<%BANNER%>

Method for Measurement of Residual Stress and Coefficient of Thermal Expansion of Laminated Composites


PAGE 1

METHOD FOR MEASUREMENT OF RESID UAL STRESS AND COEFFICIENT OF THERMAL EXPANSION OF LAMINATED COMPOSITES By DONALD G. MYERS II A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2004

PAGE 2

Copyright 2004 by Donald G. Myers II

PAGE 3

This thesis is dedicated to my family, love d ones, and in the memory of my father.

PAGE 4

ACKNOWLEDGMENTS I would like to thank my advisor, Dr. Peter Ifju, for all of his support and advice. I would also like to thank my other committee members, Dr. Raphael Haftka and Dr. Bhavani Sankar, for their advice. Thomas Singer, William Schulz, and Lucian Speriatu were fellow assistants in the Experimental Stress Analysis Lab that provided invaluable assistance in my efforts. I would also like to extend a thank you to Dr. Leishan Chen for all of his assistance. Together all of these people along with my various professors have made my time at the University of Florida a wonderful educational experience. Finally I would like to say thank you to my entire family and loved ones for everything they have done for me throughout my life. iv

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...........................................................................................................viii LIST OF FIGURES...........................................................................................................ix ABSTRACT......................................................................................................................xii CHAPTER 1 INTRODUCTION........................................................................................................1 1.1 Historical Background............................................................................................1 1.2 Failure of the X-33 Fuel Tank................................................................................2 1.3 Residual Strains and Stresses..................................................................................6 1.4 Structural Reliability...............................................................................................7 1.5 Research Objectives................................................................................................9 2 RESIDUAL STRESS CHARACTERIZATION TECHNIQUES..............................10 2.1 Introduction...........................................................................................................10 2.2 Destructive Testing Techniques...........................................................................11 2.2.1 Hole-Drilling Method.................................................................................11 2.2.2 Cutting Method...........................................................................................12 2.2.3 Ply Sectioning Method...............................................................................12 2.2.4 First Ply Failure..........................................................................................13 2.3 Non-Destructive Techniques................................................................................13 2.3.1 Warpage......................................................................................................13 2.3.2 Embedded Strain Gages.............................................................................14 2.3.3 X-ray Diffraction........................................................................................15 2.3.4 Electrical Resistance...................................................................................15 2.3.5 Cure Referencing Method..........................................................................15 3 CURE REFERENCING METHOD...........................................................................17 3.1 Selected Laminates...............................................................................................17 v

PAGE 6

3.2 Cure Referencing Method.....................................................................................18 3.3 Moir Interferometry............................................................................................19 3.4 Diffraction Gratings..............................................................................................20 3.5 Curing Procedure..................................................................................................22 3.6 Moir Measurement Technique............................................................................27 3.7 Data Analysis........................................................................................................29 3.8 Error Sources........................................................................................................30 3.9 Conclusions...........................................................................................................32 4 STRAIN GAGE METHODS.....................................................................................34 4.1 Acquisition Theory...............................................................................................34 4.2 Specimen Preparation...........................................................................................35 4.3 Testing Procedure.................................................................................................38 4.3.1 Cryogenic Tests..........................................................................................39 4.3.2 Elevated Temperature.................................................................................42 4.4 Data Analysis and Results....................................................................................44 4.5 Error Sources........................................................................................................47 4.6 Conclusions...........................................................................................................48 5 MATERIAL PROPERTY TESTING.........................................................................51 5.1 E 1 Testing..............................................................................................................51 5.2 E 2 Testing..............................................................................................................52 5.3 G 12 Testing............................................................................................................54 5.4 Data Analysis and Results....................................................................................55 5.5 Micro-mechanical Methods..................................................................................58 5.6 Error Sources........................................................................................................59 5.7 Conclusions...........................................................................................................59 6 PLY LEVEL RESIDUAL STRAINS AND STRESSES...........................................61 6.1 Residual Strain Calculation..................................................................................61 6.2 Residual Stress......................................................................................................62 6.3 Conclusions...........................................................................................................64 7 CLASSICAL LAMINATION THEORY...................................................................65 7.1 Theory Development............................................................................................65 7.2 Experimental vs. Analytical Results.....................................................................72 7.3 Conclusions...........................................................................................................76 vi

PAGE 7

APPENDIX A CRM IMAGES OF SPECIMENS..............................................................................77 B MATLAB CLT CODE...............................................................................................85 LIST OF REFERENCES...................................................................................................98 BIOGRAPHICAL SKETCH...........................................................................................103 vii

PAGE 8

LIST OF TABLES Table page 3-1 Total surface strain on unidirectional laminates at 24C.......................................30 3-2 Total surface strain on optimized angle ply at 24C..............................................30 3-3 Total surface strain on RLV laminate at 24C.......................................................31 4-1 Average COV across temperature range................................................................49 4-2 Unidirectional chemical shrinkage induced strain.................................................50 4-3 OAP chemical shrinkage induced strain................................................................50 4-4 RLV chemical shrinkage induced strain................................................................50 5-1 Results of each tension test....................................................................................58 5-2 Comparison of experimental measurements and micro-mechanics values...........59 viii

PAGE 9

LIST OF FIGURES Figure page 3-1 Schematic of a four-beam moir interferometer....................................................20 3-2 Image of moir interferometer used in testing.......................................................20 3-3 Replication schematic............................................................................................23 3-4 Tool separation device...........................................................................................24 3-5 Teflon coated epoxy film casting device...............................................................24 3-6 Autoclave oven......................................................................................................26 3-7 Vacuum bag assembly schematic..........................................................................26 3-8 IM7/977-2 cure cycle.............................................................................................28 3-9 Moir grating on specimen....................................................................................28 4-1 Schematic of strain gage alignment.......................................................................37 4-2 Gauging of sectioned specimen.............................................................................37 4-3 Gauging of reference materials..............................................................................38 4-4 Environmental chamber and LN 2 ...........................................................................40 4-5 Laminate coupons wired and placed inside environmental chamber....................40 4-6 Quarter-bridge schematic.......................................................................................41 4-7 Unidirectional laminate surface strain...................................................................45 4-8 OAP laminate surface strain..................................................................................46 4-9 RLV laminate surface strain..................................................................................46 4-10 Average of laminate surface strains.......................................................................49 4-11 Coefficient of thermal expansion...........................................................................50 ix

PAGE 10

5-1 E 1 and E 2 testing fixture.........................................................................................53 5-2 Shear test specimen schematic...............................................................................56 5-3 Shear Testing Fixture.............................................................................................56 5-4 E 1 stress/strain curves............................................................................................57 5-5 E 2 stress/strain curves............................................................................................57 5-6 G 12 stress/strain curves...........................................................................................58 6-1 Residual strain on selected plies............................................................................63 6-2 Residual stress level on selected plies...................................................................63 7-1 Laminate coordinate system..................................................................................67 7-2 Layer schematic and nomenclature........................................................................67 7-3 Unidirectional CLT and experimental results........................................................73 7-4 OAP CLT and experimental results.......................................................................73 7-5 RLV CLT and experimental results.......................................................................74 7-6 Unidirectional CLT shifted for chemical shrinkage..............................................74 7-7 OAP CLT shifted for chemical shrinkage.............................................................75 7-8 RLV CLT shifted for chemical shrinkage.............................................................75 A-1 Unidirectional Laminate Run 1..............................................................................77 A-2 Unidirectional Laminate Run 2..............................................................................78 A-3 Unidirectional Laminate Run 3..............................................................................78 A-4 Unidirectional Laminate Run 4..............................................................................79 A-5 Unidirectional Laminate Run 5..............................................................................79 A-6 OAP Laminate Run 1.............................................................................................80 A-7 OAP Laminate Run 2.............................................................................................80 A-8 OAP Laminate Run 3.............................................................................................81 A-9 OAP Laminate Run 4.............................................................................................81 x

PAGE 11

A-10 OAP Laminate Run 5.............................................................................................82 A-11 RLV laminate Run 1..............................................................................................82 A-12 RLV Laminate Run 2.............................................................................................83 A-13 RLV Laminate Run 3.............................................................................................83 A-14 RLV Laminate Run 4.............................................................................................84 xi

PAGE 12

Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science METHOD FOR MEASUREMENT OF RESIDUAL STRESS AND COEFFICIENT OF THERMAL EXPANSION OF LAMINATED COMPOSITES By Donald G. Myers II May 2004 Chair: Peter G. Ifju Major Department: Mechanical and Aerospace Engineering The development of residual stresses in laminated composites is a very important area of extensive research. Residual stress can dramatically reduce the strength of a component built of a laminated composite. The existence of these residuals stresses and strength degradation can lead to catastrophic material failures as seen in the X-33. A method has been proposed to measure and quantify these stresses as well as determine the thermal expansion coefficient of laminated composites as a function of temperature. This method relies on the combination of existing techniques in a manner that will allow for the acquisition of stress values across a wide range of temperatures. The first technique employed is a novel strain measurement method know as the Cure Referencing Method. This technique is based on transferring a moir grating onto a specimen during the cure process. The grating becomes a part of the specimen at cure temperature during the polymerization of the resin and therefore carries all strain information from the curing process including temperature dependent strains as well as chemical shrinkage xii

PAGE 13

information. The tool with which the grating was transferred to the specimen is used as a reference from which all relative deformations are calculated. This process is implemented on laminates of three differing stacking sequences, each of which provides its own unique data point. The results from this method provide data points from which the next technique references the strains and stresses as a function of temperature. After strain measurements are taken from the moir specimen, they are fitted with strain gages in the x and y directions. Strain values are observed as a function of temperature as the specimens are exposed to temperatures between cure and cryogenic conditions. The strain gage readings are compensated for temperature effects and are referenced to the strain values at room temperature given by the cure referencing method. The resulting values of strain represent the total surface strain at any given temperature for the various laminates. The surface strain on the unidirectional laminate represents the free expansion state any lamina would reach if not constrained. This value is used to determine residual strain present in different layers of the multidirectional laminates. The coefficient of thermal expansion is also calculated as a function of temperature from the strain gage readings. Basic material testing is also conducted and the values of longitudinal, transverse, and shear moduli are presented. Variability in these measurements is also reported in conjunction with other work at the University of Florida to perform reliability based optimization. Once the residual strain and material properties are calculated they are transformed into stress by the usual stress-strain relations. The final ply level residual stress is plotted as a function of temperature. The results of surface strain measurements are compared to classical lamination theory analytical predictions. xiii

PAGE 14

CHAPTER 1 INTRODUCTION Throughout history one of the primary goals of engineers has been to develop structures and materials that are lighter, stronger, and tougher than those that already exist. The introduction of advanced structural composites (ASCs) has helped to achieve that goal. Composite materials are those that consist of two or more separate materials that are combined macroscopically into a structural unit [13]. The prevalence of these composite materials has skyrocketed since the 1960s and are now used in structures as advanced as satellites, space structures, and military aircraft and in everyday sporting goods around the house such as skis and golf clubs. 1.1 Historical Background Despite the boom in significance since the development of ASCs in the 1960s, basic man-made composite materials have been around for thousands of years. Some examples can be seen as early as the times of the great Egyptian kingdoms when brick was made of clay and straw, and in South and Central America where the people used plant fibers in native pottery [13]. Both of these are examples of one of the most predominant composite materials, fiber reinforced composites. Other examples of fiber-reinforced composites include glass/epoxy, steel-reinforced concrete, and graphite/epoxy, the subject of this investigation. The motivation for the use of fiber-reinforcement in materials comes from research by Griffith published in 1920 that demonstrated that some materials are stronger and 1

PAGE 15

2 stiffer as fibers than in bulk form [14]. Some reasons for this increase in strength include less material flaws, orientation of molecular chains, and lower dislocation densities. Along with all the obvious advantages, advanced fiber-reinforced composite materials have their shortcomings as well. A primary performance weakness is that most of the strength exists in the direction of the fiber and the material performs poorly in transverse directions. Transverse reinforcement is necessary to make the material useful and this is done usually by orienting the fibers at various angles. Using composite materials also poses other problems for designers. These materials are much more difficult to analyze and characterize. Many of the design and analysis techniques that have been used for years on other materials cannot be applied directly to composite materials. Because they consist of more than one material they are often orthotropic in nature. Also, once several layers of lamina are assembled into a laminate, to correct the transverse weakness or tailored to achieve desired directional strengths, a complex interaction between layers develops. This is a macroscopic issue while on the microscopic level there is also a complex interaction that takes place within each lamina between the fiber and matrix material. Due to these analysis difficulties, the behavior of composite materials is not fully understood and this lack of understanding can lead to devastating mechanical failures. It is the goal of this paper to detail a combination of methods that can be used to characterize a material system and then provide a detailed set of data and information on the IM7/977-2 material system. 1.2 Failure of the X-33 Fuel Tank For several years, efforts have been underway to develop a reusable launch vehicle (RLV) to replace the aging space shuttle fleet. The goal is to develop a vehicle that is

PAGE 16

3 entirely self-contained without the need of external fuel and propulsion components such as the external fuel tank and solid rocket boosters seen on the current space shuttles. The vehicle is to be able to takeoff, perform its mission in space, and return to earth entirely under its own power. The development of the X-33 was an attempt to achieve these goals. The design of the X-33 called for an internal liquid hydrogen (LH 2 ) fuel tank [12]. This tank was to be an integrated part of the structure designed to provide both fuel storage and carry structural loads. One of the most important factors in the design of an RLV is to reduce the takeoff weight as much as possible. Manufacturing a tank consisting of ASCs was seen as the best way to do this. The X-33s fuel tank and aft fuselage were designed to consist of two LH 2 fuel tanks, with each composed of several lobes made of graphite/epoxy and having bonded woven composite joints. Each lobe was a sandwich panel with face sheets consisting of the IM7/977-2 material system and a Korex core. The tanks were to store the LH 2 fuel as well as be part of the primary aft structure designed to distribute thrust, inertial, landing gear, and aero-control surface loads. The first of two tanks was designed and delivered by Lockheed Martin to NASA at the Marshall Space Flight Center for preflight testing. The preflight testing was designed to verify that the tank was able to withstand the cryogenic loading conditions of F (20 K) as well as the expected flight loads. The planned testing sequence consisted of: 1. A partial liquid nitrogen (LN 2 ) fill and proof test to 32 psig, followed by a leak check at 34 psi of gaseous helium 2. A LH 2 fill and structural load application. The test was to consist of six conditions: a proof pressure test at 42 psig and five loading tests at 5 to 42 psig 3. A final 34 psig gaseous helium test

PAGE 17

4 It would be seen soon after testing began that the design was not feasible for its intended purpose. In early November of 1999 the fuel tank began the first combined mechanical loading and LH 2 storage test. It had been previously been filled and leak tested with LN 2 and LH 2 in the absence of external loads. This time the initial pressure and loading tests were successful and the tank was drained of LH 2 Fifteen minutes after the tank was drained a major mechanical failure took place as the outer facesheet and core material debonded from the inner facesheet. The following lists the observations that were made during the test and time leading up to the failure. The first observation was that at the time of fueling there was an expected decrease in the pressure of the sandwich core resulting from the temperature change within the tank. The second observation was an unanticipated increase in the core pressures as the tank neared full capacity and pressurization. A third observation was the continued increase in core pressures as the tank was drained. This event was expected, as the temperature increase would reverse the initial trend seen in observation one and pressure should have returned to ambient conditions. Fourth was the incident of the composite sandwich failure. At the time of the failure core pressures were between 50 and 60 psia. These pressures rapidly dropped to atmospheric as the region failed. Finally, the remaining sections of the core that did not experience a failure had internal pressures above the expected ambient a full 13 hours later. Review and analysis of the data resulted in the following failure sequence. The rise in pressure in the second observation was most likely the result of seepage of H 2 into the core through the inner facesheet. The method of transportation was though mirocracks

PAGE 18

5 that had developed in the inner facesheet due to the stresses and strains cause by the cryogenic environment. As the tank was filled with LH 2 the cooling caused a vacuum in the core, which helped to pull in GH 2 until the area was covered with the liquid. LH 2 was then also able to penetrate the facesheet through the microcracks. As the tank was emptied some of the fluid was able to travel back through the microcracks and be removed from the tank. However, before all of the fluid had passed back into the interior of the tank the pressure and strain decreased closing the microcracks and trapping some of the fluid inside the core. As the temperature continued to climb the pressure of the vaporizing H 2 resulted in a buildup until the outer facesheet and core debonded from the inner facesheet. The properties of the laminates were estimated from existing databases. Also it was assumed that 5000 microstrain () was an adequate allowable level to limit permeability. The above failure observed in the X-33 project is a perfect example of how the behavior of ASCs are not fully understood and how this lack of information can result in catastrophic failures. The design group failed to adequately cover combined thermal/mechanical effects. It is this thermal effect, or processed induced residual effects, that are not fully understood and result in considerable structural issues. The failure of the X-33 and what it means to the future of composite implementation is the driving force behind this thesis. This work will propose combined methods that can be used to better evaluate the IM7/977-2 of the X-33 and other composite material properties as they are subjected to extreme thermal conditions. This material characterization can be used to remove the some of the assumptions that are

PAGE 19

6 made in current composite design and replace them with concrete experimentally determined values. 1.3 Residual Strains and Stresses Root failure of the X-33 was most likely the failure to understand and accurately predict the residual strains and stresses that develop in a composite as a result of curing and the exposure to changing temperatures, especially low temperatures. When a polymeric fiber reinforced composite is manufactured it usually undergoes a process where the resin is heated, the fibers are wetted, and cure takes place at high temperatures [37]. During this process and subsequent conditions residual strains and stresses will develop within the laminate. The developing stresses arise on two different scales. Stresses will occur that develop between the fiber and matrix in each lamina of a composite. The study of these effects is termed micro-mechanics, where a laminate is examined on the fiber scale. Another area of interest is the examination of what happens to a composite on the macro-mechanical or ply scale. It is on this scale that the following research will focus. On the ply scale residual strains and stress develop due to the mismatch in the thermal expansion coefficients of the fiber and matrix as well as the development of chemical shrinkage. As a lamina is cured its matrix constituent undergoes polymerization. Epoxy resins undergo condensation polymerization, where two reacting monomers are brought together to form a new molecule of the desired compound [13]. In the case of many ASCs this is a two-step process. The production of prepreg tape consists of wetting the fibers and allowing the matrix to partially cure. The polymerization process concludes when these prepreg materials are arranged into the desired stacking sequences and then are heated to the desired cure temperature. As this

PAGE 20

7 process takes place the matrix undergoes a volumetric change known as chemical shrinkage while the fibers remain volumetrically unchanged. This event adds to the mismatch in expansion of the fibers and matrix, where the matrix is subject to higher expansion and contraction than the fiber. Residual stresses develop when the desired expansion of the lamina is restricted. As the angles of the lamina are varied from ply to ply, the lower contraction of the fiber will constrain the contraction of the matrix. When temperatures are decreased, as in the case of the X-33, the matrix desires to contract but instead is subjected to tensile stress as this deformation is opposed. If all of the fibers are aligned with each other there is no stress present on the ply scale. A cross-ply [0/90] stacking sequence will result in the highest level of residual stress. The residual stresses that develop during processing and operating conditions are not negligible. They can consume large amounts of a laminates strength. If these stresses are not accurately understood they can lead to drastic material failures as in the case of the X-33, where the tensile stresses that developed in the matrix material likely exceeded its critical tensile strength. Once this occurred, micro-cracks were able to develop allowing the seepage the hydrogen into the core. The development of microcracks in other structures will allow the fibers to be exposed to possibly degrading environmental conditions as well as possible chemical attack in storage facilities. 1.4 Structural Reliability The existence of the overwhelming thermal stresses that resulted in the failure of the X-33 demonstrates the need to better understand the nature of composite materials and appropriately design for their implementation. The examination of structural

PAGE 21

8 reliability of a component during the design process is a step that must be taken to prevent such failures. Previously, the design process worked to build reliability based on a factor-of-safety approach [48]. This is a deterministic approach to the issue of reliability where worst case scenarios are proposed, investigated, and accounted for by adding a safety margin in the design. This approach has been used for years in metallic structures and can provide adequate reliability for composites but not to the same degree. This is because composite material performance exhibits a larger statistical distribution and variation than metals. Reliability-based design methods are being developed to account for this problem. Probabilistic design methods are moving to the forefront in composite design. Probabilistic design is an integrated process that works to define and develop functional relationships between various design factors and material properties. A statistical variation in one variable will have consequences to the others and those changes can be designed for. The probabilistic approach uses statistical characterization instead of proposed worst case values to define a variable and determine its magnitude and frequency. The amount of data and how well the variables are defined will influence the extreme values. A typical probabilistic design approach has the following fundamental elements: 1. Identify all possible uncertain variables on all scales of the structure, including all variables at the constituent scale, all stages of fabrication and assembly, and the possible applied loads. 2. Assign a probabilistic distribution function for each variable. 3. Process all random variables through and analyzer that can handle microand macro-mechanics laminate theories, structural mechanics, and probability theories.

PAGE 22

9 4. Extract useful information from the analysis and compare them to probabilistic design parameters. To be able to carry out the above procedure the following aspects must be characterized as random variables: Material mechanical properties External loads during operation Manufacturing process variability Environmental effects Environmental history during operation Effect of flaws or damage locations Predictive Accuracy The probabilistic approach to analysis and design has become a significant area of academic interest. Qu et al have investigated areas of interest that are particularly of importance to the future of RLV design. They performed reliability-based optimization of composite laminates for cryogenic environments [41]. It was found that under these conditions that the optimized weight was sensitive to transverse lamina strength and thermal stresses as they are related to CTE and lay-up sequence. When the confidence in these values is low the thickness of the laminate is significantly greater leading to a higher vehicle weight. 1.5 Research Objectives The work of Qu et al has been carried on at the University of Florida and parallels the work in this thesis. The goals of this thesis were to establish a simple testing method to determine residual strains, stresses, and coefficient of thermal expansion as a function of temperature, and to measure these properties on the IM7/977-2 material system that was used in the X-33 fuel tank development. In addition to this, various moduli were to be experimentally determined and statistical data was to be supplied to the optimization group for further probabilistic analysis and design.

PAGE 23

CHAPTER 2 RESIDUAL STRESS CHARACTERIZATION TECHNIQUES 2.1 Introduction Since the introduction of advanced structural composites the need to understand their behavior has lead to extensive research in the area of residual stress determination. Many existing methods have been devised to aid in residual stress characterization. Some of these methods have been derived from existing tests used on other materials while others have been developed from scratch. Each of these methods has it own advantages and disadvantages. As a whole they can be useful in characterizing certain aspects of residual stress and its effects. Most do have the same substantial flaw that will be pointed out in this research. None, with the exception of the Cure Referencing Method (CRM), developed at the University of Florida and used extensively in this work, have the ability to account for the residual stress that results from the chemical shrinkage involved in the curing of an ASC. This chapter will give a brief history of and detail the existing traditional methods that are often used to characterize residual stress in laminated composites. Existing methods are usually grouped into two broad groups, those that are destructive in nature and those that are non-destructive. Destructive testing usually involves damaging or removing a section of material from a test specimen. The result is a specimen that can no longer be used for most applications. Non-destructive tests are generally preferred over destructive tests for obvious reasons. A specimen exposed to non-destructive means of testing has not lost any usefulness as a structural component. 10

PAGE 24

11 Tests can also be repeated on the same specimen to ensure accuracy. These methods also often provide a means with which to test the effects of time on a component. 2.2 Destructive Testing Techniques 2.2.1 Hole-Drilling Method One of the oldest and most accepted methods for the determination of residual stresses is that of the hole-drilling technique. Mathar first proposed this technique as early as the middle 1930s [33]. Since that time it has been employed to characterize residual stresses present in metal structures and other isotropic materials. When a hole is introduced into a stressed body the stresses are relaxed and become zero. This results in a change in the surrounding strain field, which can be measured and correlated to the relaxed stresses. This was the heart of Mathars work. The most commonly used application of the hole-drilling techniques involved drilling a small blind hole near the application point of a strain gage rosette [35]. The rosette can then measure the relaxed strain and the data can be used to calculate the residual stress that was present before the creation of the hole. Since the middle to late 1960s efforts have been underway to extend the use of this technique from isotropic to orthotropic materials [1,28,42,43]. The extension of this method into orthotropic materials does not yield results as readily as in isotropic materials. The solution method is numerically intense. Many assumptions must be made in order to simplify the resulting solutions. The highly orthotropic nature of composites also adds further difficulty in the measurements themselves. Obtaining measurement precision around the hole is extremely difficult especially in the fiber direction, even for high precision techniques such as moir interferometry [36].

PAGE 25

12 2.2.2 Cutting Method The cutting method is another technique that relies on the same principles as those in the hole drilling method. Again, stress is relaxed by the removal of material. This time a notch is removed from a specimen resulting in the creation of a free edge. Lee devised a method where a moir interferometry grating was applied to the specimen to record the resulting strain field [30,31]. The resulting strain field was calculated and related to the residual stresses by the used of finite element analysis. This method is not without its faults and has yet to be fully accepted in the experimental community. Niu poses many questions based on the application of the grating [37]. The grating was applied after an edge of the composite had been trimmed. This resulted in some of the residual stresses being released prior to data recording and the creation of a very complex strain field under the surface. The resulting measurements that were made contained data that was based on a stress field different from the original residual stress. A second technique derived to employ the cutting method was proposed by Sunderland et al. [46]. The successive grooving technique involved cutting a groove through the thickness at successive depths. Strain gages were placed opposite the groove and recorded the changing strain field. The residual stress was then calculated from the strain by use of a numerical 2-D model for each layer. 2.2.3 Ply Sectioning Method Yet another destructive technique devised to study residual stresses in laminated composites has been proposed by Joh et al. [24]. This method made use of moir interferometry. The deformation caused by sectioning and releasing a layer from the constraints imposed by an adjacent layer was measured. The resulting release of stress

PAGE 26

13 could be easily calculated from the deformation strains. The problems resulting from the cutting method are resolved by obtaining a small strip specimen from the edge of the larger specimen resulting in a plane stress state. Another manner in which to use ply sectioning involves the nature of warping seen in an unbalanced composite laminate. The outside layers of a laminate can be machined away resulting in the unbalanced, warped structure [3,32]. The resulting warpage is measured and can be used as an input in classical lamination theory to calculate the corresponding residual stresses. 2.2.4 First Ply Failure The first ply failure method makes use of the maximum stress criterion to calculate the existing residual stresses [16,27]. Hahn and Kim recorded strains and loads while loading a cross ply, [0/90], specimen to failure. Elastic assumptions were used to calculate the load at which the first ply failed. This load was compared to the corresponding strength of a unidirectional specimen and the difference was called the residual stress. Kam also attempted to consider the visco-elastic effect on this methodology [25]. 2.3 Non-Destructive Techniques 2.3.1 Warpage Warpage will result in unbalanced, asymmetric multidirectional laminates. The destructive method of Joh et al. made use of this fact in its calculation of stress from deformation strain resulting from sectioning. Asymmetric laminates can also be intentionally manufactured so that the warpage can be measured. This method has been an area of extensive investigation [8,10,17,22,23,26,27,32,51]. CLT provides the usual means by which the stresses are calculated from the observed warpage.

PAGE 27

14 The combination of warpage and CLT has also been used to devise a method to measure and quantify polymer matrix cure shrinkage [8,9]. Daniel manufactured a laminate directly onto an existing and previously cured identical laminate having the same CTE and manufacturing procedure. Any resulting warpage was strictly the result of the chemical shrinkage. The curvature was measured and stresses were again calculated using CLT and the resulting stresses were directly related to chemical shrinkage. 2.3.2 Embedded Strain Gages Embedded strain gages function by becoming an intimate part of the laminate being measured. Any deformation of the laminate is directly seen and measured by the strain gage. The most common strain gages used for this method are the electrical resistance and fiber optic strain gage. The first efforts to measure residual stress in this manner were made by Daniel [5,6,7]. High temperature electrical resistance strain gages were cured directly in the laminate by placing them between the lamina plies. The strain gages were then directly exposed to any deformation of the laminate. A compensation method was derived to account for the varying characteristics of the gage as it went through a given temperature range. A similar method of compensation is used and described in this thesis. The main drawback to the method described by Daniel was the introduction of a foreign body to the laminate. This can drastically alter the material properties at the location as well as create a void that will lead to incorrect measurements. Fiber optics represents the second set of strain gages. This gage seems to be gaining favor in the experimental community. Several recent studies have been conducted with these gages [2,29,47,49]. In this technique a fiber optic is laid or embedded in the composite as if it were one of the fibers. Again there is a concern with

PAGE 28

15 the introduction of a foreign body into the heart of the laminate. If the optical fiber is not on the same size scale as the material fiber then there is potential for significant errors based on the change in the laminate material properties. 2.3.3 X-ray Diffraction The use of X-ray diffraction in determining stresses was proposed by Predecki and Barrett in 1979 [40]. Their work was extended to measure residual stresses by Fenn, Jones, and Wells [11]. With this technique metal particles become part of the matrix material and experience similar deformations. X-ray measurements are made on the embedded metal particles and the resulting stresses are correlated to the residual stresses in the composite. As with the embedding of strain gages, the introduction of metal particles could result in a change in the laminate properties. 2.3.4 Electrical Resistance Graphite and Carbon fibers are electrical conductors and therefore a laminate made from them will have a measure of electrical conductivity and resistance. Researchers are looking into a technique where the electrical resistance of a laminate can be used to measure the development of stresses [38,50]. Wang and Chung have observed the change in resistance across the thickness of a laminate and have set out to correlate the measured resistance with stresses present in the composite. Park, Lee, and Lee set out to achieve cure monitoring and strain-stress sensing based on a similar technique. 2.3.5 Cure Referencing Method The Cure Referencing Method is a technique developed recently by Niu, Ifju, and Kilday to record process induced strains[20,21,37]. CRM uses the full-field laser based optical method of moir interferometry to document strains on the surface of laminates that initiate during the high temperature curing process. The method involves replicating

PAGE 29

16 a high frequency diffraction grating on the specimen while in the autoclave. This method will be key to the research done in this thesis. A detailed description of the application of CRM will follow.

PAGE 30

CHAPTER 3 CURE REFERENCING METHOD Several distinct experimental techniques were used in conjunction with one another to develop residual strain, stress, and coefficient of thermal expansion values across a wide temperature range. The following chapters will describe in detail the processes and methods used in the research. These processes and experimental techniques include material lay-up procedures, basic material property characterization, the Cure Referencing Method (CRM), and basic strain gage techniques. 3.1 Selected Laminates The first step in developing a better understanding for the issues at hand was to manufacture a desired test specimen. Three different composite laminates have been chosen for study. A composite laminate consists of two or more layers or plies of lamina arranged in any stacking sequence. Three different stacking sequences of IM7/977-2 were selected for study based on their relative importance. These laminates were: 1. Unidirectional (UNI): [0 13 ] 2. Reusable Launch Vehicle (RLV): Quasi-isotropic lay-up [45/90 3 /-45/ 3 ] S 3. Optimized Angle Ply (OAP): [()] 3S The unidirectional stacking sequence was selected for its ability to provide insight into some of the material properties of the material system [19]. It will be seen later how the strain values obtained from the unidirectional material are used directly in the analysis to calculate residual strain and stress in the other laminates. The RLV was, as its name implies, taken directly from the stacking sequence selected by Lockheed Martin for the design of the X-33 LH 2 fuel tanks. It was designed 17

PAGE 31

18 as a somewhat quasi-isotropic laminate to allow for maximum strength in all directions. Quasi-isotropic lay-ups are very common in structural applications of ASCs. However, it was not designed to account for the loss in strength that is associated with the development of thermal residual stresses. The lay-up selected as the optimized angle ply was the result of work done in an effort to account for thermal stresses while still maintaining a level of mechanical strength. This effort resulted from a reliability optimization performed by Qu [41], of a LH 2 fuel tank constructed from an IM600/133 graphite/epoxy material system. The stacking sequence for the tank was [ 1 / 2 ] s Optimal designs had ply angles 1 and 2 near 25 o and nearly equal; the stacking sequence was thus simplified to [] s Monte Carlo simulation gave a probability of failure of approximately 1 in 15,000 for the [ o ] s design. It should be noted that this optimized ply sequence was for a different, but similar material system. These laminates were chosen for the study of the residual strain, stress and CTE. Different sequences were used in the basic material property determination and they will be described briefly in a following chapter. 3.2 Cure Referencing Method The CRM is a technique developed at the University of Florida by Dr. Peter Ifju and Xiaokai Niu. It is a non-destructive novel testing method designed to accurately determine process induced and residual strains that are associated with the manufacturing, cure, and thermal loading of an ASC. This method is based on a full-field measurement technique know as moir interferometry in which a diffraction grating is placed on a test piece during the cure cycle. The following section aims to describe CRM and the procedures behind its implementation.

PAGE 32

19 3.3 Moir Interferometry Moir Interferometry is a very precise and sensitive laser based optical technique that allows the development of a contour map of the in-plane displacements [39]. This technique provides a displacement sensitivity of 0.417 m per fringe order. Moir relies on the interference properties of light that results in the development of dark and light bands known as fringes. For this interference to be observed, two diffraction gratings must be compared to one another. One diffraction grating must be replicated to the desired part or specimen that is to be deformed while another grating is to remain unaltered to serve as a reference. This unaltered grating may be one applied to an un-deformed specimen or the master grating that is used to transfer the pattern to specimens as described in the next section. The moir interferometer is tuned with the reference grating. Tuned refers to aligning the interaction of the laser and reference grating in a manner so that the displacement field is nullified, or no fringes are present. This is necessary so that only the deformation of the specimen is recorded by the presence of fringes. The deformation that develops in a specimen can be observed at anytime under any loading condition as long as the reference grating is still intact. In this study the deformation given by moir was only observed for a mechanically unloaded specimen at room temperature. A four-beam moir interferometer, schematically represented in Figure 3-1, was used to record the deformation data. The interferometer used in this experiment was a 2400 line/mm four-beam system seen in Figure 3-2. The interferometer consisted of a 10 milliwatt Helium Neon laser with a 632 nm wavelength and a 108 mm diameter parabolic mirror. The field of view provided was 38.1mm or 1.5 inches in diameter.

PAGE 33

20 Figure 3-1: Schematic of a four-beam moir interferometer Figure 3-2: Image of moir interferometer used in testing 3.4 Diffraction Gratings Application of a diffraction grating to a part during cure is the central idea and procedure behind CRM. Before this grating was applied to a specimen it was carefully

PAGE 34

21 produced in a several step method derived from the process developed by Niu [37] described next. The development of an appropriate diffraction grating to be attached to a specimen followed these procedures and is seen schematically in Figure 3-3: 5. A grating was replicated at room temperature onto Astrositall glass (ULE) using silicone rubber from an original photoresist master grating. The glass had dimension of 3x 4.5x 0.5. The silicone rubber was a two-part mixture (GE RTV 615). The silicone mixture was centrifuged so that all air bubbles could be removed form the mixture. 6. The silicone grating was then placed into a vacuum deposition chamber where a very thin layer of aluminum was deposited onto its surface. This layer was deposited so that another piece of Astrositall could be easily separated from its surface in a later step. 7. The aluminum coated silicone grating was then used to replicate onto the Astrositall autoclave tool. This grating was made of 3501-6 epoxy. Both pieces of Astrositall were placed into a heating/vacuum chamber where they were heated to 177 C and allowed to soak for approximately one hour. A small piece of 3501-6 was pulled from the freezer and placed into a metal tray where it could be heated. A vacuum was applied to the chamber after heating and liquefying the epoxy for 15 minutes so that any trapped gas bubbles were removed from the epoxy. All three components, the two pieces of Astrositall and the heated epoxy, were removed from the oven. Rapidly, the liquid epoxy was poured into a small pool on the clean Astrositall and then the aluminized silicone grating was placed squarely on top. The two pieces of Astrositall were returned to the oven and a small weight was placed on top so that the grating would be very thin. They were cured at 177 C for 10 hours. 8. The autoclave tool was then separated and prepared for use. The autoclave tool was separated from the silicone tool by use of the instrument seen in Figure 3-4. After separating, the autoclave tool was sent into the vacuum deposition chamber, where two layers of aluminum were cast onto the grating. Each layer of aluminum was separated by a thin layer of dilute Kodak Photo Flo solution that was applied between the first and second deposition of aluminum. The orientation of the grating lines on the tool was determined by the use of an alignment device similar to one described by Post et. Al [39]. A straight line was marked on the tool so that the fibers of the composite specimens were easily aligned with the direction of the grating. 9. A thin layer of 3501-6 was then cast onto the autoclave tool directly on top of the grating surface. A Teflon coated tool, seen in 3.5, was then used to apply the thin film of epoxy. As in a previous step some of the components, the autoclave tool

PAGE 35

22 and Teflon coated device, were heated for one hour at 177 C. At several intervals during this hour the Teflon device was removed and the Teflon was tightened to ensure a uniform film of epoxy. After the initial hour, a small piece of 3501-6 was then placed into the metal tray where it was heated for 15 minutes and then vacuumed to remove air bubbles. The components were removed and a thin pool of epoxy was poured onto the grating surface. The Teflon device was lowered into the pool at an angle to push out any possible entrapped air bubbles. The epoxy was cured at 177 C for a period of 10 hours. After cure the autoclave tool was removed from the oven and the Teflon device was disassembled leaving only the sheet of Teflon on the tool. This sheet of Teflon was removed from the surface of the tool by pulling back at a sharp angle. After completion of the diffraction grating preparation, the grating was transferred to the specimen by the following method. 3.5 Curing Procedure As with most advanced composite materials, the laminates in this study were cured in an autoclave (Figure 3-6). The autoclave is a pressurized oven that has vacuum line connections inside the pressurized chamber. After selection of an appropriate laminate the following steps were implemented [18]. It should be noted that the following procedure was specific to the implementation of CRM. Other laminates were produced using the same basic procedure and cure cycle with the only difference being that they are cured without the addition of the ultra-low expansion Astrositall glass and grating application.

PAGE 36

23 Figure 3-3: Replication schematic

PAGE 37

24 Figure 3-4: Tool separation device Figure 3-5: Teflon coated epoxy film casting device

PAGE 38

25 10. Lay-up: Individual sheets were cut to size from a roll of prepreg. Those sheets were then arranged and stacked in the proper sequence paying heavy attention to maintaining the correct angles. 11. Vacuum bagging: Figure 3-7 shows a schematic of the vacuum bag assembly. The ULE tool with the grating side up was placed at the base of the assembly. The surface was then covered with a non-porous release film. A hole of approximately 1.5 diameter was cut into the release film. The hole allows for the grating to be exposed to the composite surface during the cure cycle. The thickness of the film was important. It was critical that it was as minimal as possible so that it did not indent the surface of the composite during the cure process as the resin began to polymerize. The prepreg laminate was then placed on top of the release film, centered over the ULE tool. The non-porous release film placed between the tool and the laminate prevented the epoxy from curing to the tool. A layer of porous release film was placed on the upper surface of the laminate, which provided two functions. It allowed for excess resin to be drawn off the laminate and into the bleeder cloth, as well as facilitated the separation from the bleeder cloth. The bleeder cloth was a porous material that absorbs excess resin drawn out of the laminate via vacuum. Absorbing this excess resin allowed for an appropriate fiber volume fraction to be achieved. The breather cloth, which was the same material as the bleeder cloth, was used to distribute the vacuum throughout the entire bag. Another layer of non-porous release film was placed between the bleeder and the breather cloths so that excess resin was not transferred into the breather material. This could have resulted in resin being drawn into the vacuum line or clogging the breather cloth which then would have prevented an even vacuum distribution on the laminate during cure. Finally the ULE and laminate assembly were placed into a vacuum bag. The bag was sealed with a sealant tape and placed in the autoclave for curing. 12. Cure preparation: The vacuum bag was placed inside the oven and connected to the vacuum line. Vacuum was applied and a leak check was performed. It was imperative that the vacuum integrity was maintained throughout the necessary portions of the curing process. 13. Cure: Execute the cure cycle appropriate for the given material. The cure cycle for the IM7/977-2 can be seen in 3-8.

PAGE 39

26 Figure 3-6: Autoclave oven Figure 3-7: Vacuum bag assembly schematic

PAGE 40

27 3.6 Moir Measurement Technique After the specimens have gone through the cure cycle the reflective grating was present on the surface (Figure 3-9). The specimens were then returned to the moir interferometer where the deformation measurements were taken. There are many sources of potential error in this technique if the proper steps were not taken to insure accurate measurements. The first step that must be taken was to properly tune the moir interferometer. The virtual grating must be tuned accurately to 2400 lines/mm in both the horizontal and vertical field, U and V respectively. This was done by directing the second order diffraction of the reference grating back to the fiber optic tip for all adjustable mirrors. Then fine-tuning was done for each field by adjusting the U and V field mirrors until both displayed null fields. The exact position of the master grating was noted so that the specimen was also placed at the exact same position. Failing to do so would have introduced errors due to the imperfection in the collimated laser beam. On each specimen a circle of know radius was scribed into the reflective grating. This established the gage length over which the fringe numbers would be counted. Each specimen was then placed into the tuned interferometer with a rotation at 90 from the reference grating. This rotation was needed because the grating surfaces of the specimens were the mirror image of the reference grating. The resulting fringes were photographed using Polaroid ISO 400/27 black and white instant sheet film. Images were taken and recorded from all of the different laminate specimens at room temperature.

PAGE 41

28 Cure Cycle-500501001502002503003504000100200300400500600MinutesValues Vac Temperature (F) Pressure (psi) Figure 3-8: IM7/977-2 cure cycle Figure 3-9: Moir grating on specimen

PAGE 42

29 3.7 Data Analysis The experimental technique and methods described previously resulted in the acquisition of the very characteristic pattern of light and dark fringes of moir interferometry. These fringe patterns were used to directly determine the in-plane displacements and strains. The relationship between fringe order N and displacement was as follows: fNUx (3.1) fNVy (3.2) where f was the frequency of the virtual grating; f=2400 lines/mm. The absolute surface strain was also determined from the fringe pattern analysis by the following relations: 11xxNU xN x fxfx (3.3) 11yyNNV y x fyfy (3.4) These equations were used to determine the strain on the surface of both the unidirectional and multidirectional laminates. These strains were a result of the thermal contraction, which was a function of the temperature difference between the cure temperature and operating temperature; chemical shrinkage of the matrix, which was not temperature dependent, occurring only during cure; and the thermal expansion of the autoclave tool, which was neglected in the analysis [19]. The resulting characteristic images of each laminate stacking sequence are displayed in Appendix A and the corresponding values of the surface strain given by

PAGE 43

30 CRM can be seen in Tables 3-1 through 3-3. The values seen in the table represent the values of surface strain that were present at room temperature. These measurements were extended throughout the range from cure to LN 2 temperatures by a method to be discussed next. 3.8 Error Sources Several sources of error were potentially present during the implementation of CRM. These sources included the improper implementation of moir interferometry, manufacturing defects in the laminate specimens, and misalignment of CRM grating surface. Without taking precautions to avoid possible error sources, the indicated strains of the CRM method could be significantly misleading. Table 3-1: Total surface strain on unidirectional laminates at 24C Specimen x-direction y-direction UNI-13-06-04 0 -6890 UNI-13-02-03 0 -6500 UNI-13-03-02 0 -7220 UNI-13-B2-02 0 -6955 UNI-13-02-05 0 -6824 Average 0 -6877.8 COV ---3.77% Table 3-2: Total surface strain on optimized angle ply at 24C Specimen x-direction y-direction OAP-12-B3-01 787 -4920 OAP-12-03-02 820 -5250 OAP-16-04-01 919 -4950 OAP-16-02-02 1030 -5510 OAP-20-07-05 1033 -5250 Average 917.8 -5176 COV 12.5% 4.72%

PAGE 44

31 Table 3-3: Total surface strain on RLV laminate at 24C Specimen x-direction y-direction RLV-A1-01 -610 -190 RLV-02-02 -525 -197 RLV-03-03 -525 -220 RLV-05-04 -590 -230 Average -562.5 -209.25 COV 6.78% 7.80% The implementation of moir interferometry has many possible sources of error itself. Improper tuning of the interferometer, misalignment of specimens, improper specimen orientation and placement, and vibration can all lead to potential problems. Fortunately, these issues are somewhat easily accounted for and all the proper steps were taken to avoid improper strain reading as a result of these problems. Other sources of errors were not as easy to account for. Manufacturing defects are a common event in laminate production. Inclusions of foreign matter can often lead to incorrect results. The existence of these inclusions on a large scale is easy to detect while testing due to the large deviations they can cause from expected or previous results. Manufacturing the laminates in a clean working area and examining each layer prior to joining eliminated this problem. One source of error that was not ruled out was misalignment of the layers themselves during manufacturing. All laminates were cut and aligned as carefully as possible by hand but the element of human error still existed. It was possible that misalignments of 5.2 through the thickness developed. This misalignment can result in a change in the characteristics of the laminate. Similar to the alignment of the fibers during manufacturing, the autoclave tool and the grating on its surface was aligned by hand onto the specimen. The greatest care was taken to apply the grating as accurately as possible, but a possible slight offset from the desired axis of measurement may have resulted. This offset would have lead to the

PAGE 45

32 measurement of strain in a direction different than what was desired. For example, if the strain in the fiber direction after cure is desired and the grating was not aligned perfectly in the x, or fiber direction, as in the case of the unidirectional laminate, the measured displacement would be higher than the true displacement. A very important aspect of accuracy was the ability to read the fringe order information. From equations (3.3) and (3.4) it is seen that any inaccuracy in the reading of the correct fringe order would have lead directly to measurement errors. Different fringe pattern densities often required adjustments on gage factor as the images must be magnified to see the fringes. Different magnifications assured that all patterns could be read to less than one fringe order. In the case where the gage length was 0.5, a misread of one fringe would have resulted in an error of approximately 33 In the case of a 0.25 gage length a misread of one fringe would have led to double the previous, up to 66 3.9 Conclusions The Cure Referencing Method has been applied to find the total surface displacement on various laminate stacking sequences. The calculated surface strains represent the sum of the contraction from thermal effects as well as chemical shrinkage that resulted from polymerization. The strains have been averaged and the coefficient of variation has been calculated for each stacking sequences. Any COV measurements will be provided to the optimization group for future reliability analysis. The experimental results were relatively repeatable. All of the COVs were quite low with the exception of the x-direction OAP values. This indicated quite consistent and reliable results.

PAGE 46

33 The unidirectional laminates showed the highest value of surface strain, followed by the OAP, and finally the RLV laminates. The high strain level indicated that the unidirectional and OAP laminates were relatively free to contract. Conversely, the RLV laminate showed a very low strain value which indicated a high level of constraint. The data from these tests are revisited in the combination of CRM and strain gage results in Chapter 4.

PAGE 47

CHAPTER 4 STRAIN GAGE METHODS The second technique of experimentation that was used in this study was that of basic thermal expansion testing methods conducted with modern strain gage technology. This method was used to gather real time surface strain data as the temperature was varied over the test range. Later it will be shown how information gathered with this testing procedure was used in conjunction with CRM and material property testing to develop residual strains, stresses, and CTE as a function of temperature. 4.1 Acquisition Theory When strain gages are used to acquire strain as a result of temperature change the output cannot be treated as the true strain on the surface. The output of the gage is actually termed the thermal output. It is caused by two factors, the change in the resistivity of the grid alloy with change in temperature and the difference in thermal expansion coefficients between the gage and the test material [34]. The resulting resistance change is the sum of resistivity and differential expansion effects are TF R RGGSG (4.1) where R/R is unit resistance change, G is the thermal coefficient of resistance of the grid material, S G is the difference in specimen and grid thermal expansions, F G is the gage factor of the strain gage, and T is an arbitrary temperature change from a reference temperature. The strain that is indicated from the change in gage resistance was then 34

PAGE 48

35 IiFRR (4.2) where F I is the gage factor setting of the instrument. The gage factor of the instrument was set to that of the gage so that the resulting thermal output, or strain, output by a gage on a specimen was GSGGSGTTF )/( (4.3) From this output it was necessary to remove the gage contribution so that only the surface expansion of the specimen was recorded. This was accomplished by placing the same gage type from the same lot number onto a reference material of a known expansion coefficient and expansion curve as a function of temperature. When the reference material was exposed to identical conditions as the specimen the reference material gives a thermal output of GRGGRGTTF )/( (4.4) The gage output was measured on the reference material then the theoretical strain at a given temperature was removed from the total output yielding the gage contribution to the thermal output [18]. This output was then subtracted from the thermal output of the specimen and gage reading, yielding only the thermal induced strain on the surface as )/()/(RGTSGTS (4.5) The acquired strain readings were later combined with CRM data. 4.2 Specimen Preparation To ensure the continuity of the measurements, the same specimens used for the CRM procedure were used for the strain gage experiments. The CRM specimens were

PAGE 49

36 sectioned so that the gratings were not damaged, yet a large enough area for strain gage application was obtained. This yielded specimens of approximately 1 x 4. The specimens were conditioned and sanded as recommended by Vishay Measurements Group for strain gage application. The test covered a wide range of temperatures that demanded special attention to gage and application techniques. For this reason M-Bond 610 was selected as the adhesive system and two WK-13-250BG-350 gages were applied to each side, one directed in the x-direction and the other in the y-direction. These gages were selected for their performance characteristics over the proposed test temperature range. Each gage was aligned in the proper orientation by use of mylar tape across the terminals. The gage backing and specimen surface were coated with M-Bond 610 and were exposed to the air for approximately 15 minutes. After drying, the gages were pressed down onto the specimen and covered with a thin Teflon film. Pressure pads were placed over the gages and then they were clamped at 40 psi with 2 spring clamps. The assembly was then placed in a cool oven and the temperature was increased to 121 C at a rate of 5 C/min. and then was allowed to cure for 3 hours. The tape was peeled off of the specimens and they were returned to the oven to be post cured for two hours at 135 C. A schematic and example of the resulting specimens are seen in Figures 4-1 and 4-2. Decoupling the strain gage expansion effects accurately was crucial to maintain the validity of testing results. The reference materials were prepared following the exact same procedures, making use of the same gages and bonding techniques, as those used for the preparation of the composite specimens (Figure 4-3).

PAGE 50

37 Figure 4-1: Schematic of strain gage alignment Figure 4-2: Gauging of sectioned specimen

PAGE 51

38 Figure 4-3: Gauging of reference materials 4.3 Testing Procedure The failure of the X-33 provided motivation for this work as well as the environmental conditions under which the residual strains, stresses, and CTE were studied. The composite specimens were observed and investigated over a wide range of temperatures, from the harsh conditions of the cryogenic realm up to the temperatures at which they were cured. The following section describes the methods that were used to test and acquire data in these conditions. After all preparations were made to the specimens and reference material they were placed in a Sun Systems Model EC1x environmental chamber. Each test consisted of three specimens and the reference material being exposed to the desired conditions as controlled by the environmental chamber. Due to the extensive amount of time that was necessary to run and collect data, each test was divided into separate cryogenic and elevated temperature runs.

PAGE 52

39 4.3.1 Cryogenic Tests A LN 2 tank was attached to the environmental chamber as seen in Figure 4-4. The chamber was then able to regulate the temperature by controlling the release of LN 2 The specimens and reference material were then supported on a rack covered with Teflon (Figure 4-5). The Teflon ensured that the specimens were free to expand and contract without being exposed to friction that might inhibit this deformation. Lead wires were passed through a port in the chamber to the data acquisition set up. The chamber door was returned and sealed while the proper connections were made to the acquisition hardware. Each gage was read in a quarter-bridge configuration (Figure 4-6). The lead wires were passed directly to a STB AD808FB Optim Electronics quarter-bridge completion module. The completed bridges were directed into a National Instruments SCXI-1322 breakout board attached to a NI SCXI-1122 sixteen-channel Isolated Transducer Multiplexer Module for Signal Conditioning. The signal conditioning board was part of a SCXI-1000 four slot chassis that directed the data into a Dell Dimension 8100 through a NI PCI 6035E data acquisition card. The data and hardware was manipulated by a NI LabVIEW program. The chamber temperature was set for 20 C and the specimens were allowed to soak there for 15 minutes. This temperature was selected as the starting point due to the material property values of the reference material. The National Institute for Standards

PAGE 53

40 Figure 4-4: Environmental chamber and LN 2 Figure 4-5: Laminate coupons wired and placed inside environmental chamber

PAGE 54

41 Figure 4-6: Quarter-bridge schematic and Testing has established the thermal expansion data for the reference material, Ti6-4Al-4V, as a deviation from this temperature. The titanium reference expands as: 10111.7807.4140.2711.1362312kkkTETETEE (4.6) This equation is accurate to within 1.5% from temperatures of 24K to 300K. A reading was taken at this point and the corresponding voltages were stored and used as tare voltages that were removed from all subsequent measurements. After acquiring the tare voltages at 20 C the temperature in the environmental chamber was reduced to 5 C. The specimens were again allowed to remain at that temperature for 15 minutes so that temperature was constant through the thickness. Another set of readings was taken at this temperature. The LabVIEW program that was written to manipulate the incoming data took the raw voltages and proceeded to compensate for several factors. The initial voltages read in the tare measurement were removed from the new voltage readings. The new gage factor for the corresponding temperature was calculated and then the raw voltage was converted into strain and adjusted for transverse sensitivity errors. This was done for all gages inside the chamber. The converted strain readings represented the total thermal output of the gages and the gage contribution was decoupled as follows. LabVIEW calculated the actual strain experienced by the reference material according to equation 4.6 and compared it to what

PAGE 55

42 was read. The difference in the readings corresponded to the gage contribution to the thermal output. This difference was calculated then subtracted from all other strain readings resulting in the true surface strain on the specimens. This was recorded and saved versus the temperature. The same procedure was followed throughout the experiment. Dwell times were increased as the temperature was lowered since it took more time for the specimens and reference material to reach thermal equilibrium. Thermal equilibrium was verified by taking several strain readings within a few minutes of each other. If the drift in the calculated strain was observed to be very small, on the order of 2 to 5 microstrain per 3 minutes then the final reading was recorded and the experiment proceeded to the next temperature. The experiment terminated with the final reading at -190 C. This was the lowest temperature that was obtainable using this environmental chamber and LN 2 The chamber temperature was returned to room temperature in several steps so as not to thermally shock the specimens. The specimens and reference material were removed from the chamber and it was heated to remove all moisture. This procedure was followed for all specimens. 4.3.2 Elevated Temperature Following the cryogenic testing the specimens were exposed to elevated temperatures from room temperature to the upper limit of the composite cure temperature. The same specimens were used for the elevated temperature tests so that a curve across the entire test range could be created for each individual specimen. The only difference was in the solder that was used to connect the lead wires, as the solder appropriate for exposure to cryogenic temperature is not applicable at elevated

PAGE 56

43 temperatures. The conductivity of both solders was very similar, thus the effect of this change was negligible. The same hardware and software programs were used to collect the data from elevated testing. The chamber with specimens and reference material inside was again set at 20 C where after a dwell time tare readings were taken. This initial point was selected in order to obtain a continuous graph of the surface strains. It was not set at 20 C to account for the reference material, as a small piece of Astrositall was gauged and used as a reference instead of the titanium. The data available for the titanium reference was not valid into the upper limits of the high temperature test. The Astrositall has an extremely low expansion coefficient so that in the test range its actual surface strain was treated as zero. The chamber was heated to 24 C and allowed to dwell there. This temperature location was necessary since it corresponds to the room temperature at which the CRM measurements were made. The need for the correspondence in temperature locations will be discussed later. The dwell time needed for the elevated temperature test was much longer than that of the cryogenic test due to the slower heat absorption of the Astrositall reference material. The LabVIEW program followed the same methodology for the calculation of the true surface stain. The only difference being that the reference material was different and had a different expansion curve. The Astrositall was treated as non-expanding throughout the entire test. Therefore, any indicated strain on the reference material was immediately subtracted from the specimen readings as the gage thermal contribution. The temperature intervals began as 15 C increments. After the temperature reached 90 C the increments were increased to 20 C. The test concluded when the

PAGE 57

44 chamber was raised to 182 C, which corresponds to the composite cure temperature. As before, when conditions began to deviate much from room temperature the dwell times were increased. Thermal equilibrium was verified with the same technique as previously discussed. 4.4 Data Analysis and Results The data obtained directly from the above methods did not directly provide the total surface strain present the laminate as a function of temperature. What was observed through the strain gage testing was the relative change in surface strain as the temperature was varied from 20 C. The strain data has to be coupled with the CRM data in order to encompass all surface strains. CRM was able to capture the entire strain induced on the surface of the laminate as it went through cure and was brought down to room temperature, which was 24 C. These strain values for the various laminates were highlighted previously (Tables 3-1 through 3-3). The data acquired through the use of strain gages must be referenced to those values. This was the reason for the strain measurement taken at 24 C by the strain gages. The total surface strain as a function of temperature was calculated and plotted in Microsoft Excel. The total surface strain present at any given temperature was: 24 CRMTlam (4.7) where T is the strain calculated by the strain gages at any temperature, CRM is the strain recorded by CRM at 24 C for a particular specimen, and 24 is the strain recorded by the strain gages at 24 C. Figures 4-4 through 4-6 show the resulting total surface strains as a function of temperature on the various specimens of each laminate sequence.

PAGE 58

45 All of the curves demonstrate fairly smooth behavior. The total surface strain results for different specimens of each stacking sequence are in good agreement with each other and no large deviations are observed among different specimens of the same laminate and the average COV across the entire temperature range is acceptable for each (Table 4-1). The surface strains have been averaged for each laminate (Figure 4-7). -12000-10000-8000-6000-4000-200002000-200-1000100200Temperature (C)Microstrain UNI-E x UNI-E y UNI-B x UNI-B y UNI-M x UNI-M y UNI-N x UNI-N y UNI-F x UNI-F y Figure 4-7: Unidirectional laminate surface strain A crucial observation to be made is that the strains do not return to zero as the temperature is returned to cure. The cure temperature is often referred to as a stress free state, hence strain is also expected to be zero at this point. This is only true during the curing of the laminate because during cure the epoxy undergoes polymerization, which is a one-time event. The resulting strain from this event, referred to as chemical shrinkage is always present, is often overlooked, and can contribute a large amount of strain to a given laminate (Tables 4-2 through 4-4). The stress measurement techniques described previously, and an analytical prediction tool discussed later, known as Classical

PAGE 59

46 Lamination Theory (CLT), do not have the ability to account for this phenomenon. CRM provides a tool that can accurately predict and calculate this effect. -12000-10000-8000-6000-4000-200002000-200-1000100200Temperature (C)Microstrain OAP-A x OAP-A y OAP-E x OAP-E y OAP-H x OAP-H y OAP-K x OAP-K y OAP-I x OAP-I y Figure 4-8: OAP laminate surface strain -1500-1300-1100-900-700-500-300-100100300500-200-1000100200Temperature (C)Microstrain RLV-C x RLV-C y RLV-L x RLV-L y RLV-G x RLV-G y RLV-Q x RLV-Q y Figure 4-9: RLV laminate surface strain The strain data was also transformed into a linear coefficient of thermal expansion from cure to the testing temperature. The data was taken in 15C and 20C intervals allowing the CTE of the laminates to be calculated as a function of temperature. The

PAGE 60

47 change in strain over the temperature interval was divided by that interval from cure to cryogenic. CTE was calculated for each direction in the various specimens as: )(1212TTCTExTxTx (4.8) )(1212TTCTEyTyTy (4.9) The CTE values as a function of temperature are displayed in Figure 4.11. Over the test range the change in CTE appears to be quite linear and trend lines were created as such. The linear fit of the CTE of the unidirectional laminate will later be used to refine Classical Lamination Theory predictions. 4.5 Error Sources Just as in the case of CRM any misalignment in the orientations during manufacturing would have led to incorrect measured values. Also, since the strain gage readings were combined with those of CRM, any problems that may have been encountered during the alignment of the displacement gratings would have propagated into the combined measurements. The most likely source for a large error would have resulted from the misalignment of a strain gage on the specimen. Strain gages are extremely sensitive to misalignment when placed on composite specimens. In some cases a 4 misalignment of a strain gage would have led to errors as large as in an axial gage measurement, while a %17 %65 error could exist in a similar transverse measurement. However, in this case the sensitivity to misalignment was not as dramatic. On the unidirectional laminates the gages were aligned along the 0 and 90 directions. It was assumed that enough effort was centered on strain gage alignment that less than offset was induced. This desired alignment dropped 2

PAGE 61

48 possible misalignment errors approximately to less than 5% in the case of gages off axis by In the case of the OAP laminate a 2 2 offset leads to an error of about 3%. The somewhat quasi-isotropic nature of the RLV laminate again helped to minimize the effects of gage misalignment keeping errors on the order of 5%. 4.6 Conclusions A method of measuring relative surface strains as they develop as a function of temperature has been outlined. Proper compensation methods to account for temperature effects on measurement techniques have been discussed and automated in a LabVIEW data acquisition program. Measurements of surface strain have been made across a large temperature range and the CRM and gage values have been combined in a manner to account for the total surface strain at any given temperature. The surface strains for all specimens have been plotted as a function of temperature and the averages of the different laminates have been compared to each other. The chemical shrinkage component of the total surface strain has been decoupled. Finally, the average CTE of the different laminates has been computed as a function of temperature. The different runs of the various specimens were quite consistent. The surface strain values were reproducible as a function of temperature from run to run. As in the results from CRM in chapter 3, the unidirectional and OAP laminates continued to display a higher strain level than the RLV laminate. The same conclusion is reached, that the increasing offset of ply angles has lead to a lower strain which resulted from high constraint imposed in the RLV laminate. The decoupled chemical shrinkage terms were quite large for the transverse direction of the unidirectional and OAP laminates. These large values indicate that any

PAGE 62

49 work that neglects the chemical shrinkage components will neglect a significant strain event. These values were not seen in the RLV due to the usual constraint mechanisms. The CTE was seen to tend toward zero as the temperature was decreased. This event was observed less in the RLV laminate since the CTE was already quite low, again due to high constraint. The trend toward zero was expected as molecules will tend to slow in all motion as the temperature is reduced. The CTE observed in the unidirectional laminates was taken to be the CTE of the material system and is used in CLT calculations in Chapter 7 as 1 and 2 Table 4-1: Average COV of surface strains across temperature range Laminate COV x-direction COV y-direction UNI ----4.26% OAP 13.55% 6.10% RLV 9.59% 10.10% -12000-10000-8000-6000-4000-200002000-200-1000100200Temperature (C)Microstrain UNI x UNI y OAP x OAP y RLV x RLV y Figure 4-10: Average of laminate surface strains

PAGE 63

50 -10-505101520253035-200-150-100-50050100150200Temperature (C)CTE (microstrain/C) UNI x UNI y OAP x OAP y RLV x RLV y Figure 4-11: Coefficient of thermal expansion Table 4-2: Unidirectional chemical shrinkage induced strain Laminate x-direction y-direction UNI B --------UNI E 48.5 -2978.5 UNI F 99.3 -2338.5 UNI M 155.9 -2831.1 UNI N 131.1 -2702.0 Average 108.6 -2712.5 Table 4-3: OAP chemical shrinkage induced strain Laminate x-direction y-direction OAP A -----2164.4 OAP E --------OAP H 298.5 -1359.5 OAP I 550.2 -2153.1 OAP K 582.1 -2430.7 Average 476.9 2026.9 Table 4-4: RLV chemical shrinkage induced strain Laminate x-direction y-direction RLV C --------RLV G --------RLV L -169.3 29.3 RLV Q -83.8 -47.8 Average -126.55 -9.25

PAGE 64

CHAPTER 5 MATERIAL PROPERTY TESTING Basic material tests include those that provide some of the most elementary yet critical values that can be assigned to a material. The following experiments were performed to obtain the Youngs Moduli, E 1 and E 2 Poissons ratio, 12 and the shear modulus, G 12 These properties are necessary for the prediction of the strains and stresses provided by Classical Lamination Theory (CLT). They must also be used to transform the experimentally measured strains into stresses. The American Society for Testing and Materials (ASTM) book of standards was used as a guideline for the testing procedures implemented. Standard D 3039 was used as the model for the E 1 and E 2 tests, while standard D 5379 was used to determine the shear modulus. 5.1 E 1 Testing ASTM standard D3039 is a test method designed to determine the in-plane tensile properties of a polymer matrix composite with high modulus fiber reinforcement [44]. A panel of unidirectional IM7/977-2 was manufactured as discussed in previous sections without the previsions made for the implementation of CRM. The panel had dimensions of 10 in the fiber direction by 6 transversely. The panel was seven layers thick to approach the ASTM recommended thickness of 0.040. After the panel was cured, test coupons were cut from it having nominal dimensions of 10 x 1. Each coupon was then sanded and conditioned at its midpoint in accordance with Vishay Measurements Group techniques for strain gage application. 51

PAGE 65

52 One CEA-13-125UT-350 tee gage from Vishay was placed on each side of the coupon with the standard M-bond 200 cyanoacrylate glue. One gage located on each side allows for cancellation of bending effects. The branches of the tee were placed precisely in the fiber and transverse directions allowing for the acquisition of information needed to calculate both the E 1 and 12 properties. After the coupons were gauged they were then placed in a MTI Phoenix machine where they could be loaded (Figure 5-1). Each specimen was loaded axially in the fiber direction with a crosshead displacement rate of 0.01 per minute. Load versus strain data was acquired nominally at 200-pound intervals with National Instruments data acquisition hardware. The hardware consisted of an SCXI-1000 four slot chassis, SCXI-1322 breakout board, and a SCXI-1122 sixteen-channel Isolated Transducer Multiplexer Module for Signal Conditioning all read into a Dell Dimension 8100 through a NI PCI 6035E data acquisition card. NI LabVIEW software was used to operate and read the acquisition hardware. 5.2 E 2 Testing ASTM standard 3039 was also used to establish the E 2 testing procedures [44]. In this case a panel of IM7/977-2 was manufactured with dimensions of 6 in the fiber direction by 7 in the transverse direction. The panel was 14 layers thick to bring the thickness nominally to 0.080. Again the panel was cut into test coupons after the curing process with the test coupons nominally of dimensions 7 x 1. Each specimen was sanded and conditioned as before at its midpoint. One CEA-06-250UN-350 unidirectional gage was applied to

PAGE 66

53 each side of the specimen. The gage was placed on the coupon precisely in the matrix direction, transverse to the fibers. Figure 5-1: E 1 and E 2 testing fixture

PAGE 67

54 As before, after the specimens were prepared they were subjected to load in the MTI Phoenix displacement controlled testing machine. The displacement rate was again 0.01 per minute with load versus strain data acquired at approximately 200-pound intervals. The same NI data acquisition hardware was used for this procedure as in the E 1 testing. 5.3 G 12 Testing As with the previous material property tests, the ASTM book of standards was used as the guideline for establishing test setup and procedures. ASTM D 5379 is designed to determine the shear properties of high-modulus fiber reinforced composite materials [45]. A 6 x 6 13 layer panel of unidirectional IM7/977-2 was manufactured and then cured in the autoclave with the appropriate cure cycle. After the panel is cured it was sectioned and machined into the proper shape. The testing coupons were machined into the form of a rectangular flat strips with two v-shaped notches centered along the length. A diagram of the appropriate geometry is seen in Figure 5-2. The panel was sent to TMR Engineering for machining of the correct geometry. The v-notched specimens were then prepared for testing through the application of Vishay micro-measurements N2A-06-C032A-500 shear gages. The specimens were conditioned appropriately and then one gage was applied to each side precisely in the area spanning the concentric notches. The gauged coupons were then placed in the MTI machine with a special testing fixture seen in Figure 5-3. The specimens were inserted in the fixture and then placed onto the machine in a manner where the notches were along the line-of-action of the

PAGE 68

55 testing machine. The two halves were then tensioned at a rate of 0.01 per minute while the load-shear strain data was collected using the NI hardware. 5.4 Data Analysis and Results Before each E 1 E 2 and G 12 test was run the dimensions of the all test coupons were measured. The stress, in each coupon was calculated as a function of cross sectional area, A, and the load applied by the MTI machine, P: AP (5.1) The stress was then plotted versus the strain in each coupon. The slopes of the resulting graphs (Figures 5-4 and 5-5) were the Youngs moduli of each coupon. The various moduli were very similar in value and were averaged. The results of each test are seen in Table 5.1. The E 1 testing yielded an average value of 150.2 GPa for the modulus of elasticity in the fiber direction, while the E 2 testing yielded an average of 8.04 GPa for the transverse or matrix modulus. The shear specimen results are plotted in Figure 5.6. The shear modulus was calculated in accordance with ASTM Standard D 5379 where a shear chord between the values of 1000 and 3000 was calculated as chordG (5.2) The resulting average value for the shear modulus was 4.77 GPa. The Poissons ratio was also calculated from the E 1 test by use of the strain readings from the transverse gages. The strains displayed by the longitudinal and transverse gages were compared and Poissons ratio was calculated as

PAGE 69

56 1212 (5.3) All calculated Poissons ratios were very similar and average to 0.342. Figure 5-2: Shear test specimen schematic Figure 5-3: Shear Testing Fixture

PAGE 70

57 0500001000001500002000002500000200040006000800010000MicrostrainStress (psi) Test 1 Test 2 Test 3 Test 4 Figure 5-4: E 1 stress/strain curves 0100020003000400050006000700080009000010002000300040005000600070008000MicrostrainStress (psi) Test 1 Test 2 Test3 Figure 5-5: E 2 stress/strain curves

PAGE 71

58 010002000300040005000600070008000900002000400060008000MicrostrainStress (Pa) test 1 test 2 test 3 test 4 Figure 5-6: G 12 stress/strain curves Table 5-1: Results of each tension test Run E 1 (GPa) E 2 (GPa) 12 G 12 (GPa) 1 151.5 7.67 0.345 4.05 2 151.6 8.40 0.343 5.10 3 147.1 8.05 0.343 4.93 4 150.5 ----0.338 5.00 Ave 150.2 8.04 0.342 4.77 COV 1.40% 4.54% 0.87% 10.16% 5.5 Micro-mechanical Methods A detailed micro-mechanical approach is also being used by others to develop characteristics of the IM7/977-2 material system [4]. The micro-mechanics approach uses a unit-cell model of the material system. The unit-cell is modeled and analyzed using finite element analysis of solid elements. Periodic boundary conditions are applied to the surface of each cell. This micro-mechanical approach has yielded the E 1 E 2 12 and G 12 properties as well. Table 5-2 compares the results of the micromechanical approach with experimentally determined values.

PAGE 72

59 Table 5-2: Comparison of experimental measurements and micro-mechanics values Material Property Experimental Average General Dynamics Rockwell Micro-mechanics Max Percent Difference Min Percent Difference E 1 (GPa) 150.2 169.6 148.2 159 12.1 1.34 E 2 (GPa) 8.04 ----9.31 9.43 15.9 14.6 12 .342 --------.253 29.9 29.9 G 12 (GPa) 4.77 4.80 5.17 4.34 9.45 0.63 5.6 Error Sources Potential sources for error in the determination of material properties included many of those listed in pervious chapters. These potential sources were manufacturing defects, especially layer orientation mistakes, and gage misalignment. Again, misalignment of the strain gages is the most likely to result in the development of significant errors. For the most part the potential errors were reduced to below 5% due to the orientation of the gage with the fibers in the specific tests. The only case of potential significant error was in the 12 measurement made by measuring the transverse strain on the E 1 specimen. Any transverse measurement was much more sensitive to misalignment, however, again the desired angle helped to alleviate possible error. Transverse measurements were expected to have less than 10% potential error. 5.7 Conclusions Various material tests have been conducted using ASTM testing standards. The material properties of Young Modulus in both the fiber and transverse direction have been calculated as well as the resulting Poisons ratio. Further testing was conducted to evaluate the shear modulus of the material system.

PAGE 73

60 The data was very consistent and possessed little scatter in most of the measurements. The COV of the various material properties were very low with the exception of the shear modulus at 10.16%. Comparisons were made between the experimentally determined material properties and those calculated using a micro-mechanics based simulation. This comparison of material properties yielded widely varying results. The E 1 results agreed quite well and the E 2 results were not extremely different. However, the results for 12 did not yield good agreement. There were several potential sources for the dramatic differences. One significant source of difference could have been the assumptions made in the micro-mechanic model. This is a FE based model that is very depended on boundary conditions. Also this model was applied on a single unit cell assuming a square packing arrangement. These may have been poor assumptions. Just as likely to yield differing results were the potential error sources listed previously. All of the samples for each test were sectioned from one plate specific to that test. If any errors had been made in the lay-up of the plate then those would have been seen in all tested specimens yielding a precise but not accurate result. More material testing should be conducted on specimens from more plates to investigate these differences.

PAGE 74

CHAPTER 6 PLY LEVEL RESIDUAL STRAINS AND STRESSES The residual strains and stresses in a composite result from the mismatch in contraction of the matrix and fibers. This mismatch in contraction is from the combination of drastically different CTEs and the chemical shrinkage event that is limited to the polymer matrix. The follow section will analyze the data acquired from the previous experimental techniques and present the resulting macro-mechanical residual strains and stresses. 6.1 Residual Strain Calculation By determining the state that each ply would reach if the bonds between them were liberated the stresses that held them together can be defined [18]. This liberated strain will be referred to as the residual strain. Using the assumptions of laminate theory, where plane sections remain plane, for symmetric lay-ups, the strain on the surface matches that at any point in the interior, as long as the location is a few ply thicknesses away from the free edge. The surface strain in each laminate is first transformed into its fiber and matrix components of each ply: 12_2_1_12_2_1_lamlamlamkplyplyplyT (6.1) where lam is the total surface strain and [T] k is the transformation matrix of each ply: 61

PAGE 75

62 222222sincossincossincossincos2cossinsincos2sincosT (6.2) The residual strain in each ply then becomes: 12_2_1_12_2_1_12_2_1_plyplylplyuniuniunikresresres (6.3) where { uni } is the free expansion surface strain measured from the unidirectional specimens. This surface strain represents the strain that a ply would reach if not constrained by adjacent layers. These calculations are made for each temperature in the experimental test range. Matrix properties are significantly weaker than those of the fiber and the first sign of damage is typically mirocracks in the matrix. For this reason, the residual strains reported will be those in the transverse direction. The residual strains are plotted as a function of temperature for each layer in the OAP and RLV laminates (Figure. 6-1). 6.2 Residual Stress After residual strain was calculated the stress-strain relations were used to define the corresponding residual stresses: kresresreskkresresresQQQQQ12_2_1_662221121112_2_1_0000 (6.4) The resulting residual stresses in the ply of the OAP and RLV laminates are plotted versus temperature (Figure 6.2).

PAGE 76

63 -1.00E-031.00E-033.00E-035.00E-037.00E-039.00E-031.10E-02-200-150-100-50050100150200Temperature (C)Strain OAP RLV 45 RLV 90 RLV 0 Figure 6-1: Residual strain on selected plies 0.00E+001.00E+072.00E+073.00E+074.00E+075.00E+076.00E+077.00E+078.00E+079.00E+07-200-1000100200Temperature (C) Stress (Pa) OAP RLV 45 RLV 90 RLV 0 Figure 6-2: Residual stress level on selected plies

PAGE 77

64 6.3 Conclusions A method to quantify residual strains and stresses based on surface strains has been discussed. This method has been used to display the resulting residual stresses that were present in the different laminates. As expected, the high level of constraint that has been discussed throughout this thesis has lead to high level of residual stress. The OAP has performed much better than the RLV lay-up in reference to the development of thermally induced stresses. The often neglected chemical shrinkage has induced as much as 15.4 MPa of residual stress in the RLV laminate. From these results it has been made apparent that the chemical shrinkage is not a negligible term and must be determined to fully quantify the extent of residual stress development. It has also been made quite apparent that there is a great need for the development of optimization techniques to account for the interaction of both mechanically and thermally induced stresses.

PAGE 78

CHAPTER 7 CLASSICAL LAMINATION THEORY Classical Lamination Theory (CLT) is a predictive tool that makes it possible to analyze complex coupling effects that may occur in composites. CLT is able to predict strains, displacements, and curvatures that develop in a laminate as it is mechanically loaded as well as those resulting from hygrothermal stresses. It is the thermal contribution to strain that will be investigated in the following section. CLT is derived from many assumptions, the most basic of which are that individual lamina are perfectly bonded together so that they behave as a unitary, nonhomogeneous, anisotropic plate [13]. The most important assumption in CLT is that each ply is assumed to be in a state of plane stress and that interlaminar stresses are neglected. 7.1 Theory Development Using the coordinate system in Figure 7-1, the laminated plate geometry and ply numbering scheme defined in Figure 7-2, along with several basic assumptions, a set of equations are derived to predict deformations and stresses. The basic assumptions needed in the derivation are: 14. The plate consists of orthotropic lamina bonded together, with principal material axis directed in arbitrary directions with respect to the x and y axes. 15. Edge lengths of the plate must be much greater than the individual ply thicknesses. 16. The u,v,w displacements are small in comparison to the total plate thickness. 17. The in-plane strains x y and xy are much smaller than one. 18. Transverse shear strains xz and yz are neglected. 19. The u and v displacements are linear functions of the z coordinate. 65

PAGE 79

66 20. The transverse normal strain z is neglected. 21. Hookes Law applies to each ply. 22. Total plate thickness is constant. 23. At the plate surface the transverse shear stresses xz and yz are zero. Assumptions 6 and 7 define the displacements as ),(),(),(),(),(),(02010yxwyxwwyxzFyxvvyxzFyxuu (7.1) where u and are the displacements in the x and y directions at the middle of the plate. The displacement is the same at any thickness in the plate with the same x and y coordinates due to assumption 7. Substituting equations 7.1 into the strain-displacement equations for the transverse shear strains and imposing assumption 5 yields 0 0v w 0),(0),(21 ywyxFywzvxwyxFxwzuyzxy (7.2) and ywyxFxwyxF ),(),(21 (7.3)

PAGE 80

67 x y z Figure 7-1: Laminate coordinate system k 3 2 1 N z2 z1 z0 z3 zN-1 zk zk-1 Middle S urf ace t zN Figure 7-2: Layer schematic and nomenclature Combining the strain-displacement relations for in-plane strains with Eqs. (7.1) and (7.3) results in xyxyxyyyyxxxzxvyuzyvzxu000 (7.4)

PAGE 81

68 where mid-laminate strains are xvyuyvxuxyyx0000000 (7.5) and mid-laminate curvatures are given by yxwywxwxyyx222222 (7.6) These represent the curvatures associated with the bending at the middle of the laminate in the xz and yz planes as well as the out of plane twisting of the middle surface of the laminate. In our case, as with any symmetric laminate, these curvatures will be zero. They are left in the derivation to give the solution for any arbitrary stacking sequence where curvature may be present. Equation (7.4) provides a solution for the strain at any given distance from mid-laminate. If the various lamina are of known thickness then the distance of each from mid-laminate is known and the stress-strain relations for the k th lamina can be expressed as xyxyyyxxkxyyxzzzQQQQQQQQQ000662616262212161211 (7.7) where

PAGE 82

69 TQTQ1 (7.8) and 12662112222212112212122112111111GQEQQEQEQ (7.9) When determining strains and stresses in a laminated plate it is convenient to look at the forces and moments per unit length as they act on a plate. The force per unit length, N x is given by NkzzkxttxxkkdzdzN1221 (7.10) and the moment per unit length, is given by xM NkzzkxttxxkkzdzdzzM1221 (7.11) where t is the laminate thickness, kx is the stress in the k th lamina, z k-1 is the distance from mid-laminate to inner surface of the k th lamina, and z k is the distance to the outer surface of the same lamina. Combining (7.7) with both (7.10) and (7.11) yields the following equations for the force and moment per unit length. dzzzQzQzQMdzzQzQzQNNkzzxyxykyykxxkxNkzzxyxykyykxxkxkkkk1016012011101601201111 (7.12) after rearranging (7.12) the force and moment equations can be written as

PAGE 83

70 xyyxxyyxxxyyxxyyxxDDDBBBMBBBAAAN161211016012011161211016012011 (7.13) where the laminate extensional stiffness is given by 1122kkkNkijttkijijzzQdzQA (7.14) the laminate coupling stiffness is given by 21212221kkkNkijttkijijzzQdzzQB (7.15) and the laminate bending stiffness is 313122231kkkNkijttkijijzzQdzzQD (7.16) Finally the entire set of equations can be expressed as xyyxxyyxxyyxxyyxDDDBBBDDDBBBDDDBBBBBBAAABBBAAABBBAAAMMMNNN000662616662616262211262211161211161211662616662616262212262211161211161211 (7.17) The above solution exists for the state in which laminates are not subjected to any thermal or moisture effects. Together these effects are known as hygrothermal effects. It is the thermal component of this stress that is of the greatest concern in this case and therefore the stresses and strains in the laminate must be solved for this hygrothermal effect. When the moisture and thermal effects are incorporated the resulting stresses on each lamina are

PAGE 84

71 cTQkkkkk (7.18) and the total lamina strains equal zk0 (7.19) Equation (7.19) is substituted into Eqn. (7.18) giving cTzQkkkk0 (7.20) The force and moment per unit length are then solved in the same manner but this time by integrating (7.20). MTkkkkMTkkkkMMDBMzdzcTzQzdzMNNBANdzcTzQdzN0000 (7.21) The contributions of the thermal and moisture components are 1111kkkkNkkkMkkkkNkkkTzzQccdzQNzzQTTdzQN (7.22) and 2121212122kkkkNkkkMkkkkNkkkTzzQcczdzQMzzQTTzdzQM (7.23) Finally, the entire solution can be written in matrix form as

PAGE 85

72 xyyxxyyxMxyTxyxyMyTyyMxTxxMxyTxyxyMyTyyMxTxxDDDBBBDDDBBBDDDBBBBBBAAABBBAAABBBAAAMMMMMMMMMNNNNNNNNN000662616662616262211262211161211161211662616662616262212262211161211161211 (7.24) This approach was used to predict the strains on the surface of the laminates and compare them to those measured via experiment. A MatLAB code was authored, primarily by Thomas Singer and William Schulz, to encompass all of the assumptions and calculations involved in CLT (Appendix B). The code was executed at each temperature in which experimental data was collected. CLT makes its predictions based on the material properties of individual lamina. The material properties of E 1 E 2 12 G 12 and 1 and 2 were obtained from previous experiments. The CTE values of 1 and 2 have been calculated as functions of temperature from previous analysis. The apparent change in CTE is linear and therefore an average CTE across the test range was used as the input for the CLT program. 7.2 Experimental vs. Analytical Results CLT results were plotted as a function of temperature and compared to the experimental results (Figures 7-3through 7-5). It can be seen in all cases that there is substantial difference between the results of CLT and experimental data. Several factors could lead to potential differences in the measured and predicted values. The first potential source is in the assumptions that are made in CLT formulation. The second and most crucial difference is that CLT lacks the ability to interpret the event of chemical shrinkage and the effect it has on laminate strain. If the values at the cure temperature are compared it is seen that CLT predicts no strain at this point, while the

PAGE 86

73 combined experimental techniques show strains that are the result of the chemical shrinkage. Since chemical shrinkage is a one-time event and the resulting strains are an invariant part of the laminate strain they can be added to the results of CLT. Upon the introduction of the chemical shrinkage term into the CLT results it was seen that the graphs collapsed onto each other (Figures 7-6 through 7-8 ). -12000-10000-8000-6000-4000-200002000-200-1000100200Temperature (C)Microstrain x CLT y ClLT x y Figure 7-3: Unidirectional CLT and experimental results -10000-8000-6000-4000-200002000-200-1000100200Temperature (C)Microstrain x CLT y CLT x y Figure 7-4: OAP CLT and experimental results

PAGE 87

74 -1300-1100-900-700-500-300-100100-200-1000100200Temperature (C)Microstrain x CLT y CLT x y Figure 7-5: RLV CLT and experimental results The resulting agreement between the CLT and experimental values provides proof of testing concept and a high degree of measurement accuracy. Figure 7-6: Unidirectional CLT shifted for chemical shrinkage

PAGE 88

75 Figure 7-7: OAP CLT shifted for chemical shrinkage Figure 7-8: RLV CLT shifted for chemical shrinkage

PAGE 89

76 7.3 Conclusions Classical Laminate theory has been examined and used to predict the surface strains on the different laminates. This examination has taken place as an attempt to verify the experimentally determined results. It is a well known fact that the traditional approach to lamination theory does not have the means to account for chemical shrinkage. When the two results were compared an interesting phenomenon was observed. The measured and predicted curves appeared to follow nearly the same trends, separated by an offset. When the curves were examined it was seen that this offset was almost exactly equal to that of the chemical shrinkage. When the previously decoupled chemical shrinkage term was added to the analytical results form CLT the offset in the curves vanished and the two results were nearly identically. Several substantial conclusions have been drawn from these results. First, the fact that the resulting curves coincided with each other throughout the testing range confirmed that the chemical shrinkage can be treated as a one time event, completely independent of temperature effects. Secondly, the fact that the two curves coincided very well provided proof of concept in the strain gage measurements and temperature compensation technique. Last, and perhaps the most important conclusion was that the CRM method did indeed provide a means by which the true total strain present in a laminated composite could be measured. From this result, it is seen that once the chemical shrinkage term is decoupled and defined, the computer based analytical tool of CLT can be used to predict the development of strains anywhere in a given temperature range.

PAGE 90

77 APPENDIX A CRM IMAGES OF SPECIMENS U V Figure A-1: Unidirectional Laminate Run 1 V, y U, x 77

PAGE 91

78 U V Figure A-2: Unidirectional Laminate Run 2 V Figure A-3: Unidirectional Laminate Run 3 V, y U, x V, y U, x

PAGE 92

79 V Figure A-4: Unidirectional Laminate Run 4 V Figure A-5: Unidirectional Laminate Run 5 V, y U, x V, y U, x

PAGE 93

80 U V0 Figure A-6: OAP Laminate Run 1 U V V, y U, x V, y U, x Figure A-7: OAP Laminate Run 2

PAGE 94

81 U V Figure A-8: OAP Laminate Run 3 U V V, y U, x V, y U, x Figure A-9: OAP Laminate Run 4

PAGE 95

82 U V Figure A-10: OAP Laminate Run 5 U V V, y U, x V, y U, x Figure A-11: RLV laminate Run 1

PAGE 96

83 U V Figure A-12: RLV Laminate Run 2 U V Figure A-13: RLV Laminate Run 3 V, y U, x V, y U, x

PAGE 97

84 U V Figure A-14: RLV Laminate Run 4 V, y U, x

PAGE 98

APPENDIX B MATLAB CLT CODE Main Routine: laminatestress function [strain0,curve,stressx,stress1]=laminatestress(plydir,thick,E1,E2,G12,nu12,NM,deltaT,c,alpha12,beta12,plots) % laminatestress uses classical laminate theory to give midplane strain and curvature, and % stress (in both the coordinates of the laminate and of the ply) at the top and bottom of each ply. % All referenced equations are from Gibson, Principles of Composite Material Mechanics, 1994 % strain0: midplane strain, in laminate coords. [e_x;e_y;g_xy]. % curve: curvature. [K_x;K_y;K_xy]. % stressx: stress in laminate coords. stressx is 6 x k, where k is the number of plys. % The kth column gives stress in the x and y directions and shear in the xy plane at % the top and bottom of the kth ply; [s_x_top;s_y_top;s_xy_top;s_x_bot;s_y_bot;s_xy_bot]. % stress1: stress in ply coords. Same as stressx, but the kth column is in the kth ply % coords, with the 1-direction longitudinal, and 2-direction transverse. % plydir: string of the form '[90/+-45/(-+30)3/90bar]s' or column matrix of ply directions. Note % that the variables make more sense with the z-axis down; following this convention, the % first ply is the most negative. % plythick: column vector of ply thicknesses. If thick is a scalar, the thickness is applied to % all plys. % E1: column vector of Young's modulii in the fiber direction of each ply. If E1 is a scalar, % the modulus is applied to all plys. % E2: same as E1, in transverse directions. % G12: same as E1, but shear modulii. % nu12: same as E1, but Poisson's ratios. % NM: column vector of mechanical loading conditions [N_x;N_y,N_xy;M_x;M_y;M_xy] per unit % length. % deltaT: change in temperature from cure temp. % c: moisture concentration. % alpha12: matrix of coeffs of thermal expansion, in material coords. alpha12 % is 2xn or 3xn, where n is the number of plies. If alpha12 is 2x1 or 85

PAGE 99

86 % 3x1, the CTEs are assumed to apply to all plies. Row 1 is CTE in the % fiber direction, row 2 is CTE in the transverse direction, and row 3 % is 0. % beta12: matrix of coeffs of hygroscopic expansion. Same format as alpha12. % plots: 1 plots stressx and stress1 through thickness; 0 or blank doesn't plot. % Note: the last 5 inputs are optional. If only one is entered, it is taken to be plots. % If 2 are entered, they are taken as deltaT and alpha. If 3 are entered, they are taken % as deltaT, alpha12, and plots. If 4 are entered, they are taken as deltaT, c, alpha12, % and beta12. %----------Housekeeping---------if exist('deltaT')==0 deltaT=0; c=0; alpha12=[0;0;0]; beta12=[0;0;0]; plots=0; elseif exist('c')==0 plots=deltaT; deltaT=0; c=0; alpha12=[0;0;0]; beta12=[0;0;0]; elseif exist('alpha12')==0 alpha12=c; c=0; beta12=[0;0;0]; plots=0; elseif exist('beta12')==0 plots=alpha12; alpha12=c; c=0; beta12=[0;0;0]; elseif exist('plots')==0 plots=0; end %Sets variables not assigned in input [a,b]=size(NM); for i=a:5 NM(i+1,1)=0; end %Fills in missing force and moment terms if ischar(plydir) plydir=directions(plydir); end %Converts shorthand ply orientation to matrix

PAGE 100

87 %----------Computation------------[m,n]=size(plydir); [stiff,Q,S,Qbar,Sbar,T,z,z0]=ABD(plydir,thick,E1,E2,G12,nu12); %Calculates ABD matrix [Nh,Mh,alpha,beta]=hygrothermal(deltaT,c,alpha12,beta12,Qbar,T,z0,z); NM=NM+[Nh;Mh]; %Adds hygrothermal forces and moments. Eqs 7.95, 7.96 invstiff=inv(stiff); ek=invstiff*NM; %Solves for midplane strain and curvature. Eq 7.46 strain0=ek(1:3); curve=ek(4:6); %Extracting midplane strain and curvature strainxtop=strain0+z0*curve; %Pesky 0-index problem...see below for k=1:m strainx(:,k)=strain0+z(k)*curve; botstressx(:,k)=Qbar(:,:,k)*(strainx(:,k)-alpha(:,k)*deltaT-beta(:,k)*c); if k==1 topstressx(:,k)=Qbar(:,:,k)*(strainxtop-alpha(:,k)*deltaT-beta(:,k)*c); else topstressx(:,k)=Qbar(:,:,k)*(strainx(:,k-1)-alpha(:,k)*deltaT-beta(:,k)*c); end %Stress in general (x-y) coords. Eq 7.85 botstress1(:,k)=T(:,:,k)*botstressx(:,k); topstress1(:,k)=T(:,:,k)*topstressx(:,k); %Stress in ply (1-2) coords. Eq 2.31 end stressx=[topstressx;botstressx]; stress1=[topstress1;botstress1]; %---------Plotting--------if plots==1 for k=1:m stressxx(2*k-1,1)=stressx(1,k); stressxx(2*k,1)=stressx(4,k); stressxy(2*k-1,1)=stressx(2,k); stressxy(2*k,1)=stressx(5,k); stressxxy(2*k-1,1)=stressx(3,k); stressxxy(2*k,1)=stressx(6,k); stress11(2*k-1,1)=stress1(1,k); stress11(2*k,1)=stress1(4,k); stress12(2*k-1,1)=stress1(2,k); stress12(2*k,1)=stress1(5,k);

PAGE 101

88 stress112(2*k-1,1)=stress1(3,k); stress112(2*k,1)=stress1(6,k); end zplot=z0; for k=1:m zplot(2*k,1)=z(k); if k~=m zplot(2*k+1,1)=z(k); end end scalex=max(stressxx); scaley=max(stressxy); scalexy=max(stressxxy); scale1=max(stress11); scale2=max(stress12); scale12=max(stress112); scalex=max([scalex,scaley,scalexy])+.1*abs(max([scalex,scaley,scalexy])); scale1=max([scale1,scale2,scale12])+.1*abs(max([scale1,scale2,scale12])); scalelox=min(stressxx); scaleloy=min(stressxy); scaleloxy=min(stressxxy); scalelo1=min(stress11); scalelo2=min(stress12); scalelo12=min(stress112); scalelox=min(0,min([scalelox,scaleloy,scaleloxy])-.1*abs(min([scalelox,scaleloy,scaleloxy]))); scalelo1=min(0,min([scalelo1,scalelo2,scalelo12])-.1*abs(min([scalelo1,scalelo2,scalelo12]))); strainxplot=[strain0(1)+curve(1)*z0;strain0(1)+curve(1)*z(m)]; strainyplot=[strain0(2)+curve(2)*z0;strain0(2)+curve(2)*z(m)]; strainxyplot=[strain0(3)+curve(3)*z0;strain0(3)+curve(3)*z(m)]; strainzplot=[z0;z(m)]; figure; plot(strainxplot,strainzplot) title('x-strains throughout the thickness, (x,y) coords') xlabel('Strain') ylabel('Postion') axis([-10e-3,20e-3,z0,z(m)]) figure; plot(strainyplot,strainzplot) title('y-strains throughout the thickness, (x,y) coords') xlabel('Strain') ylabel('Postion') axis([-10e-3,20e-3,z0,z(m)]) figure; plot(strainxyplot,strainzplot) title('Shear strains throughout the thickness, (x,y) coords') xlabel('Strain')

PAGE 102

89 ylabel('Postion') axis([-10e-3,20e-3,z0,z(m)]) figure; plot(stressxx,zplot) title('x-stresses throughout the thickness, (x,y) coords') xlabel('Stress') ylabel('Position') axis([scalelox,scalex,z0,z(m)]) figure; plot(stressxy,zplot) title('y-stresses throughout the thickness, (x,y) coords') xlabel('Stress') ylabel('Position') axis([scalelox,scalex,z0,z(m)]) figure; plot(stressxxy,zplot) title('Shear stresses throughout the thickness, (x,y) coords') xlabel('Stress') ylabel('Position') axis([scalelox,scalex,z0,z(m)]) figure; plot(stress11,zplot) title('1-stresses throughout the thickness, ply (1,2) coords') xlabel('Stress') ylabel('Position') axis([scalelo1,scale1,z0,z(m)]) figure; plot(stress12,zplot) title('2-stresses throughout the thickness, ply (1,2) coords') xlabel('Stress') ylabel('Position') axis([scalelo1,scale1,z0,z(m)]) figure; plot(stress112,zplot) title('Shear stresses throughout the thickness, ply (1,2) coords') xlabel('Stress') ylabel('Position') axis([scalelo1,scale1,z0,z(m)]) end Subroutine: directions function dir=directions(str) % DIRECTIONS Converts from shorthand ply angle notation % DIRECTIONS(str) outputs a column vector of ply angles as given by str. % str should be a string of the form '[(25)2/+-45/(-+60/0)3/90bar]s' % In extended notation, this is % [25/25/+45/-45/-60/+60/0/-60/+60/0/-60/+60/0/90/0/+60/-60/0/+60/-60/0/+60/-60/-45/+45/25/25] % Always start with [. Use / to seperate ply angles. Use (plys)n to % indicate a set of plys repeated n times. Use +-theta to indicate % +theta/-theta, and -+theta to indicate -theta/+theta. Use bar after % the last ply to indicate a symmetric laminate in which the midplane

PAGE 103

90 % passes through the middle of the ply. End the ply sequence with ]. % Use s at the end to indicate symmetry (must be used with bar); t or % nothing denotes that the full ply sequence has been given. % % It may sound confusing, but it should actually be quite natural if % you are familiar with the shorthand notation. % % See also PARENTHESES str=lower(str); len=length(str); if str(len)=='t' & str(len-4:len-2)=='bar' error('Hey, dummy, is it symmetric or not?') end sym=0; oddsym=0; if str(len)=='s' sym=1; str=str(1:len-2); elseif str(len)=='t' str=str(1:len-2); else str=str(1:len-1); len=len+1; end if str(len-4:len-2)=='bar' sym=1; oddsym=1; str=str(2:len-5); else str=str(2:len-2); end len=length(str); str=strcat(str,'/'); slash=findstr('/',str); strpos=1; plypos=1; slashpos=1; plydir=0; while strpos
PAGE 104

91 strpos=slash(slashpos)+1; elseif str(strpos)=='(' closepar=findstr(')',str(strpos:len))+strpos-1; slash2=findstr('/',str(closepar(1):len+1))+closepar(1)-1; digits=slash2(1)-closepar(1)-1; [newplydir,slashnum]=parentheses(str(strpos:closepar(1)+1)); if slashpos==1 plydir=newplydir; else plydir=[plydir,newplydir]; end slashpos=slashpos+slashnum; strpos=slash(slashpos)+1; [y,z]=size(plydir); plypos=z+1; else plydir(plypos)=str2num(str(strpos:slash(slashpos)-1)); plypos=plypos+1; strpos=slash(slashpos)+1; end slashpos=slashpos+1; end if sym==1 if oddsym==1 [y,z]=size(plydir); plydir=[plydir(1:z-1),fliplr(plydir)]; else plydir=[plydir,fliplr(plydir)]; end end dir=plydir.'; Subroutine: parentheses function [plydir,slashnum]=parentheses(str) % PARENTHESES converts parenthetical shorthand ply angle notation % [plydir,slashnum]=PARENTHESES(str) outputs the ply direction plydir % and the number of slashes slashnum in the string str. % % See also DIRECTIONS len=length(str); str=strcat(str(1:len-2),'/',str(len-1:len)); mult=str2num(str(len+1)); slash=findstr('/',str); [z,slashnum]=size(slash); slashnum=slashnum-1; strpos=2; plypos=1; slashpos=1; while strpos
PAGE 105

92 if str(strpos:strpos+1)~='+-' & str(strpos:strpos+1)~='-+' plydir(plypos)=str2num(str(strpos:slash(slashpos)-1)); plypos=plypos+1; strpos=slash(slashpos)+1; elseif str(strpos:strpos+1)=='+-' plydir(plypos)=str2num(str(strpos+2:slash(slashpos)-1)); plydir(plypos+1)=-str2num(str(strpos+2:slash(slashpos)-1)); plypos=plypos+2; strpos=slash(slashpos)+1; else plydir(plypos)=-str2num(str(strpos+2:slash(slashpos)-1)); plydir(plypos+1)=str2num(str(strpos+2:slash(slashpos)-1)); plypos=plypos+2; strpos=slash(slashpos)+1; end slashpos=slashpos+1; end plydir1=plydir; for a=2:mult plydir=[plydir,plydir1]; end Subroutine: ABD function [K,Q,S,Qbar,Sbar,T,z,z0]=ABD(plydir,plythick,E1,E2,G12,nu12) % ABD calculates the laminate stiffness matrix of the form [A,B;B,D]. % All referenced equations are from Gibson, Principles of Composite Material Mechanics, 1994 % [K,Q,S,Qbar,Sbar,T,z,z0]=ABD(plydir,plythick,E1,E2,G12,nu12) % % Inputs: % plydir: string of the form '[90/+-45/(-+30)3/90bar]s' or column matrix of ply directions. Note % that the variables make more sense with the z-axis down; following this convention, the % first ply is the most negative. % plythick: column vector of ply thicknesses. If thick is a scalar, the thickness is applied to % all plys. % E1: column vector of Young's modulii in the fiber direction of each ply. If E1 is a scalar, % the modulus is applied to all plys. % E2: same as E1, in transverse directions. % G12: same as E1, but shear modulii. % nu12: same as E1, but Poisson's ratios. % % Outputs: % K: laminate stiffness (ABD) matrix % Q: ply stiffness. 3D array with Q(:,:,k) the stiffness matrix for the kth ply. % S: ply compliance. Also 3D array, same convention. % Qbar: transformed ply stiffness. Also 3D array, same convention. % Sbar: transformed ply compliance. Also 3D array, same convention.

PAGE 106

93 % T: transformation matrix. Also 3D array, same convention. Q, S, Qbar, Sbar, and T are % throughput for use by other functions. % z: column vector of z postions, kth entry is z postion of bottom of kth ply. % z0: z position of top of first ply (most negative). % % See also DIRECTIONS, PLANESTRESS %---------Housekeeping--------%Converts shorthand ply orientation to matrix if ischar(plydir) plydir=directions(plydir); end %Checks laminate property sizes and orients in column vector form [m,n]=size(plydir); [o,p]=size(plythick); if p ~= 1 plythick = plythick.'; o = p; end [q,r]=size(E1); if r ~= 1 E1 = E1.'; q = r; end [s,t]=size(E2); if t ~= 1 E2 = E2.'; s = t; end [u,v]=size(G12); if v ~= 1 G12 = G12.'; u = v; end [w,x]=size(nu12); if x ~= 1 nu12 = nu12.'; w = x; end %Expands uniform ply properties if o==1 plythick=ones(m,1)*plythick; elseif o~=m error('Thickness should be entered either as a scalar or as a vector of length n, where n is the number of plys in the laminate') end

PAGE 107

94 if q==1 E1=ones(m,1)*E1; elseif q~=m error('E1 should be entered either as a scalar or as a vector of length n, where n is the number of plys in the laminate') end if s==1 E2=ones(m,1)*E2; elseif s~=m error('E2 should be entered either as a scalar or as a vector of length n, where n is the number of plys in the laminate') end if u==1 G12=ones(m,1)*G12; elseif u~=m error('G12 should be entered either as a scalar or as a vector of length n, where n is the number of plys in the laminate') end if w==1 nu12=ones(m,1)*nu12; elseif w~=m error('nu12 should be entered either as a scalar or as a vector of length n, where n is the number of plys in the laminate') end %---------Computation----------%Sets z values for bottom of each ply. z0 is top of 1st ply. z0=-sum(plythick)/2; z(1)=z0+plythick(1); for k=2:m z(k)=z(k-1)+plythick(k); end %Calcs stiffnesses and compliances. Each is placed into a 3-D matrix, where Q(:,:,k) corresponds to the kth ply. for k=1:m [Q(:,:,k),S(:,:,k),Qbar(:,:,k),Sbar(:,:,k),T(:,:,k)]=planestress(E1(k),E2(k),G12(k),nu12(k),plydir(k)); end %Assembling A, B, and D. Eqs 7.38, 7.39, 7.40 for i=1:3 for j=1:3 A(i,j)=Qbar(i,j,1)*(z(1)-z0); B(i,j)=1/2*Qbar(i,j,1)*(z(1)^2-z0^2); D(i,j)=1/3*Qbar(i,j,1)*(z(1)^3-z0^3); for k=2:m A(i,j)=A(i,j)+Qbar(i,j,k)*(z(k)-z(k-1)); B(i,j)=B(i,j)+1/2*Qbar(i,j,k)*(z(k)^2-z(k-1)^2); D(i,j)=D(i,j)+1/3*Qbar(i,j,k)*(z(k)^3-z(k-1)^3); end

PAGE 108

95 end end %Putting it all together. K=[A,B;B,D]; Subroutine: planestress function [Q,S,Qbar,Sbar,T]=planestress(E1,E2,G12,nu12,theta) % PLANESTRESS calculates the stiffness and compliance matrices of a lamina % All referenced equations are from Gibson, Principles of Composite Material Mechanics, % 1994 except as noted. % % Inputs: % E1: Young's modulus in the fiber direction. % E2: Young's modulus in the transverse direction. % G12: Shear modulus. % nu12: Poisson's ratio. % theta: angle between ply fibers (1-direction) and x-direction, in degrees. % % Outputs: % Q: stiffness matrix in 12 (fiber) coordinates. % S: compliance matrix in 12 (fiber) coordinates. % Qbar: transformed stiffness matrix in xy (general) coordinates. % Sbar: transformed compliance matrix in xy (general) coordinates. % T: transformation matrix. %--------Computation------%Change theta to radians. theta=theta*pi/180; nu21 = nu12*E2/E1; %Calculating S and Q. Eqs 2.24, 2.4, 2.6 S=[1/E1, -nu12/E1, 0; -nu12/E1, 1/E2, 0; 0, 0, 1/G12]; Q=inv(S); %Q = [E1/(1-nu12*nu21), nu12*E2/(1-nu12*nu21), 0;nu12*E2/(1-nu12*nu21), E2/(1-nu12*nu21),0;0,0,G12]; %Transformation matrices. Eq 2.32, Ifju's class notes, 2/7/02 T=[(cos(theta))^2, (sin(theta))^2, 2*cos(theta)*sin(theta); (sin(theta))^2, (cos(theta))^2, -2*cos(theta)*sin(theta); -cos(theta)*sin(theta), cos(theta)*sin(theta), (cos(theta))^2-(sin(theta))^2]; R=[1,0,0;0,1,0;0,0,2]; %Calculating Qbar and Sbar. Ifju's class notes, 2/7/02 Qbar=inv(T)*Q*R*T*inv(R); Sbar=R*inv(T)*inv(R)*S*T;

PAGE 109

96 Subroutine: hygrothermal function [Nh,Mh,alpha,beta]=hygrothermal(deltaT,c,alpha12,beta12,Qbar,T,z0,z) % hygrothermal outputs the thermal and moisture induced forces and moments % on a laminate. % All referenced equations are from Gibson, Principles of Composite Material Mechanics, 1994 % Nh: hygrothermal force. % Mh: hygrothermal moment. % alpha: matrix of coeff thermal expansions in laminate coords. alpha is 3xn, % where n is the number of plies. The kth column is [alpha_x;alpha_y;alpha_xy] % for the kth ply. % beta: same as alpha, but coeff hygroscopic expansions. % deltaT: change in temperature from cure temp. % c: moisture concentration. % alpha12: matrix of coeffs of thermal expansion, in material coords. alpha12 % is 2xn or 3xn, where n is the number of plies. If alpha12 is 2x1 or % 3x1, the CTEs are assumed to apply to all plies. Row 1 is CTE in the % fiber direction, row 2 is CTE in the transverse direction, and row 3 % is 0. % beta12: matrix of coeffs of hygroscopic expansion. Same format as alpha12. % Qbar: transformed stiffness matrix. % T: transformation matrices, 3x3xn. % z0: z position of top of 1st ply. % z: column matrix of z positions of bottom of plies. %--------Housekeeping-------[m,n]=size(z); %determining number of plies [o,p]=size(alpha12); if o==2 alpha12(3,:)=0; end if p==1 for k=2:n alpha12(:,k)=alpha12(:,1); end end [o,p]=size(beta12); if o==2 beta12(3,:)=0; end if p==1 for k=2:n

PAGE 110

97 beta12(:,k)=beta12(:,1); end end %adding shear component to CTE and CHE matrices for calculations % and expanding uniform ply properties %--------Computation-------for k=1:n alpha(:,k)=inv(T(:,:,k))*alpha12(:,k); beta(:,k)=inv(T(:,:,k))*beta12(:,k); end alpha(3,:)=alpha(3,:)*2; beta(3,:)=beta(3,:)*2; %calculating laminate direction CTEs and CHEs--note that % alpha12*inv(T)=[alpha_x;alpha_y;alpha_xy/2], and the third row % must be multiplied by 2 to give the shear CTE. Same for beta. % Eq. 5.22 Nt = deltaT*Qbar(:,:,1)*alpha(:,1)*(z(1)-z0); Nm = c*Qbar(:,:,1)*beta(:,1)*(z(1)-z0); Mt = deltaT/2*Qbar(:,:,1)*alpha(:,1)*(z(1)^2-z0^2); Mm = c/2*Qbar(:,:,1)*beta(:,1)*(z(1)^2-z0^2); %0-index workaround again...see below for k=2:n Nt = Nt + deltaT*Qbar(:,:,k)*alpha(:,k)*(z(k)-z(k-1)); Nm = Nm + c*Qbar(:,:,k)*beta(:,k)*(z(k)-z(k-1)); Mt = Mt + deltaT/2*Qbar(:,:,k)*alpha(:,k)*(z(k)^2-z(k-1)^2); Mm = Mm + c/2*Qbar(:,:,k)*beta(:,k)*(z(k)^2-z(k-1)^2); end % calculating thermal and hygroscopic forces and moments. Eqs 7.87-7.91 Nh=Nt+Nm; Mh=Mt+Mm; % summing for hygrothermal forces and moments

PAGE 111

LIST OF REFERENCES 1. Bert, C. W., and Thompson, G. L., A Method for Measuring Planar Residual Stresses in Rectangularly Orthotropic Materials, Journal of Composite Materials, Vol. 2, No. 2, April, pp. 244-253, 1968. 2. Carman, G. P., and Sendechyj, G. P., Review of the Mechanics of Embedded Optical Sensors, Journal Composites Technology & Research, Vol. 17, No. 3, pp. 183-193, 1995. 3. Chapman, T. J., Gillespie, J. W., Pipes, R. B., Manson, J. E., "Prediction of Process-Induced Residual Stresses in Thermoplastic Composites, Journal Composite Materials, Vol. 24, pp. 616-643, 1990. 4. Choi, S., Sankar, B.V., A Micromechanics Method to Predict the Microcracking of the LH 2 Composite Tank at Cryogenic Temperature, Department of Mechanical and Aerospace Engineering, University of Florida, 2003. 5. Daniel, I. M., and Liber, T., Effect of Laminate Construction on Residual Stresses in Graphite/Polyimide Composites, Experimental Mechanics, Vol. 1, pp. 21-25, 1977. 6. Daniel, I. M., and Liber, T., Lamination Residual Stresses in Fiber Composites, IITRI Report D6073-I, for NASA-Lewis Research Center, NASA CR-134826, 1975. 7. Daniel, I. M., Liber, T., and Chamis, C. C., Measurement of Residual Strains in Boron/Epoxy and Glass/Epoxy laminates, Composite Reliability, ASTM STP 580, American Society for Testing and Materials, pp. 340-351, 1975. 8. Daniel, I. M., Wang, T., Determination of Chemical Cure Shrinkage in Composite laminates, Journal of Composites Technology & Research, Vol. 12, No. 3, pp. 172-176, 1989. 9. Daniel, I. M., Wang, T. M., Karalekas, D. and Gotro, J. T., Determination of Chemical Cure Shrinkage in Woven_Glass/Epoxy Laminates, Annual Technical Conference-Society of Plastics Engineers, New York, USA, May 1-4, pp. 632-634, 1989. 10. Fakuda, H., Takahashi, K., and Toda, S., Thermal Deformation of Anti-Symmetric Laminates At Cure, Proceedings of ICCM-10, Whistler, B.C., Canada, August, Vol.3, pp. 141-148, 1995. 98

PAGE 112

99 11. Fenn, R. H., Jones, A. M., and Wells, G. M., X-Ray Diffraction Investigation of Triaxial Residual Stresses in Composite Materials, Journal Composite Materials, Vol. 27, No. 14, pp. 1338-1351, 1993. 12. Final Report of the X-33 Liquid Hydrogen Tank Test Investigation Team, NASA George C Marshall Space Flight Center, Huntsville, AL 35812, May 2000. 13. Gibson, R. F., Principles of Composite Material Mechanics, McGraw-Hill, New York, New York, 1994. 14. Griffith, A.A., The Phenomena of Rupture and Flow in Solids, Philosophical Transactions of the Royal Society, Vol. 221A, pp. 163-198, 1920. 15. Groover, M.P., Fundamentals of Modern Manufacturing, Prentice Hall, Upper Saddle River, New Jersey, 1996. 16. Hahn, H. T., Residual Stresses in Polymer Matrix Composite Laminates, Composite Materials, Vol. 10, pp. 265-277, 1976. 17. Hyer, M. W., Calculation of Room-Temperature Shapes of Unsymmetrical Laminates, Journal Composite materials, Vol. 15, July, pp. 296-310, 1981. 18. Ifju, P.G., Myers, D., Schulz, W., Residual Stress and Thermal Expansion of Graphite Epoxy Laminates Subjected to Cryogenic Temperatures, Proceedings of the American Society for Composites 18 th Technical Conference, October 19-23, 2003. 19. Ifju, P.G., Haftka, R., Myers, D., Schulz, W., Singer, T., Statistics Properties and Reliability Based Optimiztion of Laminates for Cryogenic Environments, AIAA-2003-1935, 44th AIAA/ASME/ASCE/AHS Structures, Structural Dynamics, and Materials Conference, 7-10 April 2003. 20. Ifju, P.G., Niu, X., Kilday, B.C., Liu, S.C., Ettinger, S.M., Residual Strain Measurement in Composites Using the Cure-referencing Method, Journal of Experimental Mechanics, Vol. 40, No. 1, pp. 22-30, 2000. 21. Ifju, P.G., Niu, X., Kilday, B.C., Liu, S.C., "A Novel Means to Determine Residual Stress In Laminated Composites," Journal of Composite Materials, Vol. 33, No. 16, 1999. 22. Jain, L. K., and Mai, Y. W., On Residual Stress Induced Distortions during Fabrication of Composite Shells, Proceedings of the American Society for Composites-Tenth Technical Conference, Santa Monica, CA, USA, Oct. 18-20, pp.261-270, 1995. 23. Jeronimidis, G., and Parkyn, A. T., "Residual Stresses in Carbon Fiber-Thermoplastic Matrix Laminates, Journal Composites Materials, Vol. 22, pp. 401-415, May 1988.

PAGE 113

100 24. Joh, D., Byun, K. Y., and Ha, J., Thermal Residual Stresses in Thick Graphite/Epoxy Composite Laminates-Uniaxial Approach, Experimental Mechanics, Vol. 33, No., pp. 70-76, 1993. 25. Kam, T. Y., Nonlinear and First-Ply Failure Analyses of laminated Composite Cross-Ply Plates, Journal of Composite Materials, Vol. 29, No.4, pp. 463-482, 1995. 26. Kim, K. S., and Hahn, H.T., Residual Stress Development during Processing of Graphite/ Epoxy Composite, Composite Science and Technology, 36, pp. 121-132, 1989. 27. Kim, R. Y., and Hahn, H. T., Effect of Curing Stresses on the First Ply-failure in Composite Laminates, Journal Composite Materials, Vol. 13, January, pp. 2-16, 1979. 28. Lake, B. R., Appl, F. J., and Bert, C. W., An Investigation of the Hole-drilling Technique for Measuring Planar Residual Stress in Rectangularly Orthotropic Materials, Experimental Mechanics, Vol. 10, June, pp. 233-239, 1970. 29. Lawrence, C., Nelson, D.V., Bennett, T.E., Spingarn, J.R., Embedded Fiber Optic Sensor Method for Determining Residual Stresses in Fiber-Reinforced Composite Materials, Journal of Intelligent Material Systems and Structures, Vol. 9, pp. 788-799, 1999. 30. Lee, J., and Czarnek, R., Measuring Residual Strains in Composites Laminates Using Moir Interferometry, Proceedings of the 1991 SEM Conference on Experimental Mechanics, Milwaukee, WI, USA, June 1-3, pp. 405-415, 1991. 31. Lee, J., Czarnek, R., and Guo, Y. Interferometric Study of Residual Strains in Thick Composites, Proceedings of the 1989 SEM Conference on Experimental Mechanics, Cambridge, MA, USA, June 7-10, pp. 356-364, 1989. 32. Mnson, J. E., and Seferis, J.C. Process Simulated Laminate (PSL) A Methodology to Internal Stress Characterization in Advanced Composite Materials, Journal Composite Materials, Vol. 26, No. 3, pp. 405-431, 1992. 33. Mathar, J., Determination of Initial Stress by Measuring the Deformations Around Drilled Holes, Transaction of ASME, Vol. 56, p.249, 1934. 34. Measurement of Thermal Expansion Coefficient, Tech Note TN-513, Measurements Group, pp. 1-24. 35. Measurement of Residual Stresses by the Hole-Drilling Strain-Gage Method, Tech Note TN-503, Measurement Group, pp. 1-15, 1988.

PAGE 114

101 36. Nicoletto, G., Moir Interferometry Determination of Residual Stresses in the Presence of Gradients, Experimental Mechanics, Vol. 31, No. 3, pp. 252-256, 1991. 37. Niu, X., Process Induced Residual Stresses and Dimensional Distortions in Advanced Laminated Composites, University of Florida, 1999. 38. Park, J., Lee, S., Lee, J., Nondestructive Sensing Evaluation of Single Carbon Fiber/Polymer Composites Using Electro-Micromechanical Technique, International SAMPE Symposium and Exhibition (Proceedings), Vol. 47, pp190-203, 2002. 39. Post, D., Han, B., and Ifju, P. G., High Sensitivity Moir: Experimental Analysis for Mechanics and Materials, Springer Verlag, New York, 1994. 40. Predecki, P., and Barrett, C. S., Stress Measurement in Graphite/Epoxy Composites by X-Ray Diffraction from Fillers, Journal Composite Materials, Vol. 13, pp. 61-71, 1979. 41. Qu, X., Venkataraman, S., Haftka, R.T., Johnson, T.F., Reliability, Weight, and Cost Tradeoffs in the Design of Composite Laminates for Cryogenic Environments, AIAA Paper 2001-1327, Proceedings 42nd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Material Conference, Seattle, WA, April, 2001. 42. Rendler, N. J., and Vigness, I., Hole-drilling Strain-gage Method of Measuring Residual Stresses, Experimental Mechanics, pp. 577-586, December 1966. 43. Schajer, G. S., and Yang, L., Residual-stress Measurement in Orthotropic Materials Using the Hole-drilling Method, Experimental Mechanics, Vol. 34, No. 4, pp. 324-333, 1994. 44. Standard Test Method for Tensile Properties of Polymer Matrix Composite Materials, ASTM Standards D3039, Vol. 15.03, pp 98-108. 45. Standard Test Method for Shear Properties of Composite Materials by the V-Notched Beam Method, ASTM Standards D5379, Vol. 15.03, pp 226-238. 46. Sunderland, P., Yu, W. J., and Mnson, J. E., "A Technique for the Measurement of Process-Induced Internal Stresses in Polymers and Polymer Composites, Proceedings of ICCM-10, Whistler, B. C., Canada, August, Vol. 3, pp. 125-132, 1995. 47. Tanaka, N., Okabe, Y., Takeda, N., Temperature-Compensated Strain Measurement Using FBG Sensors Embedded in Composite Laminates, Proceedings of SPIE-The International Society for Optical Engineering, Vol. 4694, pp. 304-313, 2002.

PAGE 115

102 48. Polymer Matrix Composites: Materials, Usage, Design, and Analysis, The Composite Materials Handbook Mil 17, Vol. 3, pp. 9.1-9.12, ASTM, West Conshohocken, PA, 2002. 49. Unger, W. J., and Hansen, J. S., The Effect of Thermal Processing on Residual Strain Development in Unidirectional Graphite Fiber Reinforced PEEK, Journal Composite Materials, Vol. 27, No. 1, pp. 59-81, 1993. 50. Wang, S., Chung, D.D.L., Electrical Behavior of Carbon Fiber Polyer-Matrix Composites in the Through-Thickness Direction, Journal of Materials Science, Vol. 35, pp. 91-100, 2000. 51. Wang, T.-M., Daniel, I. M., and Gotro, J. T., Thermoviscoelastic Analysis of Residual Stresses and Warpage in Composite Laminates, Journal of Composite Materials, Vol. 26, No. 6, pp. 883-899, 1992

PAGE 116

BIOGRAPHICAL SKETCH Donald Myers was born on August 26, 1978, in North Miami Beach, Florida. He lived in Miramar, Florida, an area between Miami and Ft. Lauderdale, where he attended Miramar High School and graduated as Valedictorian in the spring of 1996. Later that year he began his undergraduate education at the University of Florida in the Department of Aerospace Engineering, Mechanics, and Engineering Science. He graduated with honors in May 2001 with a Bachelor of Science in Aerospace Engineering. After obtaining the B.S., he continued his education at the University of Florida in August of 2001 pursuing a Master of Science from the Department of Mechanical and Aerospace Engineering. He served as a Graduate Research Assistant under Dr. Bhavani Sankar in the Advanced Center for Composites Laboratory, from August 2001 till May of 2002. In May of 2002 he transitioned to the Experimental Stress Analysis Laboratory where he has been serving as a Graduate Research Assistant under the tutelage of Dr. Peter Ifju. While in the Experimental Stress Analysis Lab he has been conducting research into material characterization of graphite/epoxy composites involved in the storage of cryogenic fuels. 103


Permanent Link: http://ufdc.ufl.edu/UFE0003180/00001

Material Information

Title: Method for Measurement of Residual Stress and Coefficient of Thermal Expansion of Laminated Composites
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0003180:00001

Permanent Link: http://ufdc.ufl.edu/UFE0003180/00001

Material Information

Title: Method for Measurement of Residual Stress and Coefficient of Thermal Expansion of Laminated Composites
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0003180:00001


This item has the following downloads:


Full Text












METHOD FOR MEASUREMENT OF RESIDUAL STRESS AND COEFFICIENT OF
THERMAL EXPANSION OF LAMINATED COMPOSITES















By

DONALD G. MYERS II


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA


2004

































Copyright 2004

by

Donald G. Myers II
































This thesis is dedicated to my family, loved ones, and in the memory of my father.















ACKNOWLEDGMENTS

I would like to thank my advisor, Dr. Peter Ifju, for all of his support and advice. I

would also like to thank my other committee members, Dr. Raphael Haftka and Dr.

Bhavani Sankar, for their advice. Thomas Singer, William Schulz, and Lucian Speriatu

were fellow assistants in the Experimental Stress Analysis Lab that provided invaluable

assistance in my efforts. I would also like to extend a thank you to Dr. Leishan Chen for

all of his assistance. Together all of these people along with my various professors have

made my time at the University of Florida a wonderful educational experience. Finally I

would like to say thank you to my entire family and loved ones for everything they have

done for me throughout my life.
















TABLE OF CONTENTS



A C K N O W L E D G M E N T S ................................................................................................. iv

LIST OF TABLES .......................................... viii

LIST OF FIGURES ....................................... ........... ............................ ix

ABSTRACT ................................................... ................. xii

CHAPTER

1 IN TR O D U C T IO N ........ .. ......................................... ..........................................1.

1.1 H historical B background .................................................................... .............. ....
1.2 F failure of the X -33 Fuel T ank ........................................................... ...............2...
1.3 R esidual Strains and Stresses............................................................. ...............6...
1.4 Structural Reliability..................................................................... .7
1.5 R research O bjectives... ................................................................... ... .......... ... 9


2 RESIDUAL STRESS CHARACTERIZATION TECHNIQUES...........................10

2 .1 In tro d u ctio n ...........................................................................................................1 0
2.2 D destructive Testing Techniques ...................................................... ................ 11
2.2.1 H ole-D killing M ethod ...................................................... ............... 11
2.2.2 Cutting M ethod ....................................................................... 12
2.2.3 Ply Sectioning M ethod ....................... ............................................... 12
2.2.4 First Ply Failure .. .. ...... ............ .............................................. 13
2.3 N on-D destructive Techniques ........................................................... ................ 13
2 .3 .1 W a rp a g e ......................................................................................................1 3
2.3.2 E m bedded Strain G ages ........................................................ ................ 14
2.3.3 X -ray D iffraction .................................................................................. 15
2 .3.4 E electrical R esistance.............................................................. ............... 15
2.3.5 Cure R eferencing M ethod ....................... .......................................... 15


3 CURE REFEREN CING M ETH OD ...................................................... ................ 17

3.1 Selected L am inmates ............................ .......................................... 17


v









3.2 C ure R eferencing M ethod....................................... ...................... ............... 18
3.3 M oire Interferom etry ................................................................... ............... 19
3.4 D iffraction G ratings ...................................................................... ................ 20
3 .5 C during P procedure ... ..................................................................... ...............22
3.6 M oire M easurem ent Technique....................................................... ................ 27
3 .7 D ata A n a ly sis ........................................................................................................2 9
3 .8 E rro r S o u rc e s ........................................................................................................ 3 0
3 .9 C o n c lu sio n s ...........................................................................................................3 2


4 STRAIN GAGE METHODS ....................... ........................................................34

4 .1 A acquisition T heory .......................................................................... ...............34
4.2 Specim en Preparation ........................................................................................35
4 .3 T testing P rocedu re ............................................................................ ...............3 8
4 .3 .1 C ry og enic T ests ..................................................................... ...............39
4.3.2 Elevated Temperature..............................................................................42
4.4 Data Analysis and Results .................................................................. ...............44
4 .5 E rro r S o u rc e s ........................................................................................................4 7
4 .6 C o n c lu sio n s ...........................................................................................................4 8


5 MATERIAL PROPERTY TESTING....................................................... ...............51

5 .1 E 1 T e stin g ..............................................................................................................5 1
5 .2 E 2 T e stin g ..............................................................................................................5 2
5 .3 G 12 T e stin g ............................................................................................................5 4
5.4 Data Analysis and Results .................................................................. ...............55
5.5 Micro-mechanical Methods................................................................ ...............58
5 .6 E rro r S o u rc e s ........................................................................................................5 9
5 .7 C o n c lu sio n s ...........................................................................................................5 9


6 PLY LEVEL RESIDUAL STRAINS AND STRESSES ........................................... 61

6.1 Residual Strain Calculation ...............................................................................61
6 .2 R e sid u al S tre ss ......................................................................................................6 2
6 .3 C o n c lu sio n s ...........................................................................................................6 4


7 CLASSICAL LAMINATION THEORY ................................................................65

7.1 Theory Development ............................................65
7.2 Experimental vs. Analytical Results ..................................................................72
7 .3 C o n c lu sio n s ...........................................................................................................7 6









APPENDIX

A CRM IM A GES OF SPECIM EN S......................................................... ................ 77

B M A TL A B CL T C O D E ...................................................................... ................ 85

L IST O F R EFE R E N C E S .... ....................................................................... ................ 98

BIO GRAPH ICAL SK ETCH .... .............................................................. ............. 103















LIST OF TABLES


Table page

3-1 Total surface strain on unidirectional laminates at 24C..................................30

3-2 Total surface strain on optimized angle ply at 24C.........................................30

3-3 Total surface strain on RLV laminate at 240C.................................. ................ 31

4-1 Average COV across tem perature range........................................... ................ 49

4-2 Unidirectional chemical shrinkage induced strain...........................................50

4-3 OAP chem ical shrinkage induced strain........................................... ................ 50

4-4 RLV chem ical shrinkage induced strain........................................... ............... 50

5-1 R results of each tension test ...................................... ...................... ................ 58

5-2 Comparison of experimental measurements and micro-mechanics values ...........59















LIST OF FIGURES

Figure page

3-1 Schematic of a four-beam moire interferometer ..............................................20

3-2 Image of moire interferometer used in testing..................................................20

3-3 R eplication schem atic .................................................................... ................ 23

3-4 T ool separation device ................................................................... ................ 24

3-5 Teflon coated epoxy film casting device ............... ....................................24

3-6 Autoclave oven ...................... .. ........... ............................. 26

3-7 Vacuum bag assembly schem atic ................................................... 26

3-8 IM 7/977-2 cure cycle ..................................................................... ................ 28

3-9 M oire grating on specim en ...................................... ...................... ................ 28

4-1 Schem atic of strain gage alignm ent.................................................. ................ 37

4-2 G auging of sectioned specim en ........................................................ ................ 37

4-3 G auging of reference m aterials......................................................... ................ 38

4-4 Environmental chamber and LN2.....................................................40

4-5 Laminate coupons wired and placed inside environmental chamber .................40

4-6 Q uarter-bridge schem atic........................................ ....................... ................ 4 1

4-7 U nidirectional lam inate surface strain .............................................. ................ 45

4-8 O A P lam inate surface strain ............................................................. ................ 46

4-9 R L V lam inate surface strain ............................................................. ................ 46

4-10 A average of lam inate surface strains.................................................. ................ 49

4-11 Coefficient of therm al expansion...................................................... ................ 50









5-1 E i and E 2 testing fi xture ......................................... ........................ ................ 53

5-2 Shear test specim en schem atic.................. .................................................... 56

5-3 Shear Testing Fixture .................................................................. ................ 56

5-4 E i stress/strain curves .............. .... .............. .............................................. 57

5-5 E 2 stress/strain curves .............. .... .............. .............................................. 57

5-6 G 12 stress/strain curves.......................................... ......................... .............. 58

6-1 R esidual strain on selected plies ....................................................... ................ 63

6-2 Residual stress level on selected plies ..................................................63

7-1 L am inate coordinate system ............................................................. ................ 67

7-2 Layer schem atic and nom enclature................................................... ................ 67

7-3 Unidirectional CLT and experimental results...................................................73

7-4 OAP CLT and experimental results..................................................................73

7-5 RLV CLT and experimental results..................................................................74

7-6 Unidirectional CLT shifted for chemical shrinkage .........................................74

7-7 OAP CLT shifted for chemical shrinkage ........................................................75

7-8 RLV CLT shifted for chemical shrinkage ........................................................75

A -1 U nidirectional L am inate R un 1 ......................................................... ................ 77

A -2 U nidirectional L am inate R un 2......................................................... ................ 78

A -3 U nidirectional L am inate R un 3......................................................... ................ 78

A -4 U nidirectional L am inate R un 4......................................................... ................ 79

A -5 U nidirectional L am inate R un 5......................................................... ................ 79

A -6 O A P L am inate R un 1 .................................................................... ................ 80

A -7 O A P Lam inate Run 2 .................................................................. ................ 80

A -8 O A P L am inate R un 3 ..................................................................... ................ 81

A -9 O A P Lam inate R un 4 .................................................................. ..................... 81









A-10 OAP Laminate Run 5 .................................................................. ................ 82

A -11 R L V lam inate R un 1 .. ..................................................................... ................ 82

A -12 R L V L am inate R un 2 ..................................................................... ................ 83

A-13 RLV Laminate Run 3 .................................................................. ................ 83

A -14 R L V L am inate R un 4. ..................................................................... ................ 84















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

METHOD FOR MEASUREMENT OF RESIDUAL STRESS AND COEFFICIENT OF
THERMAL EXPANSION OF LAMINATED COMPOSITES

By

Donald G. Myers II

May 2004

Chair: Peter G. Ifju
Major Department: Mechanical and Aerospace Engineering

The development of residual stresses in laminated composites is a very important

area of extensive research. Residual stress can dramatically reduce the strength of a

component built of a laminated composite. The existence of these residuals stresses and

strength degradation can lead to catastrophic material failures as seen in the X-33. A

method has been proposed to measure and quantify these stresses as well as determine the

thermal expansion coefficient of laminated composites as a function of temperature. This

method relies on the combination of existing techniques in a manner that will allow for

the acquisition of stress values across a wide range of temperatures. The first technique

employed is a novel strain measurement method know as the Cure Referencing Method.

This technique is based on transferring a moire grating onto a specimen during the cure

process. The grating becomes a part of the specimen at cure temperature during the

polymerization of the resin and therefore carries all strain information from the curing

process including temperature dependent strains as well as chemical shrinkage









information. The tool with which the grating was transferred to the specimen is used as a

reference from which all relative deformations are calculated. This process is

implemented on laminates of three differing stacking sequences, each of which provides

its own unique data point. The results from this method provide data points from which

the next technique references the strains and stresses as a function of temperature. After

strain measurements are taken from the moire specimen, they are fitted with strain gages

in the x and y directions. Strain values are observed as a function of temperature as the

specimens are exposed to temperatures between cure and cryogenic conditions. The

strain gage readings are compensated for temperature effects and are referenced to the

strain values at room temperature given by the cure referencing method. The resulting

values of strain represent the total surface strain at any given temperature for the various

laminates. The surface strain on the unidirectional laminate represents the free expansion

state any lamina would reach if not constrained. This value is used to determine residual

strain present in different layers of the multidirectional laminates. The coefficient of

thermal expansion is also calculated as a function of temperature from the strain gage

readings. Basic material testing is also conducted and the values of longitudinal,

transverse, and shear moduli are presented. Variability in these measurements is also

reported in conjunction with other work at the University of Florida to perform reliability

based optimization. Once the residual strain and material properties are calculated they

are transformed into stress by the usual stress-strain relations. The final ply level residual

stress is plotted as a function of temperature. The results of surface strain measurements

are compared to classical lamination theory analytical predictions.














CHAPTER 1
INTRODUCTION

Throughout history one of the primary goals of engineers has been to develop

structures and materials that are lighter, stronger, and tougher than those that already

exist. The introduction of advanced structural composites (ASCs) has helped to achieve

that goal. Composite materials are those that consist of two or more separate materials

that are combined macroscopically into a structural unit [13]. The prevalence of these

composite materials has skyrocketed since the 1960s and are now used in structures as

advanced as satellites, space structures, and military aircraft and in everyday sporting

goods around the house such as skis and golf clubs.

1.1 Historical Background

Despite the boom in significance since the development of ASCs in the 1960s,

basic man-made composite materials have been around for thousands of years. Some

examples can be seen as early as the times of the great Egyptian kingdoms when brick

was made of clay and straw, and in South and Central America where the people used

plant fibers in native pottery [13]. Both of these are examples of one of the most

predominant composite materials, fiber reinforced composites. Other examples of fiber-

reinforced composites include glass/epoxy, steel-reinforced concrete, and graphite/epoxy,

the subject of this investigation.

The motivation for the use of fiber-reinforcement in materials comes from research

by Griffith published in 1920 that demonstrated that some materials are stronger and









stiffer as fibers than in bulk form [14]. Some reasons for this increase in strength include

less material flaws, orientation of molecular chains, and lower dislocation densities.

Along with all the obvious advantages, advanced fiber-reinforced composite

materials have their shortcomings as well. A primary performance weakness is that most

of the strength exists in the direction of the fiber and the material performs poorly in

transverse directions. Transverse reinforcement is necessary to make the material useful

and this is done usually by orienting the fibers at various angles.

Using composite materials also poses other problems for designers. These

materials are much more difficult to analyze and characterize. Many of the design and

analysis techniques that have been used for years on other materials cannot be applied

directly to composite materials. Because they consist of more than one material they are

often orthotropic in nature. Also, once several layers of lamina are assembled into a

laminate, to correct the transverse weakness or tailored to achieve desired directional

strengths, a complex interaction between layers develops. This is a macroscopic issue

while on the microscopic level there is also a complex interaction that takes place within

each lamina between the fiber and matrix material.

Due to these analysis difficulties, the behavior of composite materials is not fully

understood and this lack of understanding can lead to devastating mechanical failures. It

is the goal of this paper to detail a combination of methods that can be used to

characterize a material system and then provide a detailed set of data and information on

the IM7/977-2 material system.

1.2 Failure of the X-33 Fuel Tank

For several years, efforts have been underway to develop a reusable launch vehicle

(RLV) to replace the aging space shuttle fleet. The goal is to develop a vehicle that is









entirely self-contained without the need of external fuel and propulsion components such

as the external fuel tank and solid rocket boosters seen on the current space shuttles. The

vehicle is to be able to takeoff, perform its mission in space, and return to earth entirely

under its own power. The development of the X-33 was an attempt to achieve these

goals.

The design of the X-33 called for an internal liquid hydrogen (LH2) fuel tank [12].

This tank was to be an integrated part of the structure designed to provide both fuel

storage and carry structural loads. One of the most important factors in the design of an

RLV is to reduce the takeoff weight as much as possible. Manufacturing a tank

consisting of ASCs was seen as the best way to do this.

The X-33's fuel tank and aft fuselage were designed to consist of two LH2 fuel

tanks, with each composed of several "lobes" made of graphite/epoxy and having bonded

woven composite joints. Each lobe was a sandwich panel with face sheets consisting of

the IM7/977-2 material system and a Korex core. The tanks were to store the LH2 fuel

as well as be part of the primary aft structure designed to distribute thrust, inertial,

landing gear, and aero-control surface loads.

The first of two tanks was designed and delivered by Lockheed Martin to NASA at

the Marshall Space Flight Center for preflight testing. The preflight testing was designed

to verify that the tank was able to withstand the cryogenic loading conditions of -423 F

(200 K) as well as the expected flight loads. The planned testing sequence consisted of:

1. A partial liquid nitrogen (LN2) fill and proof test to 32 psig, followed by a leak
check at 34 psi of gaseous helium

2. A LH2 fill and structural load application. The test was to consist of six conditions:
a proof pressure test at 42 psig and five loading tests at 5 to 42 psig

3. A final 34 psig gaseous helium test









It would be seen soon after testing began that the design was not feasible for its intended

purpose.

In early November of 1999 the fuel tank began the first combined mechanical

loading and LH2 storage test. It had been previously been filled and leak tested with LN2

and LH2 in the absence of external loads. This time the initial pressure and loading tests

were successful and the tank was drained of LH2. Fifteen minutes after the tank was

drained a major mechanical failure took place as the outer facesheet and core material

debonded from the inner facesheet. The following lists the observations that were made

during the test and time leading up to the failure.

The first observation was that at the time of fueling there was an expected decrease

in the pressure of the sandwich core resulting from the temperature change within the

tank. The second observation was an unanticipated increase in the core pressures as the

tank neared full capacity and pressurization. A third observation was the continued

increase in core pressures as the tank was drained. This event was expected, as the

temperature increase would reverse the initial trend seen in observation one and pressure

should have returned to ambient conditions. Fourth was the incident of the composite

sandwich failure. At the time of the failure core pressures were between 50 and 60 psia.

These pressures rapidly dropped to atmospheric as the region failed. Finally, the

remaining sections of the core that did not experience a failure had internal pressures

above the expected ambient a full 13 hours later.

Review and analysis of the data resulted in the following failure sequence. The rise

in pressure in the second observation was most likely the result of seepage of H2 into the

core through the inner facesheet. The method of transportation was though mirocracks









that had developed in the inner facesheet due to the stresses and strains cause by the

cryogenic environment. As the tank was filled with LH2, the cooling caused a vacuum in

the core, which helped to pull in GH2 until the area was covered with the liquid. LH2 was

then also able to penetrate the facesheet through the microcracks. As the tank was

emptied some of the fluid was able to travel back through the microcracks and be

removed from the tank. However, before all of the fluid had passed back into the interior

of the tank the pressure and strain decreased closing the microcracks and trapping some

of the fluid inside the core. As the temperature continued to climb the pressure of the

vaporizing H2 resulted in a buildup until the outer facesheet and core debonded from the

inner facesheet.

The properties of the laminates were estimated from existing databases. Also it

was assumed that 5000 microstrain (,s) was an adequate allowable level to limit

permeability. The above failure observed in the X-33 project is a perfect example of how

the behavior of ASCs are not fully understood and how this lack of information can result

in catastrophic failures. The design group failed to adequately cover combined

thermal/mechanical effects. It is this thermal effect, or processed induced residual

effects, that are not fully understood and result in considerable structural issues.

The failure of the X-33 and what it means to the future of composite

implementation is the driving force behind this thesis. This work will propose combined

methods that can be used to better evaluate the IM7/977-2 of the X-33 and other

composite material properties as they are subjected to extreme thermal conditions. This

material characterization can be used to remove the some of the assumptions that are









made in current composite design and replace them with concrete experimentally

determined values.

1.3 Residual Strains and Stresses

Root failure of the X-33 was most likely the failure to understand and accurately

predict the residual strains and stresses that develop in a composite as a result of curing

and the exposure to changing temperatures, especially low temperatures. When a

polymeric fiber reinforced composite is manufactured it usually undergoes a process

where the resin is heated, the fibers are wetted, and cure takes place at high temperatures

[37]. During this process and subsequent conditions residual strains and stresses will

develop within the laminate.

The developing stresses arise on two different scales. Stresses will occur that

develop between the fiber and matrix in each lamina of a composite. The study of these

effects is termed micro-mechanics, where a laminate is examined on the fiber scale.

Another area of interest is the examination of what happens to a composite on the macro-

mechanical or ply scale. It is on this scale that the following research will focus.

On the ply scale residual strains and stress develop due to the mismatch in the

thermal expansion coefficients of the fiber and matrix as well as the development of

chemical shrinkage. As a lamina is cured its matrix constituent undergoes

polymerization. Epoxy resins undergo condensation polymerization, where two reacting

monomers are brought together to form a new molecule of the desired compound [13]. In

the case of many ASCs this is a two-step process. The production of prepreg tape

consists of wetting the fibers and allowing the matrix to partially cure. The

polymerization process concludes when these prepreg materials are arranged into the

desired stacking sequences and then are heated to the desired cure temperature. As this









process takes place the matrix undergoes a volumetric change known as chemical

shrinkage while the fibers remain volumetrically unchanged. This event adds to the

mismatch in expansion of the fibers and matrix, where the matrix is subject to higher

expansion and contraction than the fiber.

Residual stresses develop when the desired expansion of the lamina is restricted.

As the angles of the lamina are varied from ply to ply, the lower contraction of the fiber

will constrain the contraction of the matrix. When temperatures are decreased, as in the

case of the X-33, the matrix desires to contract but instead is subjected to tensile stress as

this deformation is opposed. If all of the fibers are aligned with each other there is no

stress present on the ply scale. A cross-ply [0/90] stacking sequence will result in the

highest level of residual stress.

The residual stresses that develop during processing and operating conditions are

not negligible. They can consume large amounts of a laminate's strength. If these

stresses are not accurately understood they can lead to drastic material failures as in the

case of the X-33, where the tensile stresses that developed in the matrix material likely

exceeded its critical tensile strength. Once this occurred, micro-cracks were able to

develop allowing the seepage the hydrogen into the core. The development of

microcracks in other structures will allow the fibers to be exposed to possibly degrading

environmental conditions as well as possible chemical attack in storage facilities.

1.4 Structural Reliability

The existence of the overwhelming thermal stresses that resulted in the failure of

the X-33 demonstrates the need to better understand the nature of composite materials

and appropriately design for their implementation. The examination of structural









reliability of a component during the design process is a step that must be taken to

prevent such failures.

Previously, the design process worked to build reliability based on a factor-of-

safety approach [48]. This is a deterministic approach to the issue of reliability where

worst case scenarios are proposed, investigated, and accounted for by adding a safety

margin in the design. This approach has been used for years in metallic structures and

can provide adequate reliability for composites but not to the same degree. This is

because composite material performance exhibits a larger statistical distribution and

variation than metals. Reliability-based design methods are being developed to account

for this problem.

Probabilistic design methods are moving to the forefront in composite design.

Probabilistic design is an integrated process that works to define and develop functional

relationships between various design factors and material properties. A statistical

variation in one variable will have consequences to the others and those changes can be

designed for. The probabilistic approach uses statistical characterization instead of

proposed worst case values to define a variable and determine its magnitude and

frequency. The amount of data and how well the variables are defined will influence the

extreme values.

A typical probabilistic design approach has the following fundamental elements:

1. Identify all possible uncertain variables on all scales of the structure, including all
variables at the constituent scale, all stages of fabrication and assembly, and the
possible applied loads.

2. Assign a probabilistic distribution function for each variable.

3. Process all random variables through and analyzer that can handle micro- and
macro-mechanics laminate theories, structural mechanics, and probability theories.









4. Extract useful information from the analysis and compare them to probabilistic
design parameters.

To be able to carry out the above procedure the following aspects must be
characterized as random variables:
* Material mechanical properties
* External loads during operation
* Manufacturing process variability
* Environmental effects
* Environmental history during operation
* Effect of flaws or damage locations
* Predictive Accuracy

The probabilistic approach to analysis and design has become a significant area of

academic interest. Qu et al have investigated areas of interest that are particularly of

importance to the future of RLV design. They performed reliability-based optimization

of composite laminates for cryogenic environments [41]. It was found that under these

conditions that the optimized weight was sensitive to transverse lamina strength and

thermal stresses as they are related to CTE and lay-up sequence. When the confidence in

these values is low the thickness of the laminate is significantly greater leading to a

higher vehicle weight.

1.5 Research Objectives

The work of Qu et al has been carried on at the University of Florida and parallels

the work in this thesis. The goals of this thesis were to establish a simple testing method

to determine residual strains, stresses, and coefficient of thermal expansion as a function

of temperature, and to measure these properties on the IM7/977-2 material system that

was used in the X-33 fuel tank development. In addition to this, various moduli were to

be experimentally determined and statistical data was to be supplied to the optimization

group for further probabilistic analysis and design.














CHAPTER 2
RESIDUAL STRESS CHARACTERIZATION TECHNIQUES

2.1 Introduction

Since the introduction of advanced structural composites the need to understand

their behavior has lead to extensive research in the area of residual stress determination.

Many existing methods have been devised to aid in residual stress characterization.

Some of these methods have been derived from existing tests used on other materials

while others have been developed from scratch. Each of these methods has it own

advantages and disadvantages. As a whole they can be useful in characterizing certain

aspects of residual stress and its effects. Most do have the same substantial flaw that will

be pointed out in this research. None, with the exception of the Cure Referencing

Method (CRM), developed at the University of Florida and used extensively in this work,

have the ability to account for the residual stress that results from the chemical shrinkage

involved in the curing of an ASC. This chapter will give a brief history of and detail the

existing traditional methods that are often used to characterize residual stress in laminated

composites.

Existing methods are usually grouped into two broad groups, those that are

destructive in nature and those that are non-destructive. Destructive testing usually

involves damaging or removing a section of material from a test specimen. The result is

a specimen that can no longer be used for most applications. Non-destructive tests are

generally preferred over destructive tests for obvious reasons. A specimen exposed to

non-destructive means of testing has not lost any usefulness as a structural component.









Tests can also be repeated on the same specimen to ensure accuracy. These methods also

often provide a means with which to test the effects of time on a component.

2.2 Destructive Testing Techniques

2.2.1 Hole-Drilling Method

One of the oldest and most accepted methods for the determination of residual

stresses is that of the hole-drilling technique. Mathar first proposed this technique as

early as the middle 1930s [33]. Since that time it has been employed to characterize

residual stresses present in metal structures and other isotropic materials. When a hole is

introduced into a stressed body the stresses are relaxed and become zero. This results in

a change in the surrounding strain field, which can be measured and correlated to the

relaxed stresses. This was the heart of Mathar's work.

The most commonly used application of the hole-drilling techniques involved

drilling a small blind hole near the application point of a strain gage rosette [35]. The

rosette can then measure the relaxed strain and the data can be used to calculate the

residual stress that was present before the creation of the hole. Since the middle to late

1960s efforts have been underway to extend the use of this technique from isotropic to

orthotropic materials [1,28,42,43]. The extension of this method into orthotropic

materials does not yield results as readily as in isotropic materials. The solution method

is numerically intense. Many assumptions must be made in order to simplify the

resulting solutions. The highly orthotropic nature of composites also adds further

difficulty in the measurements themselves. Obtaining measurement precision around the

hole is extremely difficult especially in the fiber direction, even for high precision

techniques such as moire interferometry [36].









2.2.2 Cutting Method

The cutting method is another technique that relies on the same principles as those

in the hole drilling method. Again, stress is relaxed by the removal of material. This

time a notch is removed from a specimen resulting in the creation of a free edge. Lee

devised a method where a moire interferometry grating was applied to the specimen to

record the resulting strain field [30,31]. The resulting strain field was calculated and

related to the residual stresses by the used of finite element analysis.

This method is not without its faults and has yet to be fully accepted in the

experimental community. Niu poses many questions based on the application of the

grating [37]. The grating was applied after an edge of the composite had been trimmed.

This resulted in some of the residual stresses being released prior to data recording and

the creation of a very complex strain field under the surface. The resulting measurements

that were made contained data that was based on a stress field different from the original

residual stress.

A second technique derived to employ the cutting method was proposed by

Sunderland et al. [46]. The successive grooving technique involved cutting a groove

through the thickness at successive depths. Strain gages were placed opposite the groove

and recorded the changing strain field. The residual stress was then calculated from the

strain by use of a numerical 2-D model for each layer.

2.2.3 Ply Sectioning Method

Yet another destructive technique devised to study residual stresses in laminated

composites has been proposed by Joh et al. [24]. This method made use of moire

interferometry. The deformation caused by sectioning and releasing a layer from the

constraints imposed by an adjacent layer was measured. The resulting release of stress









could be easily calculated from the deformation strains. The problems resulting from the

cutting method are resolved by obtaining a small strip specimen from the edge of the

larger specimen resulting in a plane stress state.

Another manner in which to use ply sectioning involves the nature of warping seen

in an unbalanced composite laminate. The outside layers of a laminate can be machined

away resulting in the unbalanced, warped structure [3,32]. The resulting warpage is

measured and can be used as an input in classical lamination theory to calculate the

corresponding residual stresses.

2.2.4 First Ply Failure

The first ply failure method makes use of the maximum stress criterion to calculate

the existing residual stresses [16,27]. Hahn and Kim recorded strains and loads while

loading a cross ply, [0/90], specimen to failure. Elastic assumptions were used to

calculate the load at which the first ply failed. This load was compared to the

corresponding strength of a unidirectional specimen and the difference was called the

residual stress. Kam also attempted to consider the visco-elastic effect on this

methodology [25].

2.3 Non-Destructive Techniques

2.3.1 Warpage

Warpage will result in unbalanced, asymmetric multidirectional laminates. The

destructive method of Joh et al. made use of this fact in its calculation of stress from

deformation strain resulting from sectioning. Asymmetric laminates can also be

intentionally manufactured so that the warpage can be measured. This method has been

an area of extensive investigation [8,10,17,22,23,26,27,32,51]. CLT provides the usual

means by which the stresses are calculated from the observed warpage.









The combination of warpage and CLT has also been used to devise a method to

measure and quantify polymer matrix cure shrinkage [8,9]. Daniel manufactured a

laminate directly onto an existing and previously cured identical laminate having the

same CTE and manufacturing procedure. Any resulting warpage was strictly the result of

the chemical shrinkage. The curvature was measured and stresses were again calculated

using CLT and the resulting stresses were directly related to chemical shrinkage.

2.3.2 Embedded Strain Gages

Embedded strain gages function by becoming an intimate part of the laminate being

measured. Any deformation of the laminate is directly seen and measured by the strain

gage. The most common strain gages used for this method are the electrical resistance

and fiber optic strain gage.

The first efforts to measure residual stress in this manner were made by Daniel

[5,6,7]. High temperature electrical resistance strain gages were cured directly in the

laminate by placing them between the lamina plies. The strain gages were then directly

exposed to any deformation of the laminate. A compensation method was derived to

account for the varying characteristics of the gage as it went through a given temperature

range. A similar method of compensation is used and described in this thesis. The main

drawback to the method described by Daniel was the introduction of a foreign body to the

laminate. This can drastically alter the material properties at the location as well as create

a void that will lead to incorrect measurements.

Fiber optics represents the second set of strain gages. This gage seems to be

gaining favor in the experimental community. Several recent studies have been

conducted with these gages [2,29,47,49]. In this technique a fiber optic is laid or

embedded in the composite as if it were one of the fibers. Again there is a concern with









the introduction of a foreign body into the heart of the laminate. If the optical fiber is not

on the same size scale as the material fiber then there is potential for significant errors

based on the change in the laminate material properties.

2.3.3 X-ray Diffraction

The use of X-ray diffraction in determining stresses was proposed by Predecki and

Barrett in 1979 [40]. Their work was extended to measure residual stresses by Fenn,

Jones, and Wells [11]. With this technique metal particles become part of the matrix

material and experience similar deformations. X-ray measurements are made on the

embedded metal particles and the resulting stresses are correlated to the residual stresses

in the composite. As with the embedding of strain gages, the introduction of metal

particles could result in a change in the laminate properties.

2.3.4 Electrical Resistance

Graphite and Carbon fibers are electrical conductors and therefore a laminate made

from them will have a measure of electrical conductivity and resistance. Researchers are

looking into a technique where the electrical resistance of a laminate can be used to

measure the development of stresses [38,50]. Wang and Chung have observed the

change in resistance across the thickness of a laminate and have set out to correlate the

measured resistance with stresses present in the composite. Park, Lee, and Lee set out to

achieve cure monitoring and strain-stress sensing based on a similar technique.

2.3.5 Cure Referencing Method

The Cure Referencing Method is a technique developed recently by Niu, Ifiu, and

Kilday to record process induced strains[20,21,37]. CRM uses the full-field laser based

optical method of moire interferometry to document strains on the surface of laminates

that initiate during the high temperature curing process. The method involves replicating






16


a high frequency diffraction grating on the specimen while in the autoclave. This method

will be key to the research done in this thesis. A detailed description of the application of

CRM will follow.














CHAPTER 3
CURE REFERENCING METHOD

Several distinct experimental techniques were used in conjunction with one another

to develop residual strain, stress, and coefficient of thermal expansion values across a

wide temperature range. The following chapters will describe in detail the processes and

methods used in the research. These processes and experimental techniques include

material lay-up procedures, basic material property characterization, the Cure

Referencing Method (CRM), and basic strain gage techniques.

3.1 Selected Laminates

The first step in developing a better understanding for the issues at hand was to

manufacture a desired test specimen. Three different composite laminates have been

chosen for study. A composite laminate consists of two or more layers or plies of lamina

arranged in any stacking sequence. Three different stacking sequences of IM7/977-2

were selected for study based on their relative importance. These laminates were:

1. Unidirectional (UNI): [013]
2. Reusable Launch Vehicle (RLV): Quasi-isotropic lay-up [45/903/-45/03]s
3. Optimized Angle Ply (OAP): [(25)]3s

The unidirectional stacking sequence was selected for its ability to provide insight

into some of the material properties of the material system [19]. It will be seen later how

the strain values obtained from the unidirectional material are used directly in the

analysis to calculate residual strain and stress in the other laminates.

The RLV was, as its name implies, taken directly from the stacking sequence

selected by Lockheed Martin for the design of the X-33 LH2 fuel tanks. It was designed









as a somewhat quasi-isotropic laminate to allow for maximum strength in all directions.

Quasi-isotropic lay-ups are very common in structural applications of ASCs. However, it

was not designed to account for the loss in strength that is associated with the

development of thermal residual stresses.

The lay-up selected as the optimized angle ply was the result of work done in an

effort to account for thermal stresses while still maintaining a level of mechanical

strength. This effort resulted from a reliability optimization performed by Qu [41], of a

LH2 fuel tank constructed from an IM600/133 graphite/epoxy material system. The

stacking sequence for the tank was [+61/62]s. Optimal designs had ply angles 01 and 02

near 25, and nearly equal; the stacking sequence was thus simplified to [+6]s. Monte

Carlo simulation gave a probability of failure of approximately 1 in 15,000 for the [250]s

design. It should be noted that this optimized ply sequence was for a different, but

similar material system.

These laminates were chosen for the study of the residual strain, stress and CTE.

Different sequences were used in the basic material property determination and they will

be described briefly in a following chapter.

3.2 Cure Referencing Method

The CRM is a technique developed at the University of Florida by Dr. Peter Ifju

and Xiaokai Niu. It is a non-destructive novel testing method designed to accurately

determine process induced and residual strains that are associated with the

manufacturing, cure, and thermal loading of an ASC. This method is based on a full-field

measurement technique know as moire interferometry in which a diffraction grating is

placed on a test piece during the cure cycle. The following section aims to describe CRM

and the procedures behind its implementation.









3.3 Moire Interferometry

Moire Interferometry is a very precise and sensitive laser based optical technique

that allows the development of a contour map of the in-plane displacements [39]. This

technique provides a displacement sensitivity of 0.417 |tm per fringe order. Moire relies

on the interference properties of light that results in the development of dark and light

bands known as fringes. For this interference to be observed, two diffraction gratings

must be compared to one another. One diffraction grating must be replicated to the

desired part or specimen that is to be deformed while another grating is to remain

unaltered to serve as a reference. This unaltered grating may be one applied to an un-

deformed specimen or the master grating that is used to transfer the pattern to specimens

as described in the next section. The moire interferometer is tuned with the reference

grating. "Tuned" refers to aligning the interaction of the laser and reference grating in a

manner so that the displacement field is nullified, or no fringes are present. This is

necessary so that only the deformation of the specimen is recorded by the presence of

fringes. The deformation that develops in a specimen can be observed at anytime under

any loading condition as long as the reference grating is still intact. In this study the

deformation given by moire was only observed for a mechanically unloaded specimen at

room temperature. A four-beam moire interferometer, schematically represented in

Figure 3-1, was used to record the deformation data.

The interferometer used in this experiment was a 2400 line/mm four-beam system

seen in Figure 3-2. The interferometer consisted of a 10 milliwatt Helium Neon laser

with a 632 nm wavelength and a 108 mm diameter parabolic mirror. The field of view

provided was 38.1mm or 1.5 inches in diameter.











Composite Panel -
or Reference Grating


Angle a Determines
the Reference Frequency



Figure 3-1: Schematic of a four-beam moire interferometer


Figure 3-2: Image of moire interferometer used in testing

3.4 Diffraction Gratings

Application of a diffraction grating to a part during cure is the central idea and

procedure behind CRM. Before this grating was applied to a specimen it was carefully









produced in a several step method derived from the process developed by Niu [37]

described next.

The development of an appropriate diffraction grating to be attached to a specimen

followed these procedures and is seen schematically in Figure 3-3:

5. A grating was replicated at room temperature onto Astrositall glass (ULE) using
silicone rubber from an original photoresist master grating. The glass had
dimension of 3"x 4.5"x 0.5". The silicone rubber was a two-part mixture (GE RTV
615). The silicone mixture was centrifuged so that all air bubbles could be
removed form the mixture.

6. The silicone grating was then placed into a vacuum deposition chamber where a
very thin layer of aluminum was deposited onto its surface. This layer was
deposited so that another piece of Astrositall could be easily separated from its
surface in a later step.

7. The aluminum coated silicone grating was then used to replicate onto the
Astrositall autoclave tool. This grating was made of 3501-6 epoxy. Both pieces of
Astrositall were placed into a heating/vacuum chamber where they were heated to
177 C and allowed to soak for approximately one hour. A small piece of 3501-6
was pulled from the freezer and placed into a metal tray where it could be heated.
A vacuum was applied to the chamber after heating and liquefying the epoxy for 15
minutes so that any trapped gas bubbles were removed from the epoxy. All three
components, the two pieces of Astrositall and the heated epoxy, were removed
from the oven. Rapidly, the liquid epoxy was poured into a small pool on the clean
Astrositall and then the aluminized silicone grating was placed squarely on top.
The two pieces of Astrositall were returned to the oven and a small weight was
placed on top so that the grating would be very thin. They were cured at 177 C for
10 hours.

8. The autoclave tool was then separated and prepared for use. The autoclave tool
was separated from the silicone tool by use of the instrument seen in Figure 3-4.
After separating, the autoclave tool was sent into the vacuum deposition chamber,
where two layers of aluminum were cast onto the grating. Each layer of aluminum
was separated by a thin layer of dilute Kodak Photo Flo solution that was applied
between the first and second deposition of aluminum. The orientation of the
grating lines on the tool was determined by the use of an alignment device similar
to one described by Post et. Al [39]. A straight line was marked on the tool so that
the fibers of the composite specimens were easily aligned with the direction of the
grating.

9. A thin layer of 3501-6 was then cast onto the autoclave tool directly on top of the
grating surface. A Teflon coated tool, seen in 3.5, was then used to apply the thin
film of epoxy. As in a previous step some of the components, the autoclave tool









and Teflon coated device, were heated for one hour at 177 C. At several intervals
during this hour the Teflon device was removed and the Teflon was tightened to
ensure a uniform film of epoxy. After the initial hour, a small piece of 3501-6 was
then placed into the metal tray where it was heated for 15 minutes and then
vacuumed to remove air bubbles. The components were removed and a thin pool
of epoxy was poured onto the grating surface. The Teflon device was lowered into
the pool at an angle to push out any possible entrapped air bubbles. The epoxy was
cured at 177 C for a period of 10 hours. After cure the autoclave tool was
removed from the oven and the Teflon device was disassembled leaving only the
sheet of Teflon on the tool. This sheet of Teflon was removed from the surface of
the tool by pulling back at a sharp angle.

After completion of the diffraction grating preparation, the grating was transferred

to the specimen by the following method.

3.5 Curing Procedure

As with most advanced composite materials, the laminates in this study were cured

in an autoclave (Figure 3-6). The autoclave is a pressurized oven that has vacuum line

connections inside the pressurized chamber. After selection of an appropriate laminate

the following steps were implemented [18]. It should be noted that the following

procedure was specific to the implementation of CRM. Other laminates were produced

using the same basic procedure and cure cycle with the only difference being that they

are cured without the addition of the ultra-low expansion Astrositall glass and grating

application.











STEP 1
Replicate Silicone
Grating on
Astrositall





STEP 2
Aluminum
Deposition





STEP3
Re plicate Epoxy
Grating onto
Autoclave Tool


Silicone grating


Astrositall





Aluminie d
silicone grating
tool



Aluminize d
silicone grating
tool


Autoclave
tool


STEP 4
Deposit Two
Layers of
Aluminum onto
Autoclave Tool


p


Aluminized
autoclave tool


STEP
Cast Thin Film of
Epoxyonto
Autoclave Tool


Teflon
casting
device


Aluminized
autoclave tool


Figure 3-3: Replication schematic



































Figure 3-4: Tool separation device


Figure 3-5: Teflon coated epoxy film casting device









10. Lay-up: Individual sheets were cut to size from a roll of prepreg. Those sheets
were then arranged and "stacked" in the proper sequence paying heavy attention to
maintaining the correct angles.

11. Vacuum bagging: Figure 3-7 shows a schematic of the vacuum bag assembly. The
ULE tool with the grating side up was placed at the base of the assembly. The
surface was then covered with a non-porous release film. A hole of approximately
1.5" diameter was cut into the release film. The hole allows for the grating to be
exposed to the composite surface during the cure cycle. The thickness of the film
was important. It was critical that it was as minimal as possible so that it did not
indent the surface of the composite during the cure process as the resin began to
polymerize. The prepreg laminate was then placed on top of the release film,
centered over the ULE tool. The non-porous release film placed between the tool
and the laminate prevented the epoxy from curing to the tool. A layer of porous
release film was placed on the upper surface of the laminate, which provided two
functions. It allowed for excess resin to be drawn off the laminate and into the
bleeder cloth, as well as facilitated the separation from the bleeder cloth. The
bleeder cloth was a porous material that absorbs excess resin drawn out of the
laminate via vacuum. Absorbing this excess resin allowed for an appropriate fiber
volume fraction to be achieved. The breather cloth, which was the same material as
the bleeder cloth, was used to distribute the vacuum throughout the entire bag.
Another layer of non-porous release film was placed between the bleeder and the
breather cloths so that excess resin was not transferred into the breather material.
This could have resulted in resin being drawn into the vacuum line or clogging the
breather cloth which then would have prevented an even vacuum distribution on
the laminate during cure. Finally the ULE and laminate assembly were placed into
a vacuum bag. The bag was sealed with a sealant tape and placed in the autoclave
for curing.

12. Cure preparation: The vacuum bag was placed inside the oven and connected to
the vacuum line. Vacuum was applied and a leak check was performed. It was
imperative that the vacuum integrity was maintained throughout the necessary
portions of the curing process.

13. Cure: Execute the cure cycle appropriate for the given material. The cure cycle for
the IM7/977-2 can be seen in 3-8.





































Figure 3-6: Autoclave oven


Porous release film Breather ply
Bleeder ply I Vacuum


I Non-porous release film
Pre-preg


- Pre-preg

. 3501-6 Epoxy layer
- Evaporated aluminum
-3501-6 Epoxy layer

-Astrosital autoclave tool


Figure 3-7: Vacuum bag assembly schematic









3.6 Moire Measurement Technique

After the specimens have gone through the cure cycle the reflective grating was

present on the surface (Figure 3-9). The specimens were then returned to the moire

interferometer where the deformation measurements were taken. There are many sources

of potential error in this technique if the proper steps were not taken to insure accurate

measurements.

The first step that must be taken was to properly tune the moire interferometer.

The virtual grating must be tuned accurately to 2400 lines/mm in both the horizontal and

vertical field, U and V respectively. This was done by directing the second order

diffraction of the reference grating back to the fiber optic tip for all adjustable mirrors.

Then fine-tuning was done for each field by adjusting the U and V field mirrors until both

displayed null fields. The exact position of the master grating was noted so that the

specimen was also placed at the exact same position. Failing to do so would have

introduced errors due to the imperfection in the collimated laser beam.

On each specimen a circle of know radius was scribed into the reflective grating.

This established the gage length over which the fringe numbers would be counted. Each

specimen was then placed into the tuned interferometer with a rotation at 900 from the

reference grating. This rotation was needed because the grating surfaces of the

specimens were the mirror image of the reference grating. The resulting fringes were

photographed using Polaroid ISO 400/270 black and white instant sheet film. Images

were taken and recorded from all of the different laminate specimens at room

temperature.











Cure Cycle


- Vac
-Temperature (F)
-Pressure (psi)


Minutes


Figure 3-8: IM7/977-2 cure cycle


Figure 3-9: Moire grating on specimen









3.7 Data Analysis

The experimental technique and methods described previously resulted in the

acquisition of the very characteristic pattern of light and dark fringes of moire

interferometry. These fringe patterns were used to directly determine the in-plane

displacements and strains. The relationship between fringe order N and displacement

was as follows:

U =-N (3.1)
f

N
V =-- (3.2)
f

wherefwas the frequency of the virtual grating; f-2400 lines/mm.

The absolute surface strain was also determined from the fringe pattern analysis by

the following relations:


S U= =- =- -i[AN] (3.3)
-x f L x f Ax

V 1[O ] 1 [AN] (3.4)
O 8x f Iy F f Ay I

These equations were used to determine the strain on the surface of both the

unidirectional and multidirectional laminates. These strains were a result of the thermal

contraction, which was a function of the temperature difference between the cure

temperature and operating temperature; chemical shrinkage of the matrix, which was not

temperature dependent, occurring only during cure; and the thermal expansion of the

autoclave tool, which was neglected in the analysis [19].

The resulting characteristic images of each laminate stacking sequence are

displayed in Appendix A and the corresponding values of the surface strain given by









CRM can be seen in Tables 3-1 through 3-3. The values seen in the table represent the

values of surface strain that were present at room temperature. These measurements were

extended throughout the range from cure to LN2 temperatures by a method to be

discussed next.

3.8 Error Sources

Several sources of error were potentially present during the implementation of

CRM. These sources included the improper implementation of moire interferometry,

manufacturing defects in the laminate specimens, and misalignment of CRM grating

surface. Without taking precautions to avoid possible error sources, the indicated strains

of the CRM method could be significantly misleading.

Table 3-1: Total surface strain on unidirectional laminates at 24C
Specimen x-direction y-direction
UNI-13-06-04 0 -6890
UNI-13-02-03 0 -6500
UNI-13-03-02 0 -7220
UNI-13-B2-02 0 -6955
UNI-13-02-05 0 -6824
Average 0 -6877.8
COV ---- 3.77%

Table 3-2: Total surface strain on optimized angle ply at 24C
Specimen x-direction y-direction
OAP-12-B3-01 787 -4920
OAP-12-03-02 820 -5250
OAP-16-04-01 919 -4950
OAP-16-02-02 1030 -5510
OAP-20-07-05 1033 -5250
Average 917.8 -5176
COV 12.5% 4.72%









Table 3-3: Total surface strain on RLV laminate at 24C
Specimen x-direction y-direction
RLV-A1-01 -610 -190
RLV-02-02 -525 -197
RLV-03-03 -525 -220
RLV-05-04 -590 -230
Average -562.5 -209.25
COV 6.78% 7.80%

The implementation of moire interferometry has many possible sources of error

itself. Improper tuning of the interferometer, misalignment of specimens, improper

specimen orientation and placement, and vibration can all lead to potential problems.

Fortunately, these issues are somewhat easily accounted for and all the proper steps were

taken to avoid improper strain reading as a result of these problems.

Other sources of errors were not as easy to account for. Manufacturing defects are

a common event in laminate production. Inclusions of foreign matter can often lead to

incorrect results. The existence of these inclusions on a large scale is easy to detect while

testing due to the large deviations they can cause from expected or previous results.

Manufacturing the laminates in a clean working area and examining each layer prior to

joining eliminated this problem. One source of error that was not ruled out was

misalignment of the layers themselves during manufacturing. All laminates were cut and

aligned as carefully as possible by hand but the element of human error still existed. It

was possible that misalignments of + 2.50 through the thickness developed. This

misalignment can result in a change in the characteristics of the laminate.

Similar to the alignment of the fibers during manufacturing, the autoclave tool and

the grating on its surface was aligned by hand onto the specimen. The greatest care was

taken to apply the grating as accurately as possible, but a possible slight offset from the

desired axis of measurement may have resulted. This offset would have lead to the









measurement of strain in a direction different than what was desired. For example, if the

strain in the fiber direction after cure is desired and the grating was not aligned perfectly

in the x, or fiber direction, as in the case of the unidirectional laminate, the measured

displacement would be higher than the true displacement.

A very important aspect of accuracy was the ability to read the fringe order

information. From equations (3.3) and (3.4) it is seen that any inaccuracy in the reading

of the correct fringe order would have lead directly to measurement errors. Different

fringe pattern densities often required adjustments on gage factor as the images must be

magnified to see the fringes. Different magnifications assured that all patterns could be

read to less than one fringe order. In the case where the gage length was 0.5", a misread

of one fringe would have resulted in an error of approximately 33 pa. In the case of a

0.25" gage length a misread of one fringe would have led to double the previous, up to 66

stg.

3.9 Conclusions

The Cure Referencing Method has been applied to find the total surface

displacement on various laminate stacking sequences. The calculated surface strains

represent the sum of the contraction from thermal effects as well as chemical shrinkage

that resulted from polymerization. The strains have been averaged and the coefficient of

variation has been calculated for each stacking sequences. Any COV measurements will

be provided to the optimization group for future reliability analysis.

The experimental results were relatively repeatable. All of the COVs were quite

low with the exception of the x-direction OAP values. This indicated quite consistent

and reliable results.









The unidirectional laminates showed the highest value of surface strain, followed

by the OAP, and finally the RLV laminates. The high strain level indicated that the

unidirectional and OAP laminates were relatively free to contract. Conversely, the RLV

laminate showed a very low strain value which indicated a high level of constraint. The

data from these tests are revisited in the combination of CRM and strain gage results in

Chapter 4.














CHAPTER 4
STRAIN GAGE METHODS

The second technique of experimentation that was used in this study was that of

basic thermal expansion testing methods conducted with modem strain gage technology.

This method was used to gather real time surface strain data as the temperature was

varied over the test range. Later it will be shown how information gathered with this

testing procedure was used in conjunction with CRM and material property testing to

develop residual strains, stresses, and CTE as a function of temperature.

4.1 Acquisition Theory

When strain gages are used to acquire strain as a result of temperature change the

output cannot be treated as the true strain on the surface. The output of the gage is

actually termed the thermal output. It is caused by two factors, the change in the

resistivity of the grid alloy with change in temperature and the difference in thermal

expansion coefficients between the gage and the test material [34]. The resulting

resistance change is the sum of resistivity and differential expansion effects are


R

where AR/R is unit resistance change, PG is the thermal coefficient of resistance of the

grid material, as UoG is the difference in specimen and grid thermal expansions, FG is the

gage factor of the strain gage, and AT is an arbitrary temperature change from a reference

temperature. The strain that is indicated from the change in gage resistance was then









AR/
= R (4.2)


where FI is the gage factor setting of the instrument. The gage factor of the instrument

was set to that of the gage so that the resulting thermal output, or strain, output by a gage

on a specimen was


T(GS) AT+ ( ) (4.3)
FG

From this output it was necessary to remove the gage contribution so that only the

surface expansion of the specimen was recorded. This was accomplished by placing the

same gage type from the same lot number onto a reference material of a known

expansion coefficient and expansion curve as a function of temperature. When the

reference material was exposed to identical conditions as the specimen the reference

material gives a thermal output of


T(G/R) = AT + (R G) (4.4)
FG

The gage output was measured on the reference material then the theoretical strain

at a given temperature was removed from the total output yielding the gage contribution

to the thermal output [18]. This output was then subtracted from the thermal output of

the specimen and gage reading, yielding only the thermal induced strain on the surface as

ES = 8T(G/S) T(G/R) (4.5)

The acquired strain readings were later combined with CRM data.

4.2 Specimen Preparation

To ensure the continuity of the measurements, the same specimens used for the

CRM procedure were used for the strain gage experiments. The CRM specimens were









sectioned so that the gratings were not damaged, yet a large enough area for strain gage

application was obtained. This yielded specimens of approximately 1" x 4".

The specimens were conditioned and sanded as recommended by Vishay

Measurements Group for strain gage application. The test covered a wide range of

temperatures that demanded special attention to gage and application techniques. For this

reason M-Bond 610 was selected as the adhesive system and two WK-13-250BG-350

gages were applied to each side, one directed in the x-direction and the other in the

y-direction. These gages were selected for their performance characteristics over the

proposed test temperature range. Each gage was aligned in the proper orientation by use

of mylar tape across the terminals. The gage backing and specimen surface were coated

with M-Bond 610 and were exposed to the air for approximately 15 minutes. After

drying, the gages were pressed down onto the specimen and covered with a thin Teflon

film. Pressure pads were placed over the gages and then they were clamped at 40 psi

with 2" spring clamps. The assembly was then placed in a cool oven and the temperature

was increased to 121 C at a rate of 5 C/min. and then was allowed to cure for 3 12 hours.

The tape was peeled off of the specimens and they were returned to the oven to be post

cured for two hours at 1350 C. A schematic and example of the resulting specimens are

seen in Figures 4-1 and 4-2.

Decoupling the strain gage expansion effects accurately was crucial to maintain the

validity of testing results. The reference materials were prepared following the exact

same procedures, making use of the same gages and bonding techniques, as those used

for the preparation of the composite specimens (Figure 4-3).















y,v

X,U

Figure 4-1: Schematic of strain gage alignment


Figure 4-2: Gauging of sectioned specimen






























Figure 4-3: Gauging of reference materials

4.3 Testing Procedure

The failure of the X-33 provided motivation for this work as well as the

environmental conditions under which the residual strains, stresses, and CTE were

studied. The composite specimens were observed and investigated over a wide range of

temperatures, from the harsh conditions of the cryogenic realm up to the temperatures at

which they were cured. The following section describes the methods that were used to

test and acquire data in these conditions.

After all preparations were made to the specimens and reference material they were

placed in a Sun Systems Model EClx environmental chamber. Each test consisted of

three specimens and the reference material being exposed to the desired conditions as

controlled by the environmental chamber. Due to the extensive amount of time that was

necessary to run and collect data, each test was divided into separate cryogenic and

elevated temperature runs.









4.3.1 Cryogenic Tests

A LN2 tank was attached to the environmental chamber as seen in Figure 4-4. The

chamber was then able to regulate the temperature by controlling the release of LN2. The

specimens and reference material were then supported on a rack covered with Teflon

(Figure 4-5). The Teflon ensured that the specimens were free to expand and contract

without being exposed to friction that might inhibit this deformation. Lead wires were

passed through a port in the chamber to the data acquisition set up. The chamber door

was returned and sealed while the proper connections were made to the acquisition

hardware.

Each gage was read in a quarter-bridge configuration (Figure 4-6). The lead wires

were passed directly to a STB AD808FB Optim Electronics quarter-bridge completion

module. The completed bridges were directed into a National Instruments SCXI-1322

breakout board attached to a NI SCXI-1122 sixteen-channel Isolated Transducer

Multiplexer Module for Signal Conditioning. The signal conditioning board was part of a

SCXI-1000 four slot chassis that directed the data into a Dell Dimension 8100 through a

NI PCI 6035E data acquisition card. The data and hardware was manipulated by a NI

LabVIEW program.

The chamber temperature was set for 200 C and the specimens were allowed to

soak there for 15 minutes. This temperature was selected as the starting point due to the

material property values of the reference material. The National Institute for Standards































Figure 4-4: Environmental chamber and LN2


Figure 4-5: Laminate coupons wired and placed inside environmental chamber









RL

---| R2
RE RL E

RL R


Figure 4-6: Quarter-bridge schematic

and Testing has established the thermal expansion data for the reference material, Ti6-

4A1-4V, as a deviation from this temperature. The titanium reference expands as:

(-1.711E2 + -2.140E x Tk +4.807E3 xT,2 -7.111E6 xT)x (4.6)

This equation is accurate to within 1.5% from temperatures of 240K to 3000K. A reading

was taken at this point and the corresponding voltages were stored and used as tare

voltages that were removed from all subsequent measurements.

After acquiring the tare voltages at 200 C the temperature in the environmental

chamber was reduced to 5 C. The specimens were again allowed to remain at that

temperature for 15 minutes so that temperature was constant through the thickness.

Another set of readings was taken at this temperature. The LabVIEW program that was

written to manipulate the incoming data took the raw voltages and proceeded to

compensate for several factors. The initial voltages read in the tare measurement were

removed from the new voltage readings. The new gage factor for the corresponding

temperature was calculated and then the raw voltage was converted into strain and

adjusted for transverse sensitivity errors. This was done for all gages inside the chamber.

The converted strain readings represented the total thermal output of the gages and the

gage contribution was decoupled as follows. LabVIEW calculated the actual strain

experienced by the reference material according to equation 4.6 and compared it to what









was read. The difference in the readings corresponded to the gage contribution to the

thermal output. This difference was calculated then subtracted from all other strain

readings resulting in the true surface strain on the specimens. This was recorded and

saved versus the temperature.

The same procedure was followed throughout the experiment. Dwell times were

increased as the temperature was lowered since it took more time for the specimens and

reference material to reach thermal equilibrium. Thermal equilibrium was verified by

taking several strain readings within a few minutes of each other. If the drift in the

calculated strain was observed to be very small, on the order of 2 to 5 microstrain per 3

minutes then the final reading was recorded and the experiment proceeded to the next

temperature.

The experiment terminated with the final reading at -190 C. This was the lowest

temperature that was obtainable using this environmental chamber and LN2. The

chamber temperature was returned to room temperature in several steps so as not to

thermally shock the specimens. The specimens and reference material were removed

from the chamber and it was heated to remove all moisture. This procedure was followed

for all specimens.

4.3.2 Elevated Temperature

Following the cryogenic testing the specimens were exposed to elevated

temperatures from room temperature to the upper limit of the composite cure

temperature. The same specimens were used for the elevated temperature tests so that a

curve across the entire test range could be created for each individual specimen. The

only difference was in the solder that was used to connect the lead wires, as the solder

appropriate for exposure to cryogenic temperature is not applicable at elevated









temperatures. The conductivity of both solders was very similar, thus the effect of this

change was negligible. The same hardware and software programs were used to collect

the data from elevated testing.

The chamber with specimens and reference material inside was again set at 200 C

where after a dwell time tare readings were taken. This initial point was selected in order

to obtain a continuous graph of the surface strains. It was not set at 200 C to account for

the reference material, as a small piece of Astrositall was gauged and used as a reference

instead of the titanium. The data available for the titanium reference was not valid into

the upper limits of the high temperature test. The Astrositall has an extremely low

expansion coefficient so that in the test range its actual surface strain was treated as zero.

The chamber was heated to 24 C and allowed to dwell there. This temperature

location was necessary since it corresponds to the room temperature at which the CRM

measurements were made. The need for the correspondence in temperature locations will

be discussed later. The dwell time needed for the elevated temperature test was much

longer than that of the cryogenic test due to the slower heat absorption of the Astrositall

reference material.

The LabVIEW program followed the same methodology for the calculation of the

true surface stain. The only difference being that the reference material was different and

had a different expansion curve. The Astrositall was treated as non-expanding

throughout the entire test. Therefore, any indicated strain on the reference material was

immediately subtracted from the specimen readings as the gage thermal contribution.

The temperature intervals began as 150 C increments. After the temperature

reached 900 C the increments were increased to 200 C. The test concluded when the









chamber was raised to 182 C, which corresponds to the composite cure temperature. As

before, when conditions began to deviate much from room temperature the dwell times

were increased. Thermal equilibrium was verified with the same technique as previously

discussed.

4.4 Data Analysis and Results

The data obtained directly from the above methods did not directly provide the total

surface strain present the laminate as a function of temperature. What was observed

through the strain gage testing was the relative change in surface strain as the temperature

was varied from 200 C. The strain data has to be coupled with the CRM data in order to

encompass all surface strains.

CRM was able to capture the entire strain induced on the surface of the laminate as

it went through cure and was brought down to room temperature, which was 24 C.

These strain values for the various laminates were highlighted previously (Tables 3-1

through 3-3). The data acquired through the use of strain gages must be referenced to

those values. This was the reason for the strain measurement taken at 24 C by the strain

gages.

The total surface strain as a function of temperature was calculated and plotted in

Microsoft Excel. The total surface strain present at any given temperature was:

Elam = sT + EcpM E24 (4.7)

where er is the strain calculated by the strain gages at any temperature, CCRM is the strain

recorded by CRM at 24 C for a particular specimen, and C24 is the strain recorded by the

strain gages at 24 C. Figures 4-4 through 4-6 show the resulting total surface strains as a

function of temperature on the various specimens of each laminate sequence.










All of the curves demonstrate fairly smooth behavior. The total surface strain results for

different specimens of each stacking sequence are in good agreement with each other and

no large deviations are observed among different specimens of the same laminate and the

average COV across the entire temperature range is acceptable for each (Table 4-1). The

surface strains have been averaged for each laminate (Figure 4-7).



2000

0.. -C-UNI-E x
-H-UNI-E y
-2000A-- UNI-B x
cc UNI-B y
-4000
-W -UNI-M x
-4-UNI-M y
-6 -I-UNI-N x
-8000 UNI-N y
UNI-F x
-10000 UNI-F y

-12000
-200 -100 0 100 200
Temperature (C)


Figure 4-7: Unidirectional laminate surface strain

A crucial observation to be made is that the strains do not return to zero as the

temperature is returned to cure. The cure temperature is often referred to as a stress free

state, hence strain is also expected to be zero at this point. This is only true during the

curing of the laminate because during cure the epoxy undergoes polymerization, which is

a one-time event. The resulting strain from this event, referred to as chemical shrinkage

is always present, is often overlooked, and can contribute a large amount of strain to a

given laminate (Tables 4-2 through 4-4). The stress measurement techniques described

previously, and an analytical prediction tool discussed later, known as Classical







46


Lamination Theory (CLT), do not have the ability to account for this phenomenon. CRM

provides a tool that can accurately predict and calculate this effect.


2000

0

-2000

-4000

-6000

-8000

-10000

-12000
-200 -100 0

Temperature (C)


Figure 4-8: OAP laminate surface strain


500
300
100
-100
'I -300 -
S-500 -
.y -700
-900
-1100
-1300
-1500
-200 -100 0


-- OAP-A x
- OAP-A y
-- OAP-Ex
OAP-Ey
- OAP-Hx
OAP-H y
OAP-K x
OAP-K y
OAP-I x
OAP-I y


100 200


-*- RLV-C x
--- RLV-C y
-A- RLV-L x
RLV-L y
-W- RLV-G x
-0- RLV-G y
--- RLV-Q x
RLV-Q y


100


Temperature (C)


Figure 4-9: RLV laminate surface strain

The strain data was also transformed into a linear coefficient of thermal expansion


from cure to the testing temperature. The data was taken in 150C and 200C intervals

allowing the CTE of the laminates to be calculated as a function of temperature. The









change in strain over the temperature interval was divided by that interval from cure to

cryogenic. CTE was calculated for each direction in the various specimens as:


CTEX (r2x rix) (4.8)



CTEY = (4.9)


The CTE values as a function of temperature are displayed in Figure 4.11. Over

the test range the change in CTE appears to be quite linear and trend lines were created as

such. The linear fit of the CTE of the unidirectional laminate will later be used to refine

Classical Lamination Theory predictions.

4.5 Error Sources

Just as in the case of CRM any misalignment in the orientations during

manufacturing would have led to incorrect measured values. Also, since the strain gage

readings were combined with those of CRM, any problems that may have been

encountered during the alignment of the displacement gratings would have propagated

into the combined measurements. The most likely source for a large error would have

resulted from the misalignment of a strain gage on the specimen.

Strain gages are extremely sensitive to misalignment when placed on composite

specimens. In some cases a + 40 misalignment of a strain gage would have led to errors

as large as + 17% in an axial gage measurement, while a+ 65% error could exist in a

similar transverse measurement. However, in this case the sensitivity to misalignment

was not as dramatic. On the unidirectional laminates the gages were aligned along the 0

and 90 directions. It was assumed that enough effort was centered on strain gage

alignment that less than 20 offset was induced. This desired alignment dropped









possible misalignment errors approximately to less than 5% in the case of gages off axis

by 2 In the case of the OAP laminate a + 20 offset leads to an error of about 3%.

The somewhat quasi-isotropic nature of the RLV laminate again helped to minimize the

effects of gage misalignment keeping errors on the order of 5%.

4.6 Conclusions

A method of measuring relative surface strains as they develop as a function of

temperature has been outlined. Proper compensation methods to account for temperature

effects on measurement techniques have been discussed and automated in a LabVIEW

data acquisition program. Measurements of surface strain have been made across a large

temperature range and the CRM and gage values have been combined in a manner to

account for the total surface strain at any given temperature. The surface strains for all

specimens have been plotted as a function of temperature and the averages of the

different laminates have been compared to each other. The chemical shrinkage

component of the total surface strain has been decoupled. Finally, the average CTE of

the different laminates has been computed as a function of temperature.

The different runs of the various specimens were quite consistent. The surface

strain values were reproducible as a function of temperature from run to run. As in the

results from CRM in chapter 3, the unidirectional and OAP laminates continued to

display a higher strain level than the RLV laminate. The same conclusion is reached, that

the increasing offset of ply angles has lead to a lower strain which resulted from high

constraint imposed in the RLV laminate.

The decoupled chemical shrinkage terms were quite large for the transverse

direction of the unidirectional and OAP laminates. These large values indicate that any







49


work that neglects the chemical shrinkage components will neglect a significant strain

event. These values were not seen in the RLV due to the usual constraint mechanisms.

The CTE was seen to tend toward zero as the temperature was decreased. This

event was observed less in the RLV laminate since the CTE was already quite low, again

due to high constraint. The trend toward zero was expected as molecules will tend to

slow in all motion as the temperature is reduced. The CTE observed in the unidirectional

laminates was taken to be the CTE of the material system and is used in CLT calculations

in Chapter 7 as ai and a2.

Table 4-1: Average COV of surface strains across temperature range
Laminate COV x-direction COV y-direction
UNI -- 4.26%
OAP 13.55% 6.10%
RLV 9.59% 10.10%


2000

0

-2000 -*-UNI x
--E--UNI y
-4000 ---OAP x
0 OAP y
0 -6000 AP
!5E -N- RLV x
-8000 -- -*-RLV y

-10000

-12000
-200 -100 0 100 200
Temperature (C)

Figure 4-10: Average of laminate surface strains













35 -
30
25
20
15
10
5
0
-5
-10 -
-200


-150 -100 -50 0 50
Temperature (C)


UNIx
M UNI y
OAP x
OAP y
X RLV x
RLV y





100 150 200


Figure 4-11: Coefficient of thermal expansion


Table 4-2: Unidirectional chemical shrinkage induced strain
Laminate x-direction


y-direction


UNI B -
UNI E 48.5 -2978.5
UNI F 99.3 -2338.5
UNI M 155.9 -2831.1
UNI N 131.1 -2702.0
Average 108.6 -2712.5

Table 4-3: OAP chemical shrinkage induced strain
Laminate x-direction y-direction
OAP A -- -2164.4
OAP E
OAP H 298.5 -1359.5
OAP I 550.2 -2153.1
OAP K 582.1 -2430.7
Average 476.9 2026.9

Table 4-4: RLV chemical shrinkage induced strain
Laminate x-direction y-direction
RLV C
RLV G
RLVL -169.3 29.3
RLV Q -83.8 -47.8
Average -126.55 -9.25














CHAPTER 5
MATERIAL PROPERTY TESTING

Basic material tests include those that provide some of the most elementary yet

critical values that can be assigned to a material. The following experiments were

performed to obtain the Young's Moduli, El and E2, Poisson's ratio, V12, and the shear

modulus, G12. These properties are necessary for the prediction of the strains and stresses

provided by Classical Lamination Theory (CLT). They must also be used to transform

the experimentally measured strains into stresses.

The American Society for Testing and Materials (ASTM) book of standards was

used as a guideline for the testing procedures implemented. Standard D 3039 was used as

the model for the El and E2 tests, while standard D 5379 was used to determine the shear

modulus.

5.1 El Testing

ASTM standard D3039 is a test method designed to determine the in-plane tensile

properties of a polymer matrix composite with high modulus fiber reinforcement [44]. A

panel of unidirectional IM7/977-2 was manufactured as discussed in previous sections

without the previsions made for the implementation of CRM. The panel had dimensions

of 10" in the fiber direction by 6" transversely. The panel was seven layers thick to

approach the ASTM recommended thickness of 0.040".

After the panel was cured, test coupons were cut from it having nominal

dimensions of 10" x 1". Each coupon was then sanded and conditioned at its midpoint in

accordance with Vishay Measurements Group techniques for strain gage application.









One CEA-13-125UT-350 tee gage from Vishay was placed on each side of the coupon

with the standard M-bond 200 cyanoacrylate glue. One gage located on each side allows

for cancellation of bending effects. The branches of the tee were placed precisely in the

fiber and transverse directions allowing for the acquisition of information needed to

calculate both the El and v12 properties.

After the coupons were gauged they were then placed in a MTI Phoenix machine

where they could be loaded (Figure 5-1). Each specimen was loaded axially in the fiber

direction with a crosshead displacement rate of 0.01" per minute. Load versus strain data

was acquired nominally at 200-pound intervals with National Instruments data

acquisition hardware. The hardware consisted of an SCXI-1000 four slot chassis, SCXI-

1322 breakout board, and a SCXI-1122 sixteen-channel Isolated Transducer Multiplexer

Module for Signal Conditioning all read into a Dell Dimension 8100 through a NI PCI

6035E data acquisition card. NI LabVIEW software was used to operate and read the

acquisition hardware.

5.2 E2 Testing

ASTM standard 3039 was also used to establish the E2 testing procedures [44]. In

this case a panel of IM7/977-2 was manufactured with dimensions of 6" in the fiber

direction by 7" in the transverse direction. The panel was 14 layers thick to bring the

thickness nominally to 0.080".

Again the panel was cut into test coupons after the curing process with the test

coupons nominally of dimensions 7" x 1". Each specimen was sanded and conditioned

as before at its midpoint. One CEA-06-250UN-350 unidirectional gage was applied to









each side of the specimen. The gage was placed on the coupon precisely in the matrix

direction, transverse to the fibers.


Figure 5-1: El and E2 testing fixture









As before, after the specimens were prepared they were subjected to load in the

MTI Phoenix displacement controlled testing machine. The displacement rate was again

0.01" per minute with load versus strain data acquired at approximately 200-pound

intervals. The same NI data acquisition hardware was used for this procedure as in the El

testing.

5.3 G12 Testing

As with the previous material property tests, the ASTM book of standards was used

as the guideline for establishing test setup and procedures. ASTM D 5379 is designed to

determine the shear properties of high-modulus fiber reinforced composite materials [45].

A 6" x 6" 13 layer panel of unidirectional IM7/977-2 was manufactured and then cured in

the autoclave with the appropriate cure cycle.

After the panel is cured it was sectioned and machined into the proper shape. The

testing coupons were machined into the form of a rectangular flat strips with two v-

shaped notches centered along the length. A diagram of the appropriate geometry is seen

in Figure 5-2. The panel was sent to TMR Engineering for machining of the correct

geometry.

The v-notched specimens were then prepared for testing through the application of

Vishay micro-measurements N2A-06-C032A-500 shear gages. The specimens were

conditioned appropriately and then one gage was applied to each side precisely in the

area spanning the concentric notches.

The gauged coupons were then placed in the MTI machine with a special testing

fixture seen in Figure 5-3. The specimens were inserted in the fixture and then placed

onto the machine in a manner where the notches were along the line-of-action of the









testing machine. The two halves were then tensioned at a rate of 0.01" per minute

while the load-shear strain data was collected using the NI hardware.

5.4 Data Analysis and Results

Before each El, E2, and G12 test was run the dimensions of the all test coupons were

measured. The stress, c, in each coupon was calculated as a function of cross sectional

area, A, and the load applied by the MTI machine, P:


S= -. (5.1)
A

The stress was then plotted versus the strain in each coupon. The slopes of the resulting

graphs (Figures 5-4 and 5-5) were the Young's moduli of each coupon. The various

moduli were very similar in value and were averaged. The results of each test are seen in

Table 5.1. The El testing yielded an average value of 150.2 GPa for the modulus of

elasticity in the fiber direction, while the E2 testing yielded an average of 8.04 GPa for

the transverse or matrix modulus.

The shear specimen results are plotted in Figure 5.6. The shear modulus was

calculated in accordance with ASTM Standard D 5379 where a "shear chord" between

the values of 1000 and 3000 ,s was calculated as

Gchord A(5.2)
Ay

The resulting average value for the shear modulus was 4.77 GPa.

The Poisson's ratio was also calculated from the El test by use of the strain

readings from the transverse gages. The strains displayed by the longitudinal and

transverse gages were compared and Poisson's ratio was calculated as





56


V12 -_ 2

All calculated Poisson's ratios were very similar and average to 0.342.

900


- 1.50"
p,


-- 3.0 0"

Figure 5-2: Shear test specimen schematic


Figure 5-3: Shear Testing Fixture


(5.3)







t
.750"

I













250000





200000

A

A
150000
















01
10000




50000






0 2000 4000 6000 8000
Microstrain


Figure 5-4: El stress/strain curves



9000

8000 -

7000

6000 -

5000

4000 -
I *
3000 /-




1000 ----

,, -________


*Test 1
* Test 2
A Test 3
X Test 4


10000


0 1000 2000 3000 4000 5000 6000 7000 8000
Microstrain


Figure 5-5: E2 stress/strain curves


*Test 1
* Test 2
A Test3










9000
8000
7000 A "
6000
6000 test 1
2- 5000 m test 2

4000 A test 3
5 O 3000 x test
2000
1000 -
0
0 2000 4000 6000 8000
M icrostrain

Figure 5-6: G12 stress/strain curves

Table 5-1: Results of each tension test
Run E1 (GPa) E2 (GPa) U12 G12 (GPa)
1 151.5 7.67 0.345 4.05
2 151.6 8.40 0.343 5.10
3 147.1 8.05 0.343 4.93
4 150.5 -- 0.338 5.00
Ave 150.2 8.04 0.342 4.77
COV 1.40% 4.54% 0.87% 10.16%

5.5 Micro-mechanical Methods

A detailed micro-mechanical approach is also being used by others to develop

characteristics of the IM7/977-2 material system [4]. The micro-mechanics approach

uses a unit-cell model of the material system. The unit-cell is modeled and analyzed

using finite element analysis of solid elements. Periodic boundary conditions are applied

to the surface of each cell. This micro-mechanical approach has yielded the El, E2, v12,

and G12 properties as well. Table 5-2 compares the results of the micromechanical

approach with experimentally determined values.









Table 5-2: Comparison of experimental measurements and micro-mechanics values
Material Experimental General Rockwell Micro- Max Min
Property Average Dynamics mechanics Percent Percent
Difference Difference
El (GPa) 150.2 169.6 148.2 159 12.1 1.34
E2(GPa) 8.04 -- 9.31 9.43 15.9 14.6
v12 .342 -- -- .253 29.9 29.9
G12 4.77 4.80 5.17 4.34 9.45 0.63
(GPa)

5.6 Error Sources

Potential sources for error in the determination of material properties included

many of those listed in pervious chapters. These potential sources were manufacturing

defects, especially layer orientation mistakes, and gage misalignment. Again,

misalignment of the strain gages is the most likely to result in the development of

significant errors.

For the most part the potential errors were reduced to below 5% due to the

orientation of the gage with the fibers in the specific tests. The only case of potential

significant error was in the v12 measurement made by measuring the transverse strain on

the El specimen. Any transverse measurement was much more sensitive to

misalignment, however, again the desired angle helped to alleviate possible error.

Transverse measurements were expected to have less than 10% potential error.

5.7 Conclusions

Various material tests have been conducted using ASTM testing standards. The

material properties of Young' Modulus in both the fiber and transverse direction have

been calculated as well as the resulting Poison's ratio. Further testing was conducted to

evaluate the shear modulus of the material system.









The data was very consistent and possessed little scatter in most of the

measurements. The COV of the various material properties were very low with the

exception of the shear modulus at 10.16%.

Comparisons were made between the experimentally determined material

properties and those calculated using a micro-mechanics based simulation. This

comparison of material properties yielded widely varying results. The El results agreed

quite well and the E2 results were not extremely different. However, the results for v12

did not yield good agreement. There were several potential sources for the dramatic

differences. One significant source of difference could have been the assumptions made

in the micro-mechanic model. This is a FE based model that is very depended on

boundary conditions. Also this model was applied on a single unit cell assuming a square

packing arrangement. These may have been poor assumptions. Just as likely to yield

differing results were the potential error sources listed previously. All of the samples for

each test were sectioned from one plate specific to that test. If any errors had been made

in the lay-up of the plate then those would have been seen in all tested specimens yielding

a precise but not accurate result. More material testing should be conducted on

specimens from more plates to investigate these differences.














CHAPTER 6
PLY LEVEL RESIDUAL STRAINS AND STRESSES

The residual strains and stresses in a composite result from the mismatch in

contraction of the matrix and fibers. This mismatch in contraction is from the

combination of drastically different CTEs and the chemical shrinkage event that is

limited to the polymer matrix. The follow section will analyze the data acquired from the

previous experimental techniques and present the resulting macro-mechanical residual

strains and stresses.

6.1 Residual Strain Calculation

By determining the state that each ply would reach if the bonds between them were

liberated the stresses that held them together can be defined [18]. This liberated strain

will be referred to as the residual strain. Using the assumptions of laminate theory, where

plane sections remain plane, for symmetric lay-ups, the strain on the surface matches that

at any point in the interior, as long as the location is a few ply thicknesses away from the

free edge.

The surface strain in each laminate is first transformed into its fiber and matrix

components of each ply:

Sply 1
SEpy 2 = [T]k *am 1 nilm 2 2'an, 12 (6.1)
Yply 12 J

where F1am is the total surface strain and [T]k is the transformation matrix of each ply:









cos2 0 sin2 0 2cos0sinO
[T]= sin2 0 cos2o -2cos sinO (6.2)
cos0sinO cosOsinO cos2 O- sin2 0

The residual strain in each ply then becomes:

res 1 unm_ 1 1py_1
Lres 2 = u sn, 2 ply 2 (6.3)
7res_12 L umn 12 ply 12

where {Suni} is the free expansion surface strain measured from the unidirectional

specimens. This surface strain represents the strain that a ply would reach if not

constrained by adjacent layers. These calculations are made for each temperature in the

experimental test range. Matrix properties are significantly weaker than those of the fiber

and the first sign of damage is typically mirocracks in the matrix. For this reason, the

residual strains reported will be those in the transverse direction. The residual strains are

plotted as a function of temperature for each layer in the OAP and RLV laminates

(Figure. 6-1).

6.2 Residual Stress

After residual strain was calculated the stress-strain relations were used to define

the corresponding residual stresses:

res Q11 Q12 0 res
tres 12 Q Q122 0 (6.4)
ires1 k 0 0 66 k res _12 k

The resulting residual stresses in the ply of the OAP and RLV laminates are plotted

versus temperature (Figure 6.2).










1.10E-02


9.00E-03


7.00E-03


5.00E-03


3.00E-03


1.OOE-03


-1.OOE-03
-200 -150 -100 -50 0 50
Temperature (oC)


-*- OAP
--- RLV 45
S-- RLV 90
RLV 0







100 150 200


Figure 6-1: Residual strain on selected plies


9.00E+07 -
8.00E+07 -
7.00E+07
6.00E+07
5.00E+07 _
4.00E+07
3.00E+07 -
2.00E+07
1.00E+07
O.OOE+00
-200 -100 0 100


200


Temperature (oC)


Figure 6-2: Residual stress level on selected plies









6.3 Conclusions

A method to quantify residual strains and stresses based on surface strains has been

discussed. This method has been used to display the resulting residual stresses that were

present in the different laminates. As expected, the high level of constraint that has been

discussed throughout this thesis has lead to high level of residual stress. The OAP has

performed much better than the RLV lay-up in reference to the development of thermally

induced stresses. The often neglected chemical shrinkage has induced as much as 15.4

MPa of residual stress in the RLV laminate.

From these results it has been made apparent that the chemical shrinkage is not a

negligible term and must be determined to fully quantify the extent of residual stress

development. It has also been made quite apparent that there is a great need for the

development of optimization techniques to account for the interaction of both

mechanically and thermally induced stresses.














CHAPTER 7
CLASSICAL LAMINATION THEORY

Classical Lamination Theory (CLT) is a predictive tool that makes it possible to

analyze complex coupling effects that may occur in composites. CLT is able to predict

strains, displacements, and curvatures that develop in a laminate as it is mechanically

loaded as well as those resulting from hygrothermal stresses. It is the thermal

contribution to strain that will be investigated in the following section.

CLT is derived from many assumptions, the most basic of which are that individual

lamina are perfectly bonded together so that they behave as a "unitary, nonhomogeneous,

anisotropic plate" [13]. The most important assumption in CLT is that each ply is

assumed to be in a state of plane stress and that interlaminar stresses are neglected.

7.1 Theory Development

Using the coordinate system in Figure 7-1, the laminated plate geometry and ply

numbering scheme defined in Figure 7-2, along with several basic assumptions, a set of

equations are derived to predict deformations and stresses. The basic assumptions

needed in the derivation are:

14. The plate consists of orthotropic lamina bonded together, with principal material
axis directed in arbitrary directions with respect to the x and y axes.

15. Edge lengths of the plate must be much greater than the individual ply thicknesses.

16. The u,v,w displacements are small in comparison to the total plate thickness.

17. The in-plane strains Fx, Sy, and xy are much smaller than one.

18. Transverse shear strains yxz and 7yz are neglected.

19. The u and v displacements are linear functions of the z coordinate.









20. The transverse normal strain Fz is neglected.

21. Hooke's Law applies to each ply.

22. Total plate thickness is constant.

23. At the plate surface the transverse shear stresses Txz and Tyz are zero.

Assumptions 6 and 7 define the displacements as

u =u (x,y) + zF(x,y)
v = v0 (x,y) + zF2 (x,y) (7.1)
w = w(x, y) = w(x, y)

where uo and v are the displacements in the x and y directions at the middle of the plate.

The displacement w is the same at any thickness in the plate with the same x and y

coordinates due to assumption 7. Substituting equations 7.1 into the strain-displacement

equations for the transverse shear strains and imposing assumption 5 yields

Bu Ow Ow
XY =-+- =F,(x,y)+-= 0
8z 9x 8x
O a O (7.2)
Qv Ow Ow
Yyz=-+-=F2(x,y)+-= 0
az ay By

and


F, (x, y) =-
O (7.3)
F2(,y)-
Sly







67













y x







Figure 7-1: Laminate coordinate system






2 zo
/ Z1
3
Z2

Middle Surface z3
Zk-1 Zk

k ZN-1ZN


NZN


Figure 7-2: Layer schematic and nomenclature

Combining the strain-displacement relations for in-plane strains with Eqs. (7.1) and (7.3)

results in

ou o
gx E+-- ZKx


E=- E + zC (7.4)


ay y
2/XY= OY+ ac= 2XY+ ZI









where mid-laminate strains are

u0


E (7.5)
9y

yo =auo v


and mid-laminate curvatures are given by


K 2x

K =- (7.6)
S OY 2


8x8y
Cc -)C-Y-


These represent the curvatures associated with the bending at the middle of the laminate

in the xz and yz planes as well as the out of plane twisting of the middle surface of the

laminate. In our case, as with any symmetric laminate, these curvatures will be zero.

They are left in the derivation to give the solution for any arbitrary stacking sequence

where curvature may be present.

Equation (7.4) provides a solution for the strain at any given distance from mid-

laminate. If the various lamina are of known thickness then the distance of each from

mid-laminate is known and the stress-strain relations for the kth lamina can be expressed

as

x Q11 Q12 Q16 EX +zK
y 12 22 Q 26 -' + z, (7.7)
Q16 Q 26 Q 66 0 r + Z
xY k 16 26 6 r xY


where









[]= [T] [QI[T] (7.8)

and

Q11 El
V12V21

1 V12V21 (7.9)
22 E2
I V12v21
Q66 = G12

When determining strains and stresses in a laminated plate it is convenient to look

at the forces and moments per unit length as they act on a plate. The force per unit

length, Nx, is given by

Nx f 2 x dz (x )kdz (7.10)
2 k= k-1

and the moment per unit length, Mx, is given by

M : 2 a -d: k ()kzdz} (7.11)
t/2 k=l -k-1

where t is the laminate thickness, (O )k is the stress in the kth lamina, Zk-I is the distance

from mid-laminate to inner surface of the kth lamina, and Zk is the distance to the outer

surface of the same lamina. Combining (7.7) with both (7.10) and (7.11) yields the

following equations for the force and moment per unit length.

Nx 1k = 1 {(Il)k -+ )-Zx + 12 )k -E + )-Z--16 )k(Xy+ ZxytZ
k k-1 (7.12)
M k 11 )k ( Z x 12 )k ( ZK Q16 )k x Z +ZK y dz
k=after rearranging (7.12) the force and moment equations can be written as

after rearranging (7.12) the force and moment equations can be written as









N =A gEO+A2 0+A1670 ++B +B12K y+B16 /C
Mx = BEl +B 120 +B 160 +D lcx +D 12y +D 16/
x 11 X 12 (7 .613 )1 12 + 6

where the laminate extensional stiffness is given by


A, f (Q kdz ) (zk Zk-1 (7.14)
2 k-1 k

the laminate coupling stiffness is given by


B, = kzdz Z= ) (z (7.15)
2 2k=1 k

and the laminate bending stiffness is


D J = k zdz _-I ) Z) (7.16)
2 k-1 k

Finally the entire set of equations can be expressed as

Nx All A12 A16 B11 B12 B16 Ex
yN11 22 A26 B12 B22 B26 cy
Nxy A16 A26 A66 16 26 B66 x
X 11 12 16 11 12 16 (7.17)
My B B22 B26 D, D22 D26 K
Mxy B16 B26 B66 D16 D26 D66 xy

The above solution exists for the state in which laminates are not subjected to any

thermal or moisture effects. Together these effects are known as hygrothermal effects. It

is the thermal component of this stress that is of the greatest concern in this case and

therefore the stresses and strains in the laminate must be solved for this hygrothermal

effect.

When the moisture and thermal effects are incorporated the resulting stresses on

each lamina are








}k = [Q ({k {}k AT {}k c) (7.18)
and the total lamina strains equal

{tk= { }+ z{ic (7.19)
Equation (7.19) is substituted into Eqn. (7.18) giving

{t}k = Qt ({ 1+ z{ {}k AT {}k c) (7.20)
The force and moment per unit length are then solved in the same manner but this time
by integrating (7.20).

{N} = J}k dz= (E 0 + z {a} k AT }k iz
{NI= [A ]o}+ [B]K- N NM} (7.21)
S= J{ }kzdz = []k ({} +z{} {- k T }kc)zdz
j=[0}+ B]O + [D]{ T_ M

The contributions of the thermal and moisture components are

{NT} J[]Qk tak ATdz = (AT) [] a}k (Zk Zk-1
k=l k (7.22)
{NM = JIk V[ k }cz= cdz ck (zk -Zk-1
k=1 k
and

PT J f]lk Ak Tzdz = (AT) N }k2z 2
2 k 1 k
(7.23)
{P }= J[tl k CZdz z= I[] {(z kz-/)
2 kthe entire solution k
Finally, the entire solution can be written in matrix form as









X +X X11 A12 16 11 B12 16N
N + NT + NM A A22 A26 B12 B22 B26 B 0
Nxy AT M 16 26 66 16 26 66 (7.24)

Mx +M +M B B12 B16 D, D12 D16 /
My +M +MM B B22 B26 D D22 D26 y
MXy +M +M 16 26 66 D16 D26 D66 xy

This approach was used to predict the strains on the surface of the laminates and

compare them to those measured via experiment. A MatLAB code was authored,

primarily by Thomas Singer and William Schulz, to encompass all of the assumptions

and calculations involved in CLT (Appendix B). The code was executed at each

temperature in which experimental data was collected. CLT makes its predictions based

on the material properties of individual lamina. The material properties of El, E2, V12,

G12, and ai and a2 were obtained from previous experiments. The CTE values of ai and

a2 have been calculated as functions of temperature from previous analysis. The apparent

change in CTE is linear and therefore an average CTE across the test range was used as

the input for the CLT program.

7.2 Experimental vs. Analytical Results

CLT results were plotted as a function of temperature and compared to the

experimental results (Figures 7-3through 7-5). It can be seen in all cases that there is

substantial difference between the results of CLT and experimental data.

Several factors could lead to potential differences in the measured and predicted

values. The first potential source is in the assumptions that are made in CLT formulation.

The second and most crucial difference is that CLT lacks the ability to interpret the event

of chemical shrinkage and the effect it has on laminate strain. If the values at the cure

temperature are compared it is seen that CLT predicts no strain at this point, while the







73


combined experimental techniques show strains that are the result of the chemical

shrinkage. Since chemical shrinkage is a one-time event and the resulting strains are an

invariant part of the laminate strain they can be added to the results of CLT. Upon the

introduction of the chemical shrinkage term into the CLT results it was seen that the

graphs collapsed onto each other (Figures 7-6 through 7-8 ).


2000 1


0

-2000

-4000

-6000

-8000

-10000

-12000


Figure 7-3:


-6-x CLT
-e-y CILT
---- x
'-- y


-200 -100 0 10
Temperature (oC)

Unidirectional CLT and experimental results


2000

0

-2000-

-4000-

-6000-

-8000

-10000


--x CLT
-e-y CLT
.-- x
--U-- y


-200 -100 0
Temperature (OC)


Figure 7-4: OAP CLT and experimental results


--------------






WAA-MI,


''I - -







74



100

-100

-300
-6-x CLT
-500 -9e-y CLT

-700 --x

-900 -

-1100 -

-1300
-200 -100 0 100 200
Temperature (C)

Figure 7-5: RLV CLT and experimental results

The resulting agreement between the CLT and experimental values provides proof

of testing concept and a high degree of measurement accuracy.



UNI CLT & Measured Surface Strains


2000

0 - --- -- -

-2000 --- xCLT
C -.-y CLT
-4000 -----

S-6000 -
2- -x RawCLT
-8000 ---y RawCLT

-10000

-12000 ....ii
-200 -150 -100 -50 0 50 100 150 200
Temperature (C)


Figure 7-6: Unidirectional CLT shifted for chemical shrinkage










CAP CLT& Measured SLuface Strains


2C000 -
0-
-200 -
-4000-
-6000 -4
-8000-

-10000-
-12C00 -
-200


-100 0 100
Terrpeatre (C)


--x CLT
--y CLT

---y
--,-x RajvCLT
-y Rav CLT


200


Figure 7-7: OAP CLT shifted for chemical shrinkage


RLV- CLT & Measured Surface Strains


-100 0 100
Temperature (C)


-- x CLT
-yCLT
--- x
-*-y
xRawCLT
--- y RawCLT


200


Figure 7-8: RLV CLT shifted for chemical shrinkage


1000 -


1o-
500
0
-500 -G


-1500 -
-2000 -
-200









7.3 Conclusions

Classical Laminate theory has been examined and used to predict the surface strains

on the different laminates. This examination has taken place as an attempt to verify the

experimentally determined results. It is a well known fact that the traditional approach to

lamination theory does not have the means to account for chemical shrinkage.

When the two results were compared an interesting phenomenon was observed.

The measured and predicted curves appeared to follow nearly the same trends, separated

by an offset. When the curves were examined it was seen that this offset was almost

exactly equal to that of the chemical shrinkage. When the previously decoupled chemical

shrinkage term was added to the analytical results form CLT the offset in the curves

vanished and the two results were nearly identically. Several substantial conclusions

have been drawn from these results.

First, the fact that the resulting curves coincided with each other throughout the

testing range confirmed that the chemical shrinkage can be treated as a one time event,

completely independent of temperature effects. Secondly, the fact that the two curves

coincided very well provided proof of concept in the strain gage measurements and

temperature compensation technique. Last, and perhaps the most important conclusion

was that the CRM method did indeed provide a means by which the true total strain

present in a laminated composite could be measured. From this result, it is seen that once

the chemical shrinkage term is decoupled and defined, the computer based analytical tool

of CLT can be used to predict the development of strains anywhere in a given

temperature range.







77








APPENDIX A
CRM IMAGES OF SPECIMENS


iN


V, y


U, x


Figure A-l: Unidirectional Laminate Run 1







78
























V, y1




Figure A-2: Unidirectional Laminate Run 2


V



, x


v
V
V, y


U,x


Figure A-3: Unidirectional Laminate Run 3

































V, y


Ux


Figure A-4: Unidirectional Laminate Run 4


V, y


U,x


Figure A-5: Unidirectional Laminate Run 5







80






















U
V, y


U, x

Figure A-6: OAP Laminate Run 1


V, y


U,x


Figure A-7: OAP Laminate Run 2







81






















U
V, y


U, x

Figure A-8: OAP Laminate Run 3


V, y


U,x


Figure A-9: OAP Laminate Run 4







82






















U
V, y


U, x

Figure A-10: OAP Laminate Run 5


V


V, y


U, x


Figure A-11: RLV laminate Run 1







83
























U

V, y


U, x

Figure A-12: RLV Laminate Run 2


V, y


U,x


Figure A- 13: RLV Laminate Run 3


,!


































V,yU,


U,x


Figure A-14: RLV Laminate Run 4


















APPENDIX B
MATLAB CLT CODE

Main Routine: laminatestress

function
[strainO,curve,stressx,stressl]=laminatestress(plydir,thick,E1,E2,G12,n
ul2,NM,deltaT,c,alphal2,betal2,plots)
% laminatestress uses classical laminate theory to give midplane strain
and curvature, and
% stress (in both the coordinates of the laminate and of the ply) at
the top and bottom of each ply.
% All referenced equations are from Gibson, Principles of Composite
Material Mechanics, 1994
% strainO: midplane strain, in laminate coords. [e x;e y;g xy].
% curve: curvature. [K x;K y;K xy] .
% stress: stress in laminate coords. stress is 6 x k, where k is
the number of plys.
% The kth column gives stress in the x and y directions and shear
in the xy plane at
% the top and bottom of the kth ply;
[s x top;s y top;s xy top;s x bot;s y bot;s xy bot].
% stress: stress in ply coords. Same as stress, but the kth column
is in the kth ply
% coords, with the 1-direction longitudinal, and 2-direction
transverse.
% plydir: string of the form '[90/+-45/(-+30)3/90bar]s' or column
matrix of ply directions. Note
% that the variables make more sense with the z-axis down;
following this convention, the
% first ply is the most negative.
% plythick: column vector of ply thicknesses. If thick is a scalar,
the thickness is applied to
% all plys.
% El: column vector of Young's modulii in the fiber direction of each
ply. If El is a scalar,
% the modulus is applied to all plys.
% E2: same as El, in transverse directions.
% G12: same as El, but shear modulii.
% nul2: same as El, but Poisson's ratios.
% NM: column vector of mechanical loading conditions
[N x;N y,N xy;M x;M y;M xy] per unit
% length.
% deltaT: change in temperature from cure temp.
% c: moisture concentration.
% alphal2: matrix of coeffs of thermal expansion, in material coords.
alphal2
% is 2xn or 3xn, where n is the number of plies. If alphal2 is
2xl or








86


% 3x1, the CTEs are assumed to apply to all plies. Row 1 is CTE
in the
% fiber direction, row 2 is CTE in the transverse direction, and
row 3
% is 0.
% betal2: matrix of coeffs of hygroscopic expansion. Same format as
alphal2.
% plots: 1 plots stress and stress through thickness; 0 or blank
doesn't plot.
% Note: the last 5 inputs are optional. If only one is entered, it is
taken to be plots.
% If 2 are entered, they are taken as deltaT and alpha. If 3 are
entered, they are taken
% as deltaT, alphal2, and plots. If 4 are entered, they are taken as
deltaT, c, alphal2,
% and betal2.


% ----------Housekeeping----------
if exist('deltaT')==0
deltaT=0;
c=0;
alphal2=[0;0;0];
betal2=[0;0;0];
plots=0;
elseif exist('c')==0
plots=deltaT;
deltaT=0;
c=0;
alphal2=[0;0;0];
betal2=[0;0;0];
elseif exist('alphal2')==0
alphal2=c;
c=0;
betal2=[0;0;0];
plots=0;
elseif exist('betal2')==0
plots=alphal2;
alphal2=c;
c=0;
betal2=[0;0;0];
elseif exist('plots')==0
plots=0;
end
%Sets variables not assigned in input

[a,b]=size(NM);
for i=a:5
NM(i+l,1)=0;
end
%Fills in missing force and moment terms

if ischar(plydir)
plydir=directions(plydir);
end
%Converts shorthand ply orientation to matrix













% ----------Computation-------------
[m,n]=size(plydir);

[stiff,Q,S,Qbar,Sbar,T,z,zO]=ABD(plydir,thick,E1,E2,G12,nul2);
%Calculates ABD matrix

[Nh,Mh,alpha,beta]=hygrothermal(deltaT,c,alphal2,betal2,Qbar,T,zO,z);
NM=NM+[Nh;Mh];
%Adds hygrothermal forces and moments. Eqs 7.95, 7.96

invstiff=inv(stiff);
ek=invstiff*NM;
%Solves for midplane strain and curvature. Eq 7.46

strain0=ek(1:3);
curve=ek(4:6);
%Extracting midplane strain and curvature

strainxtop=strainO+z0*curve;
%Pesky 0-index problem...see below

for k=l:m
strainx(:,k)=strainO+z(k)*curve;
botstressx(:,k)=Qbar(:,:,k)*(strainx(:,k)-alpha(:,k)*deltaT-
beta(:,k)*c);
if k==l
topstressx(:,k)=Qbar(:,:,k)*(strainxtop-alpha(:,k)*deltaT-
beta(:,k)*c);
else
topstressx(:,k)=Qbar(:,:,k)*(strainx(:,k-1)-alpha(:,k)*deltaT
beta(:,k)*c);
end
%Stress in general (x-y) coords. Eq 7.85
botstressl(:,k)=T(:, :,k)*botstressx(:,k);
topstressl(:,k)=T(:, :,k)*topstressx(:,k);
%Stress in ply (1-2) coords. Eq 2.31
end

stressx=[topstressx;botstressx];
stressl=[topstressl;botstressl];


% ---------Plotting---------
if plots==l
for k=l:m
stressxx(2*k-1,1)=stressx(1,k);
stressxx(2*k,1)=stressx(4,k);
stressxy(2*k-1,1)=stressx(2,k);
stressxy(2*k,1)=stressx(5,k);
stressxxy(2*k-1,1)=stressx(3,k);
stressxxy(2*k,1)=stressx(6,k);
stressll(2*k-1,1)=stressl(l,k);
stressll(2*k,l)=stressl(4,k);
stressl2(2*k-1,1)=stressl(2,k);
stressl2(2*k,1)=stressl(5,k);