15 10 5 0 Simple Insertion

Cutout Insertion

Figure 4.6 Young’s modulus of the GFRP laminates. The integration of DAFC in GFRP laminates showed an increase in Young's modulus in the far off center position Figure 4.6. Conversely, AFC integration lead to a decrease in modulus. The difference between the AFC and the DAFC modulus findings can be attributed to the fragmentation which would have occurred in the PZT fiber layer of the AFC at these strain levels. The CFRP data did not show statistically significant changes in modulus due to AFC/DAFC integration (Figure 4.8). This can be attributed to the stiffness mismatch between the 0° CFRP UD plies and the more compliant AFC/DAFC inclusion. The difference in the GFRP and CFRP behavior can be attributed to the ply arrangement of the two laminates as well as the difference in stiffness between host laminate and inclusion. Integration directly affected the load bearing plies of the woven GFRP laminate, either directly via cutting in the cutout procedure or indirectly by creating angles in the plies near the edge of the integrated element. However, in the cross-ply CFRP laminates, the majority of the load was carried by the 0° plies, while only the 90° plies were affected during integration. Since the 90° plies were largely low load-bearing in this tensile configuration the effect on the strength of the laminate is small.

CFRP Laminate Strength 160 Maximum Force (Kn)

140 Reference Center Off-Center Far Off-Center AFC Far Off Center

120 100 80 60 40 20 0 Inserted

Interlaced

Figure 4.7 Ultimate tensile strength of the CFRP laminates. CFRP Laminate Modulus

Young's Modulus (GPa)

80 70 Reference

Center

Off-Center

Far Off-Center

AFC Far Off Center

20 10 0 Inserted

Interlaced

Figure 4.8 Figure 3.4 Youngâ&#x20AC;&#x2122;s modulus of the CFRP laminates. The strength reduction observed in the current study is similar to the values reported in the literature. Mall [49] reported no influence of PZT wafer integration on the ultimate strength of CFRP laminates, while Paget and Levin [41] reported an ultimate strength degradation of 4.5% for [04/904/04/904/02(PZT)]S CFRP laminates. In the present study CFRP laminate strength decreases between 1% and 3% by integration in the center and off center positions using a [0,90]3S configuration. For the case of far off center symmetric DAFC integrated in GFRP laminates using the cutout method, a decrease of 32% in strength is observed. Results are in line with the findings from 79

Crawley and De Luis [39], who observed a reduction of 20% in the ultimate tensile strength of [0/90/0/90/0]S GFRP laminates symmetrically integrated with PZT wafers using the cutout method. The difference between the current results and those of Mall [49] could be explained considering the greater number of load bearing (0Â°) plies relative to the number of low load-bearing (90Â°) plies in the laminate as compared with the current study. The current findings combined with the literature show that the device placement through the laminate thickness is a critical design consideration concerning mechanical integrity of smart laminates. Representative force/strain curves of the different GFRP configurations are displayed in Figure 4.9 through 4.12 for each integration technique and laminate material. The center placement GFRP specimens displayed a smooth loading profile while the off center specimens exhibited an extended plateau after reaching their maximum force values with both insertion and cutout integration techniques. This "yield point" corresponded with visual observations during the test of the onset of internal damage and delamination in the tensile specimens corresponding to extensive failure in the KaptonÂŽ layer of the DAFC, followed by catastrophic failure of the laminate. As can be seen from the two curves depicted in Figure 4.9 and 4.10, the force for the far off center configuration decreases from the initial yield point, reaches an intermittent minimum, and then rises to a second force level where global laminate failure occurs. This behavior differs considerably from the center placement fracture behavior with DAFC inserted in the symmetric axis of the laminates where no such yield point was observed. The CFRP laminates as shown in Figure 4.11 and Figure 4.12 did not exhibit this type of loading behavior and no sign of failure in the integration region was observed prior to failure of the laminates. GFRP Insertion Laminate Loading Response 60

Force (kN)

50 40 30

Reference Center

Off Center

Far Off Center

0 0

Strain (%)

Figure 4.9 Load response of insertion integration GFRP laminates.

GFRP Cut Out Laminate Loading Response 70 60

Force (kN)

50 40 Reference Center Off Center Far Off Center Far Off Center Symmetric AFC Center AFC Far Off Center

30 20 10 0 0

Strain (%)

Figure 4.10 Load response of cutout integration GFRP laminates. The AFC and the DAFC laminates show comparable failure load values for both center and far off center integration with a yield point is evident before final failure. A difference is seen in the Young's modulus data however, where the initial moduli are similar at low strain (below 0.03%), above this limit the modulus of the samples with integrated AFC was observed to be lower than that of the samples with DAFC. Previously published studies [28, 72] reported a tensile strain for AFC tensile specimen in the order of 0.25%. So, from a materials standpoint, beyond 0.25% damage should have ensued in the PZT fiber layer. If cracks developed in the AFC the stiffness of the structure would logically decrease. As seen in Figure 4.10 the initial slope appears similar, but changes beyond 0.50%, possibly indicating damage in the AFC.

CFRP Inserted Laminate Loading Response 160 140

Force (kN)

120 100 Reference

Center

Off Center

Far Off Center AFC Far OFF Center

20 0 0

0.5

1 Strain (%)

1.5

Figure 4.11 Loading curves of CFRP inserted laminates.

CFRP Interlaced Laminate Loading Response 160 140 Force (kN)

120 100 80

Reference Center Off Center Far Off Center AFC Far Off Center

60 40 20 0 0

0.5

1 Strain (%)

1.5

Figure 4.12 Loading curves of CFRP interlaced laminates.

4.5.

Laminate Fracture Behavior

4.5.1. GFRP Laminate Fracture Behavior 82

Failure in the GFRP laminates with integrated DAFC occurred via damage in the Kapton® layer and subsequent transverse failure in the laminate. This behavior increased as the integration position moved from the laminate mid-line towards the laminate surface. In the off-center configuration delamination initiated at the edge of the integration region and spread over the surface of the DAFC. Ultimate failure generally occurred in the glass plies directly at a horizontal edge of the DAFC and post-failure inspection of the samples showed that failure occurred in the Kapton® layer. Part of the Kapton® remained adhered to the DAFC core and part remained with the adjacent glass ply. Prior to full laminate failure the DAFC essentially delaminated within the glass plies without complete failure to the specimen. Characteristic DAFC failure specimens are shown in Figure 4.13 where failure occurred near or on the horizontal edge (normal to the loading direction) of the DAFC.

Figure 4.13 Far off-center symmetric Figure 4.14 Center placement cutout insertion specimens with symmetric cutout insertion specimens DAFC. with AFC. Characteristic AFC laminate specimens, displayed in Figure 4.14 exhibited fracture behavior slightly different from that of the DAFC. The samples with center AFC contained fractures near the horizontal edge, as was the case with the DAFC. However, in some specimens, an additional vertically oriented fracture above the AFC appeared. Off-center DAFC failed almost universally near the DAFC edge. However, with the off-center AFC the transverse ultimate failures resided past the AFC edge, more inside the planar area of the integration region. In these cases the fracture lines passed through the Kapton® layer and more towards the AFC center. This Kapton® failure did not coincide with the DAFC failure. Rather, the Kapton® delaminated from the PZT fiber layer and remained attached to the laminate until failure. All of the AFC specimens had delamination in the glass plies, resulting from failure of the AFC. This delamination occurred after ultimate failure and is indicated as the white regions on the specimens, contrasting with the off-white color of the laminates near the tabbed ends.

Differences were seen in the fracture behavior laminates with DAFC integrated center and off-centered. Although the fracture locations were similar (transverse failure along the horizontal edge) the damage in the Kapton® layer suggests two different types of failure. Figure 4.15 displays images of a center and off-center DAFC post failure. Even at the macroscopic scale the two specimens show different failure characteristics. The center DAFC (Figure 4.15a) has a largely uniform surface consisting of a dimpled pattern, except at the horizontal edge where irregular ripping or shearing is seen. By comparison the off-center DAFC (Figure 4.15b) has a nonuniform character with large areas where the Kapton® has ripped away from the DAFC core material.

b Figure 4.15 Center (a) and off-center (b) placement DAFC post failure. The discontinuous failure nature of the off-center DAFC Kapton® layer suggests a non-uniform stress state. The non-uniform failure can be characterized as ripping, suggesting failure due to shearing forces. It is also pertinent to note that the Kapton® near the side edges of the DAFC (parallel to the tensile loading direction) do not show the same discontinuous ripping character. These side regions are largely undamaged. By comparison, both the off center and center specimens retain the discontinuous or ripping failure modes near the horizontal edge (perpendicular to the loading direction). During tensile testing it was observed that visible failure initiated at the horizontal edge of the DAFC and then propagated until failure. The ripping failure at the horizontal edge is indicative of the initial failure observed during testing and suggests failure in the laminates occurred due to shear forces at the horizontal DAFC edge. The uniformity of the center placement specimen implies that subsequent failure was influenced by normal stresses at the Kapton®/laminate interface. Conversely, shear

loading appears to have dominated the failure behavior in the off-center specimen as evidenced by the ripping patterns.

Figure 4.16 AFC specimen integrated in GFRP post failure. A center-integrated AFC is shown post failure in Figure 4.16. The AFC showed some fracture characteristics coincident with the DAFC. In most cases fractures occurred at the horizontal edge or separation occurred between the insert and surrounding laminates such that one face of the insert remained adhered to the adjacent laminate and the other face delaminated. Similarly, in the AFC the IDE remained adhered to the PZT fibers, while the Kapton速 pulled away with the adjacent ply. However, the Kapton速 in the center AFC shows signs of shear failures, without the dimple character reminiscent of the center DAFC. This discrepancy can be attributed to the fundamental differences in the AFC and DAFC core material. The PZT fibers offer a very different surface texture than the DAFC core (UD glass), which is a planar, therefore making a direct comparison between the actual element questionable.

4.5.2. CFRP Laminate Fracture Behavior In Figure 4.17 the AFC CFRP laminates with inserted (Figure 4.17a) and interlaced (Figure 4.17b) integration specimens post tensile testing are shown. The fracture patterns in the laminates were similar in that failure was seen to occur at the edges of the integration region. In general, transverse fracture through the width of the tensile specimens occurred at the top and bottom edge of the integration region. This behavior is slightly different from the GFRP specimens, which consistently exhibited failure at one edge of the integration region. Conversely, the CFRP specimens failed twice, on both edges of the integration region. It is also seen that the inserted laminates showed localized failure near the integration region, while the interlaced specimens exhibited more distribution in the failure character of the surface ply. One of the interlaced specimens even failed away from the integration region, closer to the tab. This suggests that the critical forces in the interlaced laminate were more

distributed over the laminate surface, while insertion integration lead to very high stress concentrations at the edge of the integration region. Closer inspection of the AFC CFRP specimens (Figure 4.18) showed that the Kapton速 film remained well-bonded to the CFRP plies, similar to the behavior seen with GFRP laminates. The PZT fiber core was severely fragmented and damaged, but remained bonded to the Kapton速 sheets. The fracture in the inserted laminate appeared to be more localized as compared with the interlaced specimen. Figure 4.18b shows more extensive delamination and failure on the surface of the interlaced laminate. Conversely, the insertion specimens showed more localized fracture near the AFC edge. The experimental results can be contrasted against the modeling findings to expand the characterization of laminate reliability.

Figure 4.17 CFRP Far Off Center AFC laminates with inserted (a) and interlaced (b) integration.

b Figure 4.18 Fracture behavior of AFC-CFRP laminates using the insertion (a) and interlacing (b) integration techniques.

4.6. Laminate Integrity Characterization Although a number of mechanical characterization studies have been performed concerning the integration of PZT devices in laminate materials, tools for assessing the material system integrity have not been presented. Therefore, the laminate materials and integration configurations were modeled in ANSYS, and laminate integrity was characterized by calculating the TWSI values for the experimentally observed failure forces, and comparing the results with the experimental fracture patterns. Laminate actuation ability was assessed in ANSYS to characterize smart material functionality of the laminate designs and compared with the relationship between the integration approach and laminate strength observed experimentally. In Figure 4.19 the TWSI values (computed for the maximum force experimentally observed) are plotted for the AFC integration region of the woven GFRP laminate with the AFC inserted in the middle 0째 ply of the composite. It was noted during testing that failure occurred in the Kapton layer prior to failure of the laminate. This explains the overcritical TWSI values, rather uniformly distributed over the integration area. Higher values can be noticed near the lateral edges. This corresponds closely to the observed fracture patters depicted in Figure 4.15. Figure 4.15 (a) shows the fracture of a DAFC integrated into the center of a 9 ply GFRP laminate using cutout integration and tested until tensile failure. A fairly uniform separation can be noticed across the entire Kapton interface. Conversely, the far off center fracture patterns display a more non-uniform pattern (Figure 4.15 (b)), with evidence of shearing and tearing within the Kapton layer of the integration region. This behavior is comparable with the TWSI plot shown in Figure 4.20, where the model results show critical TWSI values concentrated at the edges perpendicular to the tensile loading direction. A cross pattern is observed, with high TWSI values at the edges and moving towards the center of the integration region in a triangular pattern. This behavior is closely mimicked in the fracture pattern observed in Figure 4.15 (b), where tearing of the Kapton layer also occurred in a triangular fashion.

Figure 4.19 TWSI plot of integration region in a woven GFRP laminate with center inserted integration.

Figure 4.20 TWSI plot of integration region in a woven GFRP laminate with far off center inserted integration.

Figure 4.21 TWSI plot of the top ply of a CFRP interlaced far off center composite. Figure 4.21 depicts the TWSI plot for the top ply of the CFRP model with a far off center interlaced configuration. Stress concentrations are seen at the integration region edges, with a lower prediction of failure in the center of the integration region. These results closely follow the observed fracture patterns displayed in Figure 4.18b, where a clear fracture line follows the edge of the integration region. Sharp cracks can be observed running along the edges of the AFC region perpendicular to the direction of loading, but not directly over the AFC region. To characterize the laminate integrity in relation to the smart materials functionality, the FPF values were calculated for the laminates with varying AFC position, and compared with the actuation ability of the laminates. Maximum tip deflection of the laminate beam was defined as the measure of actuation ability. The modeling results are compiled in Table A.3 in the Appendix and graphically displayed in Figures 4.22 and 4.23. The maximum tip deflection is realized when the actuator is placed as far as possible from the neutral axis of the laminate. Tip deflection was used as an indicator of actuation ability simply because it was the easiest to implement and corresponded to the experimental test specimens. Figure 4.22 and 4.23 depict the relationship between smart laminate performance and structural integrity. Actuator performance and structural integrity were found to be two opposing design considerations. One will increase at the expense of the other. This illustrates the challenge in smart materials research, the question is how to balance the different design needs to create a robust active material system. Laminate integrity was found to be highest when the actuator was integrated in the middle of the laminate thickness. However, this placement position gives no ability to actuate the laminate. From an active laminate viewpoint, configurations with the AFC near the laminate neutral axis are unusable. Beam deflection actuation is only possible when the actuator is placed away from the neutral axis of the laminate. Both the 89

modeling and experimental results show that the integrity of the laminate decreases as the integration region is defined near the laminate surface. The FPF values decrease (as shown in the model) as do the ultimate strength values (as shown experimentally). Compared to cutout integration, insertion integration in GFRP laminates gives higher actuation capability at comparable integrity decrease. The additional fabric in the cutout configuration (see Figure 4.2) leads to a laminate with larger bending stiffness as opposed to the inserted configuration. For CFRP, the interlaced integration promises a lower affect on the integrity at comparable actuation capability compared to the inserted integration. Symmetric configurations show highest actuation capability while not further deteriorating laminate integrity. For CFRP, the symmetric configuration even shows a small integrity increase compared to the single far off center integration. In general, a larger affect of integration on FPF properties can be noted for the investigated fabric GFRP laminate compared to the cross-ply CFRP laminates. GFRP Model Results 60

m m

te r

ff C rO

te r en

ff C

ff C en

ff C

R ef er en c

ric

C en te r

te r

C en te r

Force (kN)

FPF Deflection

Inserted

Cutout

Figure 4.22 FPF plot of ANSYS results and actuation of GFRP laminates.

Deflection (mm)

CFRP Model Results FPF Deflection

C en

Fa r

O ff

C en te

er en c ef R

Insert ed

C en te r O ff C en Fa te rO r ff C en te r Sy m m et ric

0 te r O ff C en te r

0 r

Deflection (Îźm)

50 Force (kN)

Interlaced

Figure 4.23 FPF plot of ANSYS results and actuation of CFRP laminates.

4.7.

Conclusions

Evaluation of the impact of AFC integration on laminate integrity was assessed for different integration techniques and laminate materials types. As AFC were integrated away from the laminate symmetric axis, the mechanical integrity of the laminate decreased, while actuation ability of the AFC increased. However, this impact was much more pronounced in woven GFRP than in cross-ply CFRP. Actuator performance and structural reliability are two opposing design considerations. One will increase at the expense of the other if actuators are placed near the surface of the laminate to increase bending actuation ability. This illustrates the challenge in smart materials research, the question is how to balance the different design needs to create a robust active material system.

5. AFC Packaging Strategies 5.1. Introduction The AFC characterization study has shown that AFC are reliable so long as the integrity of the PZT fibers has not been compromised. Very good signal reproducibility is seen during fatigue loading below the fiber failure strain. After damage saturation above the failure strain AFC display very good long-term signal reproducibility, although at a lower sensitivity than in the undamaged state. The onset of fiber cracking leads to fragmentation in the fiber mat and a degradation of AFC sensor [72] and actuation [28] performance at higher strains. At low strain, even a fully fragmented AFC sensor can be expected to perform at an acceptable level relative to the undamaged AFC due to the closure of cracks in the fragmented fiber layer. Near identical behavior was observed in the literature for AFC actuators [28]. Even when fatigue loading is considered, the primary cause of AFC strain sensor performance degradation was attributed to loading beyond the tensile limit of the PZT fibers. It is evident, that reducing the onset of fragmentation is required to improve AFC reliability. The goal of the current investigation is the use of different AFC packaging strategies to improve the ultimate strain of the AFC and to extend the usable strain range before initiation of fragmentation in the PZT fiber layer. The study of AFC packaging strategies encompasses the different ways of designing an AFC-based smart material system, with the focus towards changing interaction between the passive material structure and the AFC. Packaging in this sense encompasses the ability to define the mechanical boundary conditions around the AFC. These can be manipulated in two ways: via the structural loading environment around the AFC, or by modifying the local mechanical interface of the AFC. Any packaging strategy must consider the impact on the electro-mechanical ability of the AFC. Therefore, the sensor and actuator function of the AFC must be balanced against the effect of the different packaging strategies. In establishing a packaging strategy, the different design requirements of the AFC as a sensor or actuator must first be considered.

5.2. Packaging Design Considerations 5.2.1. Actuator With a few exceptions, the packaging of PZT devices is not considered in their design, in favor of focusing on the piezoelectric charge coefficients and optimizing electricalmechanical abilities at the active material level. As an actuator, the AFC must be coupled with the host structure so as to transmit as much force and strain as possible 92

and thereby enact a structural change. Changing the AFC interface needs to consider the stiffness relative to the AFC and the host structures. Softening the interface could otherwise result in a shear-lag between the actuator and substrate. A thick bonding layer will increase the shear lag and therefore decrease the actuation forces transferred from the active to passive material [39, 73]. For an actuator, this means that forces and strains of the actuator might be accommodated by strains in the soft interface layer, thereby reducing its effectiveness. One PZT wafer design which considers the local loading environment is the LIPCA actuator [24]. In the LIPCA design a PZT wafer is sandwiched between two dissimilar composite layers, GFRP and CFRP. A PZT wafer is processed with these different materials, and upon cooling, due to the mismatch in coefficient of thermal expansion between the GFRP and CFRP layers, a curvature is realized. The curvature of the structure then imposes a compressive stress on the wafer element, oriented perpendicular to the polarization direction. The LIPCA design philosophy being, that the compressive loading condition ultimately leads to an improvement of the actuation efficiency of the PZT material. The use of CFRP and GFRP also allows the design of a lighter actuator than the traditional Thunder/Rainbow designs. When considering packaging strategies of the AFC, it is important to consider that the actuator needs to transfer actuation to its environment. Imposing a compressive stress along the desired direction of actuation will increase the force, which must be overcome to realize actuation of the element.

5.2.2. Sensor As a sensor, the function of the AFC is to receive mechanical stimuli from the operating environment, which is then processed by the signal processing component of the smart material system. Imposing a compression or tension onto a volume of PZT will then impose a mechanical stress in some orientation to the polarization direction. Since the AFC is primarily composed of PZT fibers, as a sensor high compressive forces are not detrimental to the mechanical integrity of the composite. As is known from bulk-PZT research, ferroelastic depolarization can occur when an external force deforms the polarized electrical domains beyond a certain threshold, termed the coercive stress. So any modification of the mechanical boundary conditions must consider how loads might affect ferroelastic depolarization behavior which could lead to depolarization over longer time periods. This is less critical for an actuator, since the polarization direction is also in line with the direction of applied driving voltage, and the electrical domains can be oriented during operation. In the case of the sensor repolarization may not be possible during use, and therefore ferroelastic depolarization must be considered in a packaging strategy.

5.2.3. Laminate Design Packaging can include modifying the interfacial properties of the AFC or by modifying the force transferred to the device via manipulation of the integration procedure or laminate design. Different integration concepts can be applied, such as 93

cutout versus simple insertion or interlacing of laminate plies as presented in the Laminate Integrity investigation. Switching between cutout and simple insertion techniques has been shown to have little effect on the sensor ability of PZT wafer devices as was seen in a study by Mall [49] where PZT wafers were integrated into CFRP [0/+-45/90]s laminates and subjected to increasing tensile strains. The output signal from the PZT wafer was monitored, and only a slight variation was seen in the signal character of the simple and cutout insertion laminates. Additionally, the laminate design can be changed by modifying the type of laminate (woven versus UD) or the orientation of plies, so as to affect the forces transferred to the integration region in response to a global strain. The sensor property of PZT is dependent on the force applied to the element. In a homogeneous laminate the force transferred to the AFC will essentially remain constant for simple and cutout insertion. However, using a different lay-up for the laminate brings a new dimension to the smart material system design. For example, in a [0/90] based composite laminate the stiffness of each ply is dependent on their orientation to the loading direction. The 0째 plies, being significantly stiffer than the 90째 plies, will carry the majority of the force when a global external strain is applied to the laminate. Integrating an AFC into a 90째 ply versus a 0째 ply would affect the force transfer to the AFC and possibly affect its performance ability.

5.2.4. Interface The mechanical properties such stiffness and ultimate failure strain or load of the interface layer between the AFC and the laminate can be larger or smaller than that of the AFC and the laminate. Given the relatively low tensile strain limit of the PZT fibers (as compared with GFRP or CFRP) simply using a stronger interface material will not lead to an improvement of the AFC ultimate strain. It would however, allow transfer of applied loads to the stiffer interface instead of to the AFC. Since AFC strain sensor performance is linked to the force transferred to the PZT fibers, it would be ideal to maintain force transfer while reducing strain transfer. A large sensor signal could then be realized, but without approaching the failure strain limit of the AFC.

5.3. Current Work Mechanical manipulation of the AFC interface properties investigates three different states: soft, stiff, and compression. The stiffness definition of the different interfaces is in relation to the stiffness of the AFC, which has been measured to be 29 GPa [72]. In the current study, AFC sensor performance was assessed in two different laminate and three different interface configurations. The laminates included a woven GFRP and cross-ply CFRP. A soft interface was created by coating AFC with silicon. The stiff interface refers to lamination of AFC with UD-CFRP. Finally a compressive interface was produced by laminating the AFC between layers of pre-tensioned UDCFRP. The three interface configurations represent three different packaging strategies: strain shielding, force shielding, and combined force and strain shielding. 94

The shielding concept implies shielding the AFC against the transfer of mechanical inputs (force and/or strain) from the surrounding integration environment, which as for the previous AFC Characterization study, constituted a woven GFRP laminate.

5.4.

Experimental

To evaluate different AFC packaging methods, active laminates were produced by integrating AFC into GFRP and CFRP laminates. This allowed for the evaluation of the effect of laminate environment on AFC sensor performance. AFC were also integrated in GFRP laminates with modified interfaces including silicon, CFRP, and pre-tensioned CFRP interface layers. The experimental investigation consisted of cyclically straining the active laminates with the different packaging strategies to evaluate their effect on AFC strain sensor performance.

5.4.1. Composite Laminates Laminates consisted of a 9-ply GFRP (Isolvolta Isopreg HR 320P-40 plain weave prepreg) and 12-ply CFRP (ELITREX EHKF 420-UD24K-40 pre-preg). Manufacturing procedures were identical to those described in the AFC Characterization and Laminate Integrity chapters. AFC were integrated in GFRP laminates using the cutout method, while AFC were integrated into CFRP laminates using the insertion and interlacing approaches as described in the Laminate Reliability chapter. The specific laminate types and integration methods are listed in Table 5.1. The CFRP plies had a thickness of 0.12mm and the GFRP plies were 0.24mm thick. Table 5.2 lists probes which were tested including the interface used for interface evaluation, including laminate type, interface type, integration approach including integration position in the laminate thickness, and the number of plies. Table 5.1 Description of laminate types used for tensile testing. Material Type Plies Stacking Integration Position GFRP Woven 9 [0]9 Cutout Center Far Off-Center CFRP UD 12 [0,90]3S Insertion CFRP UD 12 [0,90]3S Interlaced Far Off-Center Table 5.2 Description of probes and modified interface AFC test specimens. Laminate Interface Integration Position Probes GFRP Silicon Cutout Center 3 GFRP CFRP Cutout Center 3 Center 3 GFRP PTEN CFRP Cutout GFRP Cutout Center 4 CFRP Insertion Far Off-Center 5 CFRP Interlaced Far Off-Center 5

5.4.2. Modified interface AFC Modified Interface AFC (MI-AFC) included three different types of AFC with modified interfaces; stiff, soft, and pre-tensioned. To obtain a soft interface AFC were encapsulated in silicon rubber (RTV Type V - www.swiss-composite.ch). Two-part silicon was mixed and degassed in a vacuum chamber. The AFC were then dipped in the liquid silicon, removed, and allowed to cure in air. It was difficult to accurately control the thickness and uniformity of the silicon layer. Therefore the resulting AFC (Figure 5.1) had a non-uniform thickness, with an average value of 0.345 mm.

Figure 5.1 AFC encapsulated in silicon rubber. A stiff interface was obtained by laminating AFC elements with thin layers of UDCFRP (KUBD 1506 â&#x20AC;&#x201C; www.krempel.com). Four total plies were used, two above and two below the AFC. AFC were cleaned with acetone before being sandwiched between the CFRP, with the fiber direction of the AFC in alignment with that of the CFRP. The specimens were then placed on an aluminum tool plate and processed in a heated press at 140Âş C for 15 minutes. The CFRP plies had a nominal thickness of 0.14 mm, and the total specimen geometry is shown in Figure 5.2.

Figure 5.2 AFC specimen laminated with CFRP material. A pre-tensioned compressive stress interface was obtained by sandwiching AFC elements between UD-CFRP pre-pregs and curing while tension was applied to the CFRP. Upon curing and release of the applied tension, a compressive stress developed in the CFRP layers (Figure 5.3). The CFRP used was identical to that used in the untensioned stiff AFC. UD-CFRP tape was wrapped around the pre-tensioning test rig (Figure 5.4). Two layers were wrapped, followed by the application of three AFC 96

along the tape, and two more layers of CFRP tape to sandwich the AFC in place. One section of the pre-preg was quickly cured. This sealed the CFRP tape and prepared it for tensioning. A plunger was extended from the test rig to apply pressure on the CFRP tape and tension it. A pre-tension stress of approximately 600 MPa was applied while flexible heating elements located directly below the CFRP section with AFC enabled curing. A top beam was placed over the CFRP and an air chamber underneath the CFRP provided pressured to push the plies together. After a cure time of 15 minutes the pressure was released from the CFRP tape and the cured section was removed to be prepared for integration in the GFRP laminates. Specimens dimensions were the same for the normal and pre-stressed CFRP interface as shown in Figure 5.2.

Figure 5.3 Pre-tensioning concept and applied compressive force on the AFC.

Figure 5.4 CFRP pre-tensioning curing apparatus.

5.4.3. Tensile Testing Cyclic tensile testing was performed with an Instron 1251 test machine in accordance with the cycle testing described in the AFC Characterization chapter. A strain controlled program cycled the laminates at each desired strain level, and the AFC 97

output signal was recorded in LabView. Unidirectional strain gages were affixed to the specimens and a strain controlled test procedure was used to oscillate the specimens around a prescribed strain level at 1 Hz for ten seconds with strain oscillation amplitude of 0.01%. Specimens were loaded to the desired maximum strain in strain steps, which were defined as an increase of 0.05%. The specimens were then unloaded and cycled along the same loading path.

5.5.

Results-Discussion

5.5.1. Influence of laminate configuration on AFC sensor performance Figure 5.5 displays the AFC strain sensor results for AFC integrated in CFRP laminates. No difference was seen between inserted and interlaced specimens. However, a large deviation was seen between different specimens. In Figure 5.6 previous results of AFC integrated into GFRP (as discussed in the AFC Characterization chapter) are shown along with the current data collected on cross-ply CFRP specimens. The strain-sensor performance of the CFRP system differs noticeably from the GFRP system performance as reported in the AFC Characterization chapter. According to the damage model presented in the AFC Damage Evaluation chapter (Sec. 3.1.1.), the AFC performance curve exhibits a few key parameters including, damage onset, fragmentation range, saturation point, as well as unloading and recovery characteristics. The CFRP performance shows a compressed and shifted profile as compared with the GFRP curve. Furthermore, the GFRP specimens show a greater difference between the maximum and minimum values, and also show a higher sensitivity throughout the entire strain range. By comparison, the CFRP specimens display a lower sensitivity at 0.05% and performance decreased upon loading to 0.10%. The GFRP laminates showed stable signal performance up to 0.15% while The CRFP laminates displayed a noticeable decrease above 0.05% strain. Previously, the GFRP laminates showed good recovery of AFC performance upon reduction of applied strain, by comparison, the CFRP specimens did not. When strained to 0.50% and then tested at 0.05%, the CFRP specimens showed a performance recovery of only 80% whereas 95% performance recovery is characteristic of the GFRP system. The loading curves for both composite systems coincided with one another after strained beyond the fragmentation saturation limit near 0.50%.

AFC Integrated in CFRP

AFC Signal Amplitude (C)

1.2E-07 1.0E-07

Loading Unloading

8.0E-08 6.0E-08 4.0E-08 2.0E-08 0.0E+00 0.00

0.10

0.20

0.30

0.40

0.50

0.60

Strain (%)

Figure 5.5 Stain sensor performance of AFC integrated in CFRP.

AFC Signal Amplitude (C)

AFC Integrated in GFRP and CFRP 1.6E-07

CFRP Load CFRP Unload GFRP Load GFRP Unload

1.4E-07 1.2E-07 1.0E-07 8.0E-08 6.0E-08 4.0E-08 2.0E-08 0.0E+00 0.00

0.10

0.20

0.30

0.40

0.50

0.60

Strain (%)

Figure 5.6 Performance curves for AFC integrated into GFRP and CFRP. In Figure 5.5 the CFRP data displays a rather large standard deviation between specimens, much more than was observed in the GFRP system, which maintained at most a 10% deviation between specimens. Conversely, the CFRP system exhibited deviations from between 10% to 30%. The larger deviation in the CFRP system can be explained by considering fiber misalignment of the composite plies. It is evident from Figure 5.5 that a larger deviation exists during the initial loading curve, and is reduced at higher strains. As the laminate is strained, forces on the misaligned CFRP fibers would force them to become more aligned with respect to the direction of tensile loading at higher strains. Misalignment of the CFRP fibers in the 0째 plies would have a large affect on the force transfer to the AFC. During the tensile test the misaligned CFRP fibers would have an affect on the force transferred to the AFC, and hence a variation in strain sensor performance would be seen. In the GFRP laminate system, the AFC were integrated into a somewhat uniform loading environment in that the woven GFRP plies all had similar material properties and an effect due to fiber misalignment would not be seen. However, with the cross-ply CFRP system, load transfer to the AFC is dependent on the fiber layer orientation of the CFRP material.

Therefore, a slight misalignment in the 0º plies would have a pronounced effect on force transfer to and hence on strain sensor performance of the AFC. Although the GFRP and CFRP tests were identical in terms of applied strain, there existed two fundamental differences between the two host laminate systems. First the Young’s Modulus of the CFRP material was much greater than that of the AFC, while GFRP exhibits a material stiffness quiet near that of the AFC. Second, the CFRP laminate was a cross-ply, while the GFRP system was woven. The AFC were integrated into low load-bearing 90º plies in the CFRP laminates. Since the GFRP plies were woven, integration of the AFC in that system was in a higher load-bearing zone. Hence, a larger force was transferred to the AFC in the GFRP system as compared with the CFRP system. This concept can be illustrated by analytically comparing force transfer in a woven GFRP and cross-ply CFRP laminate, while assuming integration into the 90º CFRP plies. 5.5.1.1. Analytical Laminate Comparison Assuming linearly elastic behavior, forces along the loading direction of a laminate are distributed according to the modulus of each ply, and are added together as shown in Figure 5.7 to account for the total force carried by the laminate as per Equation 5.1, where F represents the force existing in each ply, numbered from 1 to n. The strain is considered to be uniform, and perfect bonding between laminate layers is assumed. For linearly elastic materials, the E modulus of the laminate can be estimated according to Equation 5.2, where V represents the volume fraction of each constituent. Stress and strain can be expressed as the modulus multiplied by the strain and as force divided by area as shown in Equation 5.3 where σ is the stress in the laminate, ε the applied global strain, E is Young’s Modulus of the laminate, A is the total laminate area. Since the area fraction is proportional to the volume fraction of the materials, the area is used for simplification. Then the force for each ply can be expressed by the modulus multiplied by cross-sectional area and the applied strain as shown in Equation 5.4. The material properties outlined in Table 5.3 were used to compute the theoretical difference in forces carried between the CFRP and GFRP laminates.

Figure 5.7 Schematic view of force distribution per ply in a laminate. F = F1 + F2 + ... + Fn E Lam = E1V1 + E 2V2 + ... + E nVn

100

(Equation 5.1) (Equation 5.2)

σ = Eε =

F A

F = [(EA)1 + (EA)2 + ... + (EA)n ]ε

(Equation 5.3) (Equation 5.4)

Table 5.3 Material properties for GFRP and CFRP laminates. GFRP 0º CFRP 90º CFRP E Modulus (GPa) 29 124 8 Thickness (mm) 0.26 0.12 0.12 Figure 5.8 provides a theoretical comparison between the forces existing in the different plies when strained to equal levels in the GFRP and CFRP laminates with different ply orientations within the pre-fragmentation strain range up to 0.10 %. Here it is illustrated quite well that for a comparable strain range, a woven GFRP ply carries significantly more load that a 90º CFRP ply. Q = −d 33 F

(Equation 5.5)

Equation 5.5 illustrates the basic force-charge relationship which exists in the AFC, where Q is the charge generated due to an applied force (F) and governed by the charge coefficient (d33). Since force transfer relates to sensor ability for piezoelectric materials, the AFC should produce a lower signal and hence, perform at a lower level when integrated into the 90º plies of the CFRP laminate as compared with the woven GFRP laminate. From a system design perspective, this is not necessarily a limitation. As was shown in the Laminate Reliability chapter, the integration of AFC had a lower impact on the degradation in mechanical integrity of the cross-ply CFRP laminates than of the GFRP laminates, due to the fact that integration occurred in the low loadbearing plies. This is an example of improving laminate reliability in favor of reducing AFC performance ability.

101

Ply Force Comparison 5000

Woven GFRP 0 deg CFRP 90 deg CFRP

Force (N)

4000 3000 2000 1000 0 0

0.02

0.04

0.06

0.08

0.1

Strain (%)

Figure 5.8 Comparison of forces transferred through a laminate ply for woven GFRP as compared with cross-ply CRFP for comparable strains. Despite the difference in signal performance between the two laminate systems, the character of the curves in Figure 5.6 both retain similar forms during the fragmentation stage of the strain history. Near 0.20% the curves in both laminates show a slope change, followed by the degradation in performance before a second slope change, signifying the point of fragmentation saturation. This behavior is exhibited more clearly in the GFRP system, but plainly also exists in the CFRP system. While the difference in sensitivity between the two systems can be explained by force transfer and the effect on sensor output, the agreement in slope changes is indicative of fragmentation initiating at similar strain levels, implying that failure of the PZT fibers is a strain, as opposed to a force dominated phenomenon. This behavior can also be seen by comparing the findings of AFC actuation ability at increasing strain levels with the findings of the AFC Characterization chapter. In their study of AFC actuation durability, Wickramasinghe and Hagood [28] laminated AFC with two plies of UD GFRP and strained the elements in 0.05% strain steps from 0.05% to 0.70% and actuated the specimens at each strain level. The observed degradation in actuation performance was statistically insignificant from the findings of sensor performance at increasing strain levels, which was presented in the AFC Characterization study. Wickramasinghe and Hagood used a 2-ply GFRP-AFC laminate, while the current work utilized a 9-ply laminate. Both studies used a laminate with uniform load transfer to the AFC, but in the current study the nonuniform CFRP cross-ply laminate displayed different behavior. By contrasting the results for AFC actuator and sensor durability, it appears as though fragmentation in the AFC is related to the applied strain, as opposed to the applied force.

102

5.5.2. Influence of AFC interface modification on sensor performance 5.5.2.1. Silicon The silicon interface AFC showed no response during the tests. The silicon thickness was no doubt too large, enabling a very large strain shielding effect, which prevented global laminate strains from being seen by the AFC. Hence, both force and strain shielding occurred, severely impacting the performance of the AFC as a strain sensor. This represented an extreme case, where AFC reliability was dramatically increased, in the sense that the fiber layer showed no sign of damage, but the performance was greatly reduced, in the sense that it could no longer be used as a strain sensor. Reducing this effect could be accomplished by using a thinner silicon layer, thereby reduce the shear-lag effect. However, another problem with the silicon is the very poor adhesion between the Kapton surface and the silicon. It was noted during manufacturing that the silicon layer could be easily pealed away if needed. Since the laminate was pressed during processing, sufficient friction between the Kapton and silicon should have existed to transfer force to the AFC. However, it can not be guaranteed that slippage would not occur at higher strains. 5.5.2.2. CFRP Interface

AFC with CFRP Interface CFRP Int Load CFRP Int Unload GFRP Load GFRP Unload

AFC Signal Amplitude (C)

1.8E-07 1.6E-07 1.4E-07 1.2E-07 1.0E-07 8.0E-08 6.0E-08 4.0E-08 2.0E-08 0.0E+00 0.00

0.10

0.20

0.30

0.40

0.50

Strain (%)

Figure 5.9 Sensor performance curve of AFC with CFRP interface. The CFRP-Interface AFC (Figure 5.9) exhibited strain performance behavior similar to the previous results for GFRP and CFRP laminates. The initial loading region showed an increase in sensitivity, followed by a performance degradation, which saturated near 0.35% to 0.40% strain. The sharp initial increase might be due to 103

loading of the CFRP interface. This effect is evident at initial strains to 0.10%, beyond which no loading region is seen. Between 0.15% and 0.35% the characteristic signal degradation in response to the applied strain is seen. Given the increased stiffness of the CFRP as compared to the AFC, force which normally would be seen by the AFC would be carried in part by the CFRP interface, leading to a slight reduction in AFC strain sensor performance. The degradation line, indicative of fiber fragmentation rate, extends from the initial damage initiation point to the saturation slope change. The slope was essentially the same as that seen in the normal GFRP laminates, suggesting no improvement in fragmentation rate. The body of the performance curve is also compressed, approaching a linear line, much more so than that of the interface-less AFC. The recovery line indicates crack closure behavior. Without an interface the recovery line has a character similar to a 1/x function. With the CFRP interface the 1/x function character is reduced and approaches more of a linear character. In contrast to the small standard deviation observed for AFC integrated in GFRP laminates (AFC Characterization chapter), the variation between specimens with CFRP interfaces was much greater, similar to the finding of AFC integrated in CFRP laminates. This made characterization of the specific sensor performance more problematic. While the general form of the curves was reproducible, the exact amplitude values were found to differ by up to 20% between specimens. Conversely, the AFC integrated into GFRP laminates showed very good reproducibility between different specimens on the order of 3%-10%. The variation in signal response can be attributed to misalignment of fibers in the CFRP interface. Integration in a homogeneous laminate ensures uniform force distribution to the AFC. However, with a CFRP interface, the load transfer to the AFC is governed by the stiffness of the interface. The uni-directional CFRP plies were aligned with the AFC fiber direction before processing, but there was no way to ensure that the CFRP fibers remained aligned during the pressing process. Misalignment in the CFRP interface would then correspond to a variation in force transfer to the AFC, and also to a variation in the AFC strain signal response. Therefore, misalignment of the CFRP interface is believed to explain the higher variability in the observed results. This is similar to the large deviation seen for cross-ply CFRP laminate, attributed to carbon fiber misalignment during processing which lead to variation in force transfer through different laminate layers, and hence a variation in AFC strain sensor performance. One serious drawback of the CFRP interface was the penalty to laminate integrity. Cracking could be heard upon loading above 0.50%, and one specimen failed at 0.80% strain, which is far below the characteristic failure strain of the GFRP laminate (about 2-3% strain). Evidence of debonding could be seen near the edge of the integration region, perpendicular to the laminate loading direction as shown in Figure 5.10.

104

Figure 5.10 AFC CFRP interface specimen which failed at 0.80%. 5.5.2.3. Pre-Tension Interface The Pre-Tensioned CFRP (PT-CFRP) interface AFC (Figure 5.11) exhibited interesting behavior, showing evidence of extreme modification to the fiber fragmentation character. In the first loading region, the change in sensitivity from 0.05% to 0.10% is much larger than that seen with the non-stressed CFRP interface. This is similar to the increase seen with a normal CFRP interface, related to the loading of the stiffer interface.

AFC Amplitude (C)

PreTension AFC Performance PreTen Load PreTen Unload Normal Load Normal Unload

2.E-07 1.E-07 5.E-08 0.E+00 0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

Strain (%)

Figure 5.11 AFC strain cycle performance curve with pre-tensioned CFRP interface and no interface. 105

Beyond the initial loading region, which is significantly different from the previous results, the sensitivity decreased very slightly between 0.10% and 0.25% strain. The performance degradation is very slight, showing a much more linear trend than the non-linear behavior seen with the CFRP or no interface. This effect is then due to the pretension of the CFRP layers as the remnant compressive stress is balanced with the global applied strain. This pretension must be overcome before damage can occur in the AFC. The slight slope decrease shows that the fiber fragmentation behavior was also affected, displaying an extended region with a very small slope until 0.30%. The pretensioned interface effectively required loading from the global strain boundary conditions, and reduced the onset and character of fragmentation in the PZT fiber layer. The area of the performance curve became very compressed and approached a more linear character. The recovery behavior was also greatly affected. The data clearly shows that the difference between performance levels at lower and higher strains are very similar, suggesting that fiber gap expansion is very small. A large number of fiber fragments and opening of cracks would be evidenced as a larger decrease from the lower to higher strain levels. However, here the hysteresis is very small. This is due in large part to the compressive state on the PZT fibers. During loading and unloading this remnant compression is resisting fiber fragmentation and would encourage crack closure. In comparison to the results for the AFC with a CFRP interface, the standard deviation for the PT-CFRP interface AFC was very low, on the order of 3%-10%, where as 10%-30% was characteristic of the CFRP interface AFC. A CFRP fiber misalignment of the interface was used to explain the rather large variation seen with CFRP interface AFC. During the pre-tensioning process, the CFRP fibers are all aligned along the fiber direction, and so long as the AFC were also aligned to the CFRP fiber direction, there would be much less variation than that seen with the non-tensioned CFRP specimens. This supports the assertion that a control of the variation in manufacturing processes is important for maintaining reproducibility between specimens with modified interfaces utilizing CFRP.

106

Prestressed Interface Signal Strength 7.E-06

AFC Signal (C)

6.E-06

Pretension Interface

5.E-06

No Interface

4.E-06 3.E-06 2.E-06 1.E-06 0.E+00 -1.E-060.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

Strain (%)

Figure 5.12 Comparison between normal AFC and prestressed CFRP interface. An improvement is also seen in the absolute sensor signal from the PT-CFRP interface AFC. Figure 5.12 compares the signal strength of the normal AFC and the PT-CFRP AFC. The PT-CFRP interface AFC shows improved linear behavior, and greatly improves the ability of AFC to act as a strain sensor. At 0.20% the normal AFC signal becomes nonlinear, while the PT-CFRP interface AFC only shows a slight slope change at 0.30%. Despite the improvement to AFC reliability, the addition of a PTCFRP interface was detrimental to the mechanical reliability of the laminate material system as a whole. During loading of the pre-tensioned AFC, cracks could be heard in the laminates, pointing to eventual early failure of the laminates. This occurred at strains as low as 0.25% and was visually observed as delamination between the GFRP laminate and the CFRP interface. For this reason, testing was only conducted until a strain of 0.35%. The delamination was no doubt due to the large stiffness difference between the GFRP and CFRP laminate plies, as compared with the AFC. The mismatch in stiffness means that large stress concentrations would occur at the interface between the two dissimilar composites. Pre-tensioning of the interface results in compression of the PZT material and works directly against the tensile loads developed due to a global external strain. This problem could be alleviated by pretensioning the entire laminate or using a CFRP laminate for integration, so that the interface exhibits a similar stiffness to the host laminate.

5.6. Conclusions The current work has shown the affect of integration environment on the performance of AFC sensors. The choice of laminate and interface material can have a profound affect on AFC performance by modifying the force transferred to the AFC. The addition of the silicon interface showed that AFC performance can be extended, at the expense of the electrical-mechanical properties. With the CFRP interface, fiber fragmentation behavior could be improved and approached a more linear character, but 107

the critical points of fragmentation onset and saturation remained very similar. The pre-tensioned CFRP interface offered elements of both force and strain shielding, changing the fragmentation onset and saturation behavior. With the addition of pretensioning to the CFRP interface, the compressive stress does not appear to be detrimental to the AFC, and appears to provide an enhancement for the electromechanical properties of the AFC. However, the pronounced difference in material stiffness of the CFRP interfaces and the GFRP laminates lead to stress concentrations at the interface edges between the composites, which seriously impinged on laminate reliability by decreasing the ultimate failure limit of the laminate.

108

6. Concluding Remarks AFC have been characterized with a focus towards their long term reliability. In the current work, AFC have been shown to be robust, smart material devices, which can be reliably integrated into and processed with GFRP and CFRP composite materials. This has resulted in materials with active sensing and actuation as well as passive load carrying capabilities. It was found that the integration of AFC in GFRP reduces the tensile strength of the laminate, but integration in cross-ply CFRP laminates showed improved mechanical performance by manipulating the force transfer characteristics between the AFC and host structure. Although the AFC were found to perform well under fatigue loading conditions, the low strain limit of the PZT fibers limited the AFC minimum strain to about 0.20% before the onset of fragmentation in the fiber layer. During fragmentation it was observed that cracks were continuously occurring near the edge of the electrode fingers. The electrical polarization character near the IDE fingers is very asymptotic, and it is known from literature that the polarization character of PZT ceramics will influence their mechanical properties. A primary finding of the characterization investigation was that the electrical polarization of PZT fibers in the AFC modify the mechanical properties of the fibers in the region of changing electrical fields near the electrode fingers. This effect could be modeled to evaluate new electrode designs for the improvement of the mechanical integrity of the devices. However, even vast modifications the electrode geometry would not influence the fact the material properties are modified during electrical polarization, leading to stress concentrations in the fibers. A packaging solution was developed to improve the AFC. While stress concentrations could not be modified, as they result from the optimal electrode geometry, the onset of fragmentation could be manipulated using packaging techniques. The performance of AFC could be improved by using a compressive packaging strategy and pre-tensioned CFRP to place the PZT fibers under a compressive load. In this solution, CFRP layers with the AFC were cured under tension. Upon release of the pre-tension load, the AFC was placed under a compressive stress. This meant that the ceramic PZT fibers were now in a more stable state than previous, and better suited to withstanding tensile loads. This resulted in a device which was still very thin and could be integrated into laminate materials, but which failed at higher tensile loads than the previous design. The compressive packaging strategy improved the AFC strain sensor performance and extended the usable strain limit of the material system by placing the PZT fibers under a compressive state, reducing the onset of fiber fragmentation during tensile loading. The compressive state, did not however hinder the electrical mechanical response of the AFC, which is critical, since doing so would have reduced the effectiveness of the sensor function of the AFC. The electrical sensor response of the pre-tension AFC was equal to that of the control AFC at very low strains, and at higher strains the pre109

tension AFC reduced the onset of fiber fragmentation. This therefore also reduced the onset of nonlinearity in the AFC when used as a strain sensor, thereby improving its use as a sensor. In total, the current study was successful in characterizing and improving the performance of AFC devices, as well as characterizing the impact of AFC integration on GFRP and CFRP laminates. Currently AFC are manufactured as standalone devices which are integrated into structures as foreign components. A desire for the future is to create fully integrated functional materials. This could be accomplished by integrating passive and active fibers together in a single structure. Combing PZT fibers with GFRP fibers and sheet electrodes could result in robust active sensing and actuation composite panels. For the adoption of AFC technology in its current form, manufacturing realities and costs create a problem, and AFC are generally considered too expensive to use on a broad scale in applications such as vibration reduction in automobiles and airplanes, or active aeroelastic control of structures. However, as the concept of using smart materials permeates into the collective knowledge of engineers, more applications and solutions to current problems will no doubt find solutions using smart material design principles.

6.1. AFC Outlook AFC have generally been used and developed for aeroelastic and structural actuation applications since they fit the profile for a planar high force actuation device, which can be easily integrated into wing or rotor structures. However, many high-tech applications are waiting to be discovered in the area of biomedical engineering and the integration of smart materials. One application of piezoelectric materials in biomedical applications is as the actuation surface in a micro pump design for fluid transport, but this concept is also playing off of traditional concepts of how the technology can be developed. A key concern in smart materials design is fitting capabilities to the ideal application. Virtually no research has been conducted in the combination of piezoelectric smart materials and biomedical applications, and this area represents an area which might be capitalized upon. It has been shown that biological materials respond to mechanical inputs such as force and strain. The ability of AFC to actuate in a certain direction and at high frequencies over millions of actuation cycles means that they could have a place in improving cell growth in tissue engineering applications. Scaffolds seeded with cells could be affixed to AFC substrates with different directions to actuate the cell layer in different directions, tensile and compression, over long time periods of time, and thereby texture the mechanical properties of the cell layer in a desired way.

110

Acknowledgements A number of colleagues and individuals provided invaluable assistance and guidance during this work including, but not limited to the following. Prof. Dr. Paolo Ermanni, head of the Centre of Structure Technologies at the ETH Zurich for being my doctoral adviser and the Examiner. Prof. Dr. Edoardo Mazza, who acted as the Co-Examiner from the ETH, and provided rigorous critic of this work. Dr. Michel Barbezat, for being my advisor at Empa and overseeing the day to day challenges of the research world. Dr. Rolf Paradies, who I worked with on AFC topics, and provided excellent guidance and cooperation on projects from active structures to modeling of AFC. Andrea Bergamini, who as a fellow researcher at Empa in the area of smart materials provided excellent input on projects and discussions. Alberto Belloli, who collaborated on AFC investigation topics including the modeling of smart laminates and the impact on mechanical integrity. Dr. Xavier Kornmann, who initiated much of the work on AFC at Empa Dübendorf, which preceded this dissertation work. Dr. Andreas J. Brunner at Empa, who performed acoustic emission testing support on various parts of this project. Dr. Niccolò Pini, who worked in the field of polymer-based electrodes for AFC and provided excellent support in our collaboration on testing his novel AFC. Peter Flueler, originally the head of composites department at Empa where I started my research path within the ETH community. The Mechanical Systems Engineering laboratory at Empa, which funded the work, and in particular Lorenzo De Boni, Daniel Voelki, Max Heusser and Marcel Rees and Marcel Birchmeier who provided indispensable support at various levels of this work. Without their knowledge and abilities the practical day to day research activities would not have been possible. Many students from ETH Zurich deserve my gratitude due to their work in this investigation including, Alain Monnin, Benjamin Schläpfer, Roberto Reitmayr and Laurent Wahl.

111

Also deserving of mention are various literary, scientific, and industrial voices for their collective inspiration, including that of James Douglas Morrison (of The Doors), Ursula K. Le Guin, Hunter S. Thompson, Howard Hughes, Ian Curtis (of Joy Division), Chuck Palahniuk (for Fight Club), and many others. My family also provided an irreplaceable source of support and inspiration for their example of how to live and overcome challenges. In final, my thanks go to Iris Sprow for continued emotional support during the most difficult times.

112

References 1. Head, I., Active Dampening System (EDS) Protector Racquet, in www.head.com/tennis/, www.head.com/tennis/, Editor. 2.

Head, Intelligence Ski System. www.head.com/ski/.

3. Wickramasinghe, V.K. and N.W. Hagood, Durability characterization of active fiber composite actuators for helicopter rotor blade applications. Journal of Aircraft, 2004. 41(4): p. 931-937. 4. Inman, D.J., M. Ahmadian, and R.O. Claus, Simultaneous active damping and health monitoring of aircraft panels. Journal of Intelligent Material Systems and Structures, 2001. 12(11): p. 775-783. 5. Tseng, K.K. and L. Wang, Smart piezoelectric transducers for in situ health monitoring of concrete. Smart Materials & Structures, 2004. 13(5): p. 1017-1024. 6. Kudva, J.N., Overview of the DARPA Smart Wing project. Journal of Intelligent Material Systems and Structures, 2004. 15(4): p. 261-267. 7. Suleman, A. and A.P. Costa, Adaptive control of an aeroelastic flight vehicle using piezoelectric actuators. Computers & Structures, 2004. 82(17-19): p. 1303-1314. 8.

Forterre, Y., et al., How the Venus flytrap snaps. Nature, 2005. 433(7024): p. 421-425.

9. Hufenbach, W., M. Gude, and L. Kroll, Design of multistable composites for application in adaptive structures. Composites Science and Technology, 2002. 62(16): p. 2201-2207. 10. Schultz, M.R. and M.W. Hyer, Snap-through of unsymmetric cross-ply laminates using piezoceramic actuators. Journal of Intelligent Material Systems and Structures, 2003. 14(12): p. 795-814. 11. Schultz, M.R. A new concept for active bistable twisting structures. in Proceedings of SPIE - The International Society for Optical Engineering. 2005. 12. Otsuka, K., Y. Xu, and X. Ren. Ti-Ni-based shape memory alloys as smart materials. in Materials Science Forum. 2003. 13. Shahinpoor, M. and K.J. Kim, Ionic polymer-metal composites: I. Fundamentals. Smart Materials and Structures, 2001. 10(4): p. 819-833. 14. Song, G., X. Zhou, and W. Binienda, Thermal deformation compensation of a composite beam using piezoelectric actuators. Smart Materials & Structures, 2004. 13(1): p. 30-37. 15. Lynch, C.S., The effect of uniaxial stress on the electro-mechanical response of 8/65/35 PLZT. Acta Materialia, 1996. 44(10): p. 4137-4148. 16. Schaufele, A.B. and K.H. Hardtl, Ferroelastic properties of lead zirconate titanate ceramics. Journal of the American Ceramic Society, 1996. 79(10): p. 2637-2640. 17. Alguero, M., et al., Degradation of the d(33) piezoelectric coefficient for PZT ceramics under static and cyclic compressive loading. Journal of the European Ceramic Society, 2001. 21(10-11): p. 1437-1440. 18. Cao, H.C. and A.G. Evans, Nonlinear Deformation of Ferroelectric Ceramics. Journal of the American Ceramic Society, 1993. 76(4): p. 890-896.

113

19. Tanimoto, T., K. Yamamoto, and T. Morii. Nonlinear stress-strain behavior of piezoelectric ceramics under tensile loading. in IEEE International Symposium on Applications of Ferroelectrics. 1994. University Park, PA, USA: IEEE. 20. Tanimoto, T., K. Yamamoto, and T. Morii, Nonlinear stress-strain behavior of PbZrO3-PbTiO3 under various temperatures. Japanese Journal of Applied Physics, Part 1: Regular Papers & Short Notes & Review Papers, 1994. 33(9 B): p. 5341-5344. 21. Tanimoto, T., K. Okazaki, and K. Yamamoto, Tensile Stress-Strain Behavior of Piezoelectric Ceramics. Japanese Journal of Applied Physics Part 1-Regular Papers Short Notes & Review Papers, 1993. 32(9B): p. 4233-4236. 22. Furman, E., G. Li, and G.H. Haertling. Electromechanical properties of rainbow devices. in IEEE International Symposium on Applications of Ferroelectrics. 1994. 23. Mossi, K.M., G.V. Selby, and R.G. Bryant, Thin-layer composite unimorph ferroelectric driver and sensor properties. Materials Letters, 1998. 35(1-2): p. 39-49. 24. Yoon, K.J., et al., Design and manufacture of a lightweight piezo-composite curved actuator. Smart Materials & Structures, 2002. 11(1): p. 163-168. 25. Wilkie, W.K., et al. Low-cost piezocomposite actuator for structural control applications. in Proceedings of SPIE - The International Society for Optical Engineering. 2000. 26. Bent, A.A., Active Fiber Composites for Structural Actuation, in Aeronautics and Astronautics. 1997, Massachusetts Institute of Technology. 27. Bent, A.A., N.W. Hagood, and J.P. Rodgers, Anisotropic Actuation with Piezoelectric Fiber Composites. Journal of Intelligent Material Systems and Structures, 1995. 6(3): p. 338-349. 28. Wickramasinghe, V.K. and N.W. Hagood, Material characterization of active fiber composites for integral twist-actuated rotor blade application. Smart Materials & Structures, 2004. 13(5): p. 1155-1165. 29. Wickramasinghe, V.K. and N.W. Hagood. Durability characterization of active fiber composite actuators for helicopter rotor blade applications. in Collection of Technical Papers AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference. 2003. 30. Lloyd, J.M., et al. An analytical model of the mechanical properties of the single crystal macro-fiber composite actuator. in Proceedings of SPIE - The International Society for Optical Engineering. 2004. 31. Hagood, N.W. and A. Pizzochero, Residual stiffness and actuation properties of piezoelectric composites: Theory and experiment. Journal of Intelligent Material Systems and Structures, 1997. 8(9): p. 724737. 32. Pini, N., Development and Processing of Novel Active Fibre Composites. 2006, Swiss Federal Institute of Technology Zurich. 33. Yoon, K.J., et al., Actuator performance degradation of Piezo-Composite Actuator LIPCA under cyclic actuation, in Advances in Fracture and Failure Prevention, Pts 1 and 2. 2004, Trans Tech Publications Ltd: Zurich-Uetikon. p. 1331-1336. 34. Kim, S.J. and Q. Jiang, Microcracking and electric fatigue of polycrystalline ferroelectric ceramics. Smart Materials & Structures, 1996. 5(3): p. 321-326. 35. Fang, F. and W. Yang, Poling-enhanced fracture resistance of lead zirconate titanate ferroelectric ceramics. Materials Letters, 2000. 46(2-3): p. 131-135. 36. Ru, C.Q., X. Mao, and M. Epstein, Electric-field induced interfacial cracking in multilayer electrostrictive actuators. Journal of the Mechanics and Physics of Solids, 1998. 46(8): p. 1301-1318.

114

37. Lupascu, D.C. Fatigue in Ferroelectric Actuators. in ACTUATOR 2004, 9th International Conference on New Actuators. 2004. Bremen, Germany. 38. Sekula, M.K., M.L. Wilbur, and W.T. Yeager Jr. Aerodynamic design study of an advanced active twist rotor. in AHS International 4th Decennial Specialists' Conference on Aeromechanics. 2004. 39. Crawley, E.F. and J. Deluis, Use of Piezoelectric Actuators as Elements of Intelligent Structures. Aiaa Journal, 1987. 25(10): p. 1373-1385. 40. Mall, S. and J.M. Coleman, Monotonic and fatigue loading behavior of quasi-isotropic graphite/epoxy laminate embedded with piezoelectric sensor. Smart Materials & Structures, 1998. 7(6): p. 822-832. 41. Paget, C.A. and K. Levin. Structural integrity of composites with embedded piezoelectric ceramic transducer. in Proceedings of SPIE - The International Society for Optical Engineering. 1999. 42. Singh, D.A. and A.J. Vizzini, Structural integrity of composite laminates with interlaces actuators. Smart Materials and Structures, 1994. 3(1): p. 71-79. 43. Shukla, D.R. and A.J. Vizzini, Interlacing for improved performance of laminates with embedded devices. Smart Materials & Structures, 1996. 5(2): p. 225-229. 44. Kim, K.S., M. Breslauer, and G.S. Springer, The Effect of Embedded Sensors on the Strength of Composite Laminates. Journal of Reinforced Plastics and Composites, 1992. 11(8): p. 949-958. 45. Schaaf, K., et al. Mechanical properties of composite materials with integrated embedded sensor networks. in Proceedings of SPIE - The International Society for Optical Engineering. 2005. 46. Mall, S. and T.L. Hsu, Electromechanical fatigue behavior of graphite/epoxy laminate embedded with piezoelectric actuator. Smart Materials & Structures, 2000. 9(1): p. 78-84. 47. Guillon, O., et al., Compressive creep of PZT ceramics: experiments and modelling. Journal of the European Ceramic Society, 2004. 24(9): p. 2547-2552. 48. Kornmann, X. and C. Huber, Microstructure and mechanical properties of PZT fibres. Journal of the European Ceramic Society, 2004. 24(7): p. 1987-1991. 49. Mall, S., Integrity of graphite/epoxy laminate embedded with piezoelectric sensor/actuator under monotonic and fatigue loads. Smart Materials & Structures, 2002. 11(4): p. 527-533. 50. Schnell, A., NONLINEAR CHARGE RELEASE OF PIEZOELECTRIC CERAMICS UNDER UNIAXIAL PRESSURE. Ferroelectrics, 1979. 28(1 /4): p. 347-350. 51. Kim, B.W. and J.A. Nairn, Observations of fiber fracture and interfacial debonding phenomena using the fragmentation test in single fiber composites. Journal of Composite Materials, 2002. 36(15): p. 1825-1858. 52. Tripathi, D. and F.R. Jones, Single fibre fragmentation test for assessing adhesion in fibre reinforced composites. Journal of Materials Science, 1998. 33(1): p. 1-16. 53. Rodgers, J.P., A.A. Bent, and N.W. Hagood. Characterization of interdigitated electrode piezoelectric fiber composites under high electrical and mechanical loading. in Proceedings of SPIE - The International Society for Optical Engineering. 1996. San Diego, CA, USA. 54. McIntire, R.M.a.P., Handbook of Nondestructive Testing. 2 ed. Vol. 5 Acoustic Emission Testing. 1987: American Society for Nondestructive Testing (ASNT). 208. 55. Brunner, A.J., R. Nordstrom, and P. Flueler, Fracture phenomena characterization in FRP-composites by acoustic emission. European conference on macromolecular physics - surfaces and interfaces in polymers and composites, Lausanne 1997, 1997: p. 84-85.

115

56. Tan, P. and L.Y. Tong, A one-dimensional model for non-linear behaviour of piezoelectric composite materials. Composite Structures, 2002. 58(4): p. 551-561. 57. Ye, R.Q. and L.H. He, Electric field and stresses concentrations at the edge of parallel electrodes in piezoelectric ceramics. International Journal of Solids and Structures, 2001. 38(38-39): p. 6941-6951. 58. Belloli, A., et al. Modeling and characterization of active fiber composites. in Proceedings of SPIE The International Society for Optical Engineering. 2004. 59. Webber, K.G., R. Zuo, and C.S. Lynch. Micromechanical modeling of PMN-32%PT ceramic based on single crystal properties. in Proceedings of SPIE - The International Society for Optical Engineering. 2006. San Diego, CA. 60. Hwang, S.C., C.S. Lynch, and R.M. McMeeking, Ferroelectric/ferroelastic interactions and a polarization switching model. Acta metallurgica et materialia, 1995. 43(5): p. 2073-2084. 61. Chen, W. and C.S. Lynch, A micro-electro-mechanical model for polarization switching of ferroelectric materials. Acta Materialia, 1998. 46(15): p. 5303-5311. 62. Paradies, R. Berechnung piezoelektrischer Festköperaktoren vom Typ AFC/MFC. in NAFEMS Seminar: Numerical simulation of elctromechanical systems. 2005. Wiesbaden. 63. Schläpfer, B., Grundlagen für Festigkeitsuntersuchungen an AFC. Institute for Mechanical Systems ETH Zurich, 2006(06-098). 64. Belloli, A., O. Thomaschewski, and P. Ermanni. Optimum placement of piezoelectric ceramic modules for vibration suppression of highly constrained structures. in Proceedings of the ASME International Design Engineering Technical Conferences and Computers and Information in Engineering Conference - DETC2005. 2005. Long Beach, CA. 65. I. M. Daniel, O.I., Engineering Mechanics of Composite Materials. 2 ed. 2005, New York: Oxford University Press. 66. Soden, P.D., M.J. Hinton, and A.S. Kaddour, A comparison of the predictive capabilities of current failure theories for composite laminates. Composites Science and Technology, 1998. 58(7): p. 1225-1254. 67. Hinton, M.J., A.S. Kaddour, and P.D. Soden, A comparison of the predictive capabilities of current failure theories for composite laminates, judged against experimental evidence. Composites Science and Technology, 2002. 62(12-13 SPECIAL ISSUE): p. 1725-1797. 68. Tsai, S.W. and E.M. Wu, GENERAL THEORY OF STRENGTH FOR ANISOTROPIC MATERIALS. Journal of Composite Materials, 1971. 5: p. 58-80. 69. Kuraishi, A., S.W. Tsai, and K.K.S. Liu, A progressive quadratic failure criterion, part B. Composites Science and Technology, 2002. 62(12-13 SPECIAL ISSUE): p. 1683-1695. 70.

Puck, A., Festigkeitsanalyse von Faser-Matrix-Laminaten. 1996, Wien: Carl Hanser.

71. Zemcik, R., et al., High performance 4node shell element with piezoelectric coupling. Proceedings of II Eccomas Thematic Conference on Smart Structures and Materials, 2005. 72. Melnykowycz, M., et al., Performance of integrated active fiber composites in fiber reinforced epoxy laminates. Smart Materials & Structures, 2006. 15(1): p. 204-212. 73. Wang, X., et al., Designing for piezoelectric ceramic wafers bonded on structures using force transfer criteria. Smart Materials and Structures, 2000. 9(2): p. 157-162.

116

Appendix Table A.1 GFRP laminate tensile results.

Cutout

Inserted

Maximum Force (kN) Reference Center Off Center Far Off Center

Mean 54.09 50.83 46.88 40.43

Reference Center Off Center Far Off Center Symmetric AFC Center AFC Far Off Center

59.89 54.57 48.96 45.37 40.45 57.21 53.93

StDev 0.94 1.22 -6.02 3.21 -13.33 3.35 -25.24 Δ (%)

-8.89 -18.26 -24.26 -32.46 -4.48 -9.96

3.78 2.05 3.16 2.03 1.25 2.24 2.05

Young’s Modulus (GPa) (%) 1.74 2.41 6.84 8.29 6.32 3.75 6.48 4.9 3.09 3.91 3.8

Mean 25.57 26.36 26.36 28.11

Δ (%)

StDev 0.2 0.37 3.09 0.2 2.57 0.15 9.91

25.15 25.69 25.07 26.5 28.2 22.71 23.05

2.15 -0.29 5.38 12.13 -9.86 -8.35

0.3 0.51 0.33 0.51 0.3 0.51 0.28

(%) 0.78 1.42 0.75 0.55 1.19 1.99 1.32 1.93 1.07 2.27 1.22

Table A.2 CFRP laminate tensile results.

Interlaced

Inserted

Maximum Force (kN)

Young’s Modulus (GPa)

Δ (%)

Reference

Mean 146.44

StDev (%) 4.47 3.05

Mean 70.88

Center

144.57

-1.28

2.19

1.52

68.01

Off Center

144.57

-1.28

3.63

2.51

69.3

Far Off Center

137.96

-5.97

4.95

3.59

67.58

AFC Far Off Center

135.22

-7.66

3.07

2.27

67.75

Center

143.07

-2.3

3.24

2.26

69.31

Off Center 1 Off Center 2

142.17 136.8

-2.92

4.73

-6.58

6.72 6.64

4.85

70.46 72.54

AFC Off Center 1

139.58

-4.69

3.82

2.74

69.79

Δ (%) 4.05 2.23 4.65 4.67 2.21 0.59 2.35 1.53

StDev (%) 3.77 5.32 1.95

2.86

3.02

4.36

2.74

4.06

2.7

0.98

1.42

2.12 1.71

3.01

3.3

4.73

2.35

117

Cutout

Inserted

Table A.3 GFRP model results. FPF Δ FPF Reference 58.4 Center 50.8 -13 Off Center 47.1 -19.3 Far Off Center 39.6 -32.3

Center Off Center Far Off Center Symmetric

51.3 48.7 41.4 40.4

-12.2 -16.6 -29.1 -30.8

Cutout

Inserted

Table A.4 CFRP model results. FPF Δ FPF Reference 56.1 Center 49.9 -11.1 Off Center 47.9 -14.6 Far Off Center 45.2 -19.4

118

Center Off Center Far Off Center Symmetric

55.1 54.1 51.9 53.7

-1.8 -3.6 -7.5 -4.3

Deflection (μm) 9.4 15.6 6.1 11.4 30.9

Deflection (μm) 1.6 2.3 1.8 2.4 6.3

Curriculum Vitae Personal Data Mark Melnykowycz 23308 Old Orchard Tr. Bingham Farms, MI 48025 USA mark.melnykowycz@klugmat.org Nationality: United States

Education Master of Science, Materials Science and Engineering, Dec., 2003 Michigan State University, East Lansing, Michigan MS Project Title: Mechanical Property Dependence on Hydroxyapatite Scaffold Porosity Advisor: Dr. Melissa M. Baumann Bachelor of Science, Mechanics, Dec., 2001 Michigan State University, East Lansing, Michigan

Experience Visiting Researcher Tokyo Institute of Technology Todoroki Laboratory, Tokyo, Japan, September-November 2007 Development of carbon fiber reinforced plastic materials for use as load-bearing antennas for wireless as well as structural health monitoring applications. Doktorand EMPA Materials Science and Technology, Mechanical Systems Engineering Laboratory, Duebendorf, Switzerland, July 2004 â&#x20AC;&#x201C; April 2008 Characterization of the reliability of Active Fiber Composite (AFC) sensor/actuator elements for smart materials applications. Research Intern Composites Laboratory, EMPA, Duebendorf, Switzerland, August 2003 March 2004 Characterization of AFC integrated into glass fiber laminates. Research Assistant Biomaterials Laboratory, Department of Chemical Engineering and Materials Science, Michigan State University, East Lansing, Michigan, January 2002 December 2002 Mechanical strength characterization of porous Hydroxyapatite (HA) foam scaffolds for tissue engineering applications. Engineering Intern Hartwick Professionals Inc., Troy, Michigan, June 2000 - August 2000, June 2001 - August 2001 Finite element modeling of automotive components for static and non-linear crash analysis.

119

Publications M. Melnykowycz, X. Kornmann, C. Huber, M. Barbezat, A.J. Brunner “Performance of Integrated Active Fiber Composites in Fiber Reinforced Epoxy Laminates” Smart Materials and Structures Vol. 15 pp. 204-212

R. Paradies and Mark Melnykowycz “Numerical Stress Investigation for Piezoelectric Elements with a Circular Cross Section and Interdigitated Electrodes” Journal of Intelligent Material Systems and Structures 2007;18 963-972 Papers to be Submitted

R. Paradies, M. Melnykowycz “Influence of the material properties on the state of stress in piezoelectric elements with interdigitated electrodes” M. Melnykowycz, A. Belloli, P. Ermanni, R. Paradies, “Integrity characterization of smart composite laminates integrated with PZT devices” Composites Science and Technology M. Melnykowycz, A.J. Brunner, “Packaging of AFC sensors for improved strain performance” Smart Materials and Structures

Oral Presentations M. Melnykowycz, X. Kornmann, C. Huber, M. Barbezat, A.J. Brunner “Integration of Active Fiber Composite (AFC) Sensors/Actuators into Glass/Epoxy Laminates” SPIE 2005 March 710th SPIE International Symposia Smart Structures & Materials/NDE Proceedings 12th SPIE, Vol. 5761 M. Melnykowycz, M. Martinez, F. Nitzsche, M. Barbezat, A. Artemev “Active Airfoil Design and Finite Element Analysis of Smart Structures for Rotor Blade Applications” ICAST 2005 Oct. 9-12th 2005 16TH INTERNATIONAL CONFERENCE ON ADAPTIVE STRUCTURES AND TECHNOLOGIES; Paris, France M. Melnykowycz, A. Belloli, P. Ermanni, M. Barbezat “Integration and Reliability of Active Fiber Composite (AFC) Sensors/Actuators in Carbon/Epoxy Laminates” SPIE 2006 Feb. 26th-March 2nd 2006 SPIE International Symposia Smart Structures & Materials/NDE Proceedings 13th SPIE, Vol. 6170 M. Melnykowycz, R. Paradies “Influence of Electrodes on Active Fiber Composite (AFC) Reliability” Sixth Korea-Japan Joint Symposium on Composite Materials, POSTECH, Pohang, Korea, Oct. 31st - Nov. 1st, 2007. M. Melnykowycz, A. Belloli, P. Ermanni, R. Paradies, “Investigation of smart composite laminates embedded with PZT-based modules” SEICO 08 SAMPE Europe 29th International Conference and Forum, Paris, France March 31st-April 2nd 2008. Invited Presentations

120

â&#x20AC;&#x153;Active Airfoil Design and Finite Element Analysis of Smart Structures for Rotor Blade Applicationsâ&#x20AC;? Presented at the Dept. of Mechanical and Aerospace Engineering at Carleton Univ. Ottawa, Canada Oct. 26th 2005

121