RESPONSE OF MARINE COMPOSITES SUBJECTED TO NEAR FIELD BLAST LOADING

Experimental studies were performed to understand t he explosive response of composite panels when exposed to near-field explosi ve loading in different environments. The panel construction under consider ation was an E-glass fiberreinforced composite laminate infused with vinyl es ter resin (Derakane 8084). The panel was layered bi-axially with plain-woven fiber orientations at 0o and 90o. Panel dimensions were approximately 203 mm x 203 mm x 1 m m (8 in x 8 in x 0.04 in). Experiments were carried out with the panel fully c amped in a holding fixture, which was in turn fastened inside a water tank. The fixtu re was fastened in such a way as to allow for explosive loading experiments in the foll owing environments: water submersion with water backing, water submersion wit h a r backing, and air immersion with air backing. Experiments were performed in roo m temperature conditions, and additional experiments in the submerged environment s were also performed at high and low water temperatures of 40 °C and 0 °C, respectively. A stereo Digital Image Correlation (DIC) system was employed to capture th e full-field dynamic behavior of the panel during the explosive event. Results indic ated that the immersion environment contributes significantly to the blast response of the material and to the specimens’ appreciable damage characteristics. The water submersion with air backing environment was found to encourage the grea test panel center point deflection and the most significant damage mechanisms around t he boundary. The air immersion with air backing environment was found to encourage less center point deflection and exhibited significant impact damage from the explos ive capsule. The water submersion with water backing environment encourage d th least panel deflection and minimal interlaminate damage around the panel bound ary and center. Water temperature was found to influence the panel center point deflection, but not damage mechanisms. Maximum positive center point deflectio ns associated with the high and room temperature water submersion with air backing e vironments, while statistically similar to each other, were found to be statistical ly different from those associated with the low temperature environment.


CHAPTER 1: Introduction
Composites have been employed in a variety of applications for many years in the marine, automotive, and other commercial industries. They are known and valued for qualities such as high strength-to-weight ratios, good corrosion resistance, resistance to water absorption, and reduced maintenance requirements. These characteristics have garnered them recent attention as effective materials in military applications.
Military structures are frequently exposed to extreme loads in the field or at sea, and these loads induce high strain rates in the materials that comprise those structures. The response of composite materials at high strain rates, and in unique environments, is not fully known as yet. This lack of understanding often prevents the designing of components that maximize their material advantages. It is thus advisable to conduct experimental research to fully understand how these materials respond to extreme loadings, so as to determine how to best take advantage of their qualities (LeBlanc, Gardner, & Shukla, 2013). The research described herein attempts to accomplish this objective.
It is of additional interest to naval and maritime engineering that the effects of particular extreme loadings be studied with special emphasis. Of singular importance are the consequences of explosive loadings. Among the varieties of explosions classified in the literature, the near-field explosion enjoys a great deal of relevance to naval design, planning, as well as tactics and maneuvers at sea. This is evidenced by the breadth of history concerning the use of torpedoes, undersea mines, and improvised explosive vessels as instruments to damage or sink ships. The near-field explosion also enjoins uniquely complex interactions on its target, and is thus a worthy candidate for study with a composite specimen.
This study examined the near-field explosive response of an E-glass/Vinyl Ester (EVE) marine composite panel specimen in different immersion environments and at different water temperatures. The panel specimen was a 2-ply plain weave laminate infused with Derekane 8084 vinyl ester resin. In each of the immersion environments, the specimen existed as a fully-clamped panel barrier between two media, either water or air. The face opposite the explosive side was referred to as the "backing" side.
Under this arrangement, the immersion environments were assigned as follows: (1) water submersion with air backing (hereafter referred to as "WA"), (2) water submersion with water backing (hereafter referred to as "WW"), and (3) air immersion with air backing (hereafter referred to as "AA"). Each environment was investigated in turn at room temperature. Following the room temperature studies, low temperature experiments were performed in the WA and WW environments for water temperatures of 0 °C, hereafter referred to as "WALT" and "WWLT," respectively. After these, high temperature experiments were performed in the WA and WW environments for water temperatures of 40 °C, hereafter referred to as "WAHT" and "WWHT," respectively. RP-503 detonator caps, with a TNT weight equivalent of 1 g, were the chosen explosives for this study. Free field pressure transducers were employed to record the spherical shockwave pressure histories underwater and in air, and high speed digital photography was employed to capture real-time, full-field deformation histories via the Digital Image Correlation (DIC) method. Panel displacements and damage mechanisms were noted and analyzed to discover the effects of immersion environment and water temperature on the blast response of the EVE composite specimens.
It is humbly submitted by the author that the results contained herein will contribute to the understanding of the dynamic thermo-mechanical properties of EVE composites, inspire similar research, and exist as a deposit of relevant information for that research. It is hoped that these results will assist the engineering and design communities in developing stronger and more efficient naval structures.
CHAPTER 2: Review of Literature

Review and General Comments
This study endeavored to examine the near-field blast response of a common marine composite, constructed from materials available on the commercial market,  (Haile, Yin, & Ifju, 2009). Explosive theory is also well-established for both underwater and in-air events and has received much attention over the last century (Cole, 1948) (Brinkley & Kirkwood, 1945) (Hartmann & Laboratory, 1976) (Smith & Hetherington, 1994). The uniqueness of the study under consideration is that it incorporated the effects of (1) a near-field explosion, (2) loading environments characterized by water submersion with water backing, water submersion with air backing, and full air exposure, (3) water temperature variation, and (4) a fully-clamped boundary condition around a flat panel.
Available literature does not address this engineering problem exactly.
Torabizadeh examined the mechanical behavior of unidirectional glass/epoxy composites, for temperatures ranging from -60 °C to 25 °C, when subjected to static loading (Torabizadeh, 2013). Wang, Zhou, and Mallick demonstrated the temperature and strain rate-dependence of Polyamide-6, an E-glass-reinforced composite, over temperatures ranging from 21 °C to 100 °C and slow strain rates of 0.05/min, 0.5/min, and 5/min (Wang, Zhou, & Mallick, 2002). Van Lear studied the effects of fluid structure interaction in an underwater explosive (UNDEX) event using numerical methods (Van Lear, 2008), while Espinosa et al examined UNDEX fluid structure interaction experimentally using a scaled water piston apparatus instead of an explosive charge (Espinosa, Lee, & Moldovan, 2006). Shin simulated the effects of a far-field explosion on the hull of a warship using the LS-DYNA finite element software (Shin, 2004 Rajendran and Narasimhan studied the effects of contact UNDEX loading on curved steel plates in an experimental environment characterized by water submersion with air backing (Rajendran & Narasimhan, 2001). Cichocki employed numerical software to model the loading effects of a 40 kg spherical charge on a closely-nearby protective structure around a pipe (Cichocki, 1999  Hassan, in separate studies, examined the blast resistance of fiber-reinforced composites in air and water environments using specialized numerical software (Batra & Hassan, 2008) (Batra & Hassan, 2007).
Much of the foregoing literature touched on aspects similar to those found in this proposed study, but none addressed each integral part at once, and none examined the effects of temperature variation on the specimen's dynamic response.

Literature Review of Basic Explosion Theory
An explosion is defined in broad terms as a "large scale, rapid, and sudden release of energy" (Ngo et al., 2007). Ngo et al. list several possible mechanisms for such discharge, namely those associated with chemical, nuclear, or physical events. This present study considered only explosions of chemical origin-that is to say, explosions produced as a result of the chemical combustion of a parent compound into gasses of very high temperature, pressure, and density. In this context, detonation of the explosive material releases energy via the rapid expansion of the combustion gasses. This rapid expansion produces a supersonic wave front, referred to as a shockwave, which is spherical in geometry and is characterized by discontinuities in temperature, pressure, density, and particle velocities through its thickness (Rajendran & Lee, 2009) (Shin, 2004)(Van Lear, 2008) (Spranghers et al., 2012) (Ngo et al., 2007) (Cole, 1948) (Smith & Hetherington, 1994). As a result of the gas expansion, the shockwave propagates radially and increases in size; its intensity also deteriorates radially, influenced by the medium in which is transmitted.

Review of UNDEX Theory
The shockwave produced in an UNDEX event propagates at a velocity much faster than that of the explosive gasses. Upon interfacing with the water medium, the initial shock front's velocity moves typically on the order of several thousand meters per second-that is, 3 to 5 times the acoustic velocity in water-and its initial pressure is similarly very high (Cole, 1948) (Shin, 2004) (Rajendran & Lee, 2009). In contrast, the spherical gas bubble initially retains a reduced internal pressure after shockwave emission, though this pressure is still of much greater magnitude than the equilibrium ambient plus hydrostatic pressure (hereafter referred to as the "ambient/hydrostatic" pressure). The bubble also expands radially, thus displacing the surrounding water in the process. The bubble expands in radius until a time just after its internal pressure reduces to the ambient/hydrostatic pressure, when, owing to fluid inertial effects, the bubble is caused to "over-expand" to a radius at which the internal pressure falls short of the ambient/hydrostatic pressure (Shin, 2004) (Cole, 1948)(Van Lear, 2008. The resultant pressure differential reverses the motion of the bubble, forcing it from expansion into contraction, during which time the internal pressure begins to increase once more. The effects of the gasses' compressibility, while negligible during expansion, are significant in the final stages of contraction and act to abruptly reverse the bubble's motion at the point of maximum collapse. This abrupt reversal of motion generates a new pressure wave in the water, referred to as the "bubble pulse." Upon collapse and emission of a bubble pulse, the gas bubble will continue in a periodic cycle of expansion and contraction until all of the explosion energy is released into the surrounding water or vented through the surface, if the event depth permits. During these cycles, called "bubble periods," the bubble is known to migrate upwards due to buoyancy effects. For reasons that are less obvious, the bubble is also known to

The Near-Field Problem
This concise review of basic explosion theory allows for a more detailed description of the criteria that this study selected to constitute a near-field explosion, and how these explosions differ from the far-field variety. In this study, a near-field explosion is one for which the standoff distance between the charge and the target are sufficiently short that the following conditions are true: 1. The influence of the shockwave curvature is significant during interaction with the target. In this way, the shockwave's spherical geometry is a prominent characteristic in its contact with the target and prevents planar wave approximations during analysis.
2. The target is either nearby or within a region of the blast zone where fluid structure interactions are made complex by the flow processes encouraged during shockwave generation.
UNDEX events in the near-field may further have special characteristics enjoined by the gas bubble behavior, characterized by either or both of the following: 3. In UNDEX events, the target is either nearby or within a region of the blast zone where bubble expansions and contractions directly and significantly influence the bulk fluid flow around the target.

Material/Specimen
The material considered in this analysis was an E-glass/Vinyl Ester (EVE) laminate, manufactured at TPI Composites in Warren, R.I. The laminate was layered biaxially using two plies of plain weave fiber sheets, and possessed an areal density of 0.61 kg/m 2 (18 oz/yd 2 ). The mass density of the E-glass fibers was taken to be 2.56 g/cm 3 . A vacuum-assisted resin transfer process was employed to saturate the layered sheets with Derakane 8084 vinyl ester resin, the mass density for which was taken to

Experimental Apparatus
The apparatus used in this study is illustrated in Figure 3.1-1. The depicted components are discussed in greater detail in sections 3.1.1 through 3.1.6.
Figure 3.1-1: Experimental apparatus, depicting the specimen, explosive charge, pressure sensors, DIC and side view cameras, specimen fixture, and heating/cooling system. The heating system was composed of 5 heating rods; the cooling system was composed of ice.

Experimental Water Tank
The experimental environment was contained within a special water tank. The

Experimental Fixture
Experiments required a unique specimen holding fixture so as to accommodate the panel geometry and provide the appropriate boundary conditions. This fixture consisted of a steel raised platform that was bolted to the floor of the water tank. The   The experimental holding fixture with a specimen clamped beneath the mounting bracket. This photo was taken just after an experiment and the specimen's concavity can be readily seen.

Underwater Blast Sensors
The pressure sensors employed in the WA and WW experiments were two series

Air Blast Sensors
The pressure sensors employed in the AA experiments were two series 137A21 integrated circuit piezoelectric (ICP 2 ) free field blast probes, serial numbers 10044 and 10045, produced by PCB Piezotronics, Inc. The distance from the conical probe tip to the quartz sensing element was 157 mm (6.2 in), and the overall probe length was 406 mm (16 in). The sensors were mounted horizontally with the sensing diaphragm oriented sideways.

Explosive Charge
The detonator used was an RP

RP-503 Pressure Decay
Placement of the aforementioned blast probes enabled the data at fixed standoff distances the sensors, these instruments could not capture pressure information at distances close to the specimen. Available UNDEX literature 19

Explosive Charge
The detonator used was an RP-503 secondary explosive. RP-503 is a plastic

Pressure Decay Test
Placement of the aforementioned blast probes enabled the registering of pressure data at fixed standoff distances from the explosive; but, due to concerns of damaging the sensors, these instruments could not capture pressure information at distances close to the specimen. Available UNDEX literature was consulted to find closed registering of pressure ; but, due to concerns of damaging the sensors, these instruments could not capture pressure information at distances d to find closed-form expressions for pressure wave decay under water, but the parameters associated with these solutions did not yield theoretical results that matched the experimental results after detonation (Cole, 1948) (Shin, 2004) (Batra & Hassan, 2007). It was supposed that these discrepancies could be attributed to the small dimensions of RP-503-the size of the bridge wire or detonator cap, for example, relative to the size of the explosive proper, could perhaps influence the detonation behavior in more drastic ways than those components would for heavier explosives. It was decided that the RP-503 pressure decay rate should be experimentally ascertained to produce an expression unique to the conditions used in this study, so as to accurately predict shockwave overpressures at the instant that they interface with the specimen.
Pursuant to this, a line of six tourmaline blast probes were arranged underwater on laboratory stands in the experimental water tank, and were placed at varying distances from an RP-503 explosive, as depicted in  The explosive was detonated and the pressure histories were recorded for each of the six standoff distances, the plots for which are overlaid on each other in Figure   3.1.4.1-2. The resulting peak pressures were plotted together in MATLAB as a function of standoff distance, and a least-squares curve fitting method was employed to establish a trend line function to extrapolate the data across a wider range of standoff distances at 95% confidence, as seen in Figure 3.1.4.1-3.  As can be seen in Figure 3.1.4.1-3, the pressure was seen to decay as a function of approximately 1 ‫ݎ‬ ൗ . Key results from the pressure decay test are included in Table   3.1.4.1-1. The trend line function expressed in Figure 3.1.4.1-3 was used to predict an overpressure of approximately 50.4 MPa experienced at the target at 76 mm. It must be noted however that extrapolating the data so far from the measured range between 120 and 305 mm introduced heightened uncertainty. This is indicated by the MATLAB curve fitting readouts in Figure 3.1.4.1-4, which indicate that the prediction bounds at 95% confidence, when evaluated at approximately 76 mm, are ± 17 MPa. Pressure in MPa is plotted on the Y axis, and standoff distance in mm is plotted on the X axis.

Digital Image Correlation
Digital Image Correlation (DIC) is a powerful optical tool in experimental mechanics that seeks to non-intrusively ascertain full-field displacement, strain, and velocity fields associated with a deforming body. The method is accomplished via use of high speed digital cameras and specialized analytical software (Sutton et al., 2009).
To record the deformation in real time, two high speed digital cameras are employed in a synchronized stereo arrangement. Calibration of the cameras is achieved via use of an image calibration grid, consisting of a white field and a distinct pattern of evenly-spaced points. The grid is rotated and translated in-and out-of-plane, as a series of individual photographs are taken by the stereo camera system. Since the spacing of the calibration points on the grid is predetermined, the analytical software is allowed to track the points' displacement. These displacements are tracked in a coordinate plane unique to both cameras. The software then correlates the images in these planes to establish a real-world, global coordinate system from which full-field, three-dimensional deformation measurements are made (LeBlanc et al., 2013). A calibration error of 10% or less is generally considered acceptable. As Haile notes, "A [DIC] camera is considered calibrated if the principal distance, principal point offset and lens distortion parameters are known" (Haile et al., 2009). This global threedimensional coordinate system is unique to the stereo camera layout, and any subsequent alteration of the camera layout necessarily invalidates the calibration (such alterations might include shifting of one or both cameras, replacement of one or both camera lenses, or placement/removal of additional transparent media in front of the cameras). Before experiments, the observation side of the deforming body is painted with a random black-and-white speckle pattern. This random pattern creates a diverse field of unique pixel intensity subsets whose displacements, during specimen deformation, are photographically captured by the high speed cameras. The analytical software tracks and interprets these displacements, which allows consequently for the three-dimensional assessment of strain, velocity, and other applicable parameters (Shukla & Dally, 2010).

DIC Cameras
The high speed cameras employed for the DIC technique were two Photron Fastcam SA1.1 units, of model number 675K-M1, with 8GB internal memory. These cameras can achieve frame rates between 1,000 and 675,000 frames per second with corresponding image resolutions between 1,024 x 1,024 and 64 x 16, respectively.

DIC Software
The analytical post-processing software employed for the DIC technique was "Vic-3D," produced by Correlated Solutions, Inc. Vic-3D uses the DIC method to employ various strain tensors in providing full-field, three-dimensional deformation, strain, and shape measurements across the surface of the deforming body.

Heating and Cooling Devices
Experiments investigating the influences of high and low water temperaturesnamely WAHT, WALT, WWHT, and WWLT-required unique heating and cooling techniques to achieve the desired environments. These are discussed in sections 3.1.6.1 and 3.1.6.2.

Heating Elements
The heating elements used in the WAHT and WWHT experiments were Allied Precision Industries 742G bucket heaters, delivering 1 kW of power each, as seen in  The heaters were suspended from rods above the surface of the water, with the heating elements completely submerged. This arrangement can be seen in figure   3.1.6.1-2.

Cooling Elements
The cooling element used in the WALT and WWLT experiments was cubed ice.
Again, noting the nominal water entrance temperature as approximately 23 °C, the operational water volume in the tank, and the desired water temperature of 0 °C, and assuming ice cubes of -10 °C, equation 3.1.6.2-1 was used to calculate the required mass of ice to sufficiently lower the water temperature, as follows: in which h is the latent heat of ice. Appropriate application of equation 3.1.6.2-1 yielded a required mass of 150 kg (330 lb.) of ice per experiment to adequately cool the water.

Water Submersion, Air Backing (WA)
Experiments were performed to investigate the response of the EVE composite specimen to an UNDEX event in an experimental environment characterized by water submersion and air backing-an environment referred to hereafter as "WA." For each experiment conducted thus, silicone caulk was used to fill the 3.18 mm ( 203 mm (8 in.) from the detonator, respectively. The minimum probe standoff distance of 127 mm was chosen so as not to damage the instrument. The blast probe that was positioned closer to the detonator, hereafter referred to as "sensor 1," possessed a tourmaline sensing element located at a distance of 97 mm (3.8 in.) 3 directly above the conical tip. The tourmaline element for sensor 1 therefore stood at a radial distance of 160 mm (6.3 in.) from the detonator. In a similar way, the blast probe that was positioned farther from the detonator, hereafter referred to as "sensor 2," possessed a tourmaline sensing element located at a distance of 99 mm (3.9 in.) 4 directly above the conical tip. The tourmaline element for sensor 2 therefore stood at a radial distance of 226 mm (8.9 in.) from the detonator. 100 mm DIC camera lenses were employed and the camera frame rate was set to 20,000 frames per second (FPS), rendering a DIC image resolution of 512 x 512 with an inter-frame time of 50 µsec. The RP-503 charge was detonated by an independent firing box, which was wired to an isolated electrical circuit to minimize the risk of power surges influencing the recording oscilloscope.
The deflections of the speckle patterns on each specimen were observed by the DIC cameras and processed by the DIC software. Out-of-plane deflections were measured from the plane of the un-deformed specimen. The pressure waves induced by the RP-503 explosion were detected by the tourmaline blast probes, whose millivolt signals were amplified to ± 10 VDC signals by an in-line conditioner and relayed to a recording oscilloscope. The oscilloscope was commanded to trigger upon a rising voltage of 400 mV from sensor 1. Key experimental apparatus values are presented for convenience in Table 3

WA Experiment 1
The first WA experiment was performed with the recording oscilloscope set to a sampling frequency of 10 MHz. The DIC camera angles of incidence with the transparent Lexan viewing window were not recorded during this experiment. The     Table   3.2.1.1-1.

WA Experiment 2
The second WA experiment, due to an instrument malfunction, was performed without the use of an oscilloscope to record pressure data. The DIC camera angles of incidence with the transparent Lexan viewing window were also not recorded during this experiment, but the DIC calibration error was found to be 10.0%. Experiment 2 produced DIC results that were repeatable from experiment 1; the center point deflection profile bore remarkable similarity to that of experiment 1, and the X-shaped plateau was again plainly observable and responsible for a brief outward rebound of the center point (see

WA Experiment 3
The third WA experiment was performed with the recording oscilloscope set to a sampling frequency of 10 MHz. The DIC camera angles of incidence with the transparent Lexan viewing window were 7º for both the Master and Slave 1 cameras, and the corresponding DIC calibration error was found to be 9.0%. A side view camera with a 28 mm lens was employed to observe the expansion and collapse of the

WA Experiment 4
The fourth WA experiment was performed with the recording oscilloscope set to a sampling frequency of 10 MHz. The DIC camera angles of incidence with the transparent Lexan viewing window were 7º for both the Master and Slave 1 cameras, and the corresponding DIC calibration error was found to be 9.5%. A side view camera with a 28 mm lens was employed to observe the expansion and collapse of the gas bubble produced as a result of the detonation. Due to a faulty instrument cable connection, sensor 2 was not able to register any meaningful signal. In spite of this, experiment 4 produced results that were repeatable from experiments 1, 2, and 3. Key  The gas bubble was observed to expand in radius until an elapsed time between

Water Submersion, Water Backing
Experiments were performed to investigate the response of the EVE composite specimen to an UNDEX event in an experimental environment characterized by water submersion and water backing-an environment referred to hereafter as "WW". The holding box was moved backwards on the mounting stand and re-bolted, so as to provide sufficient clearance to allow proper water circulation while the panel flexed.
The mounting/boundary conditions, the detonator/blast probes and their positioning, the camera lenses, settings, and software, and the supporting data acquisition devices employed in the WW experiment set remained identical to those employed in the WA set. These invariable parameters are reflected in Table 3.2.1-1.

WW Experiment 1
The first WW experiment was performed with the recording oscilloscope set to a sampling frequency of 10 MHz. The DIC camera angles of incidence with the transparent Lexan viewing window were 6º for the Master camera and 7º for the Slave 1 camera, and the corresponding DIC calibration error was found to be 8.5%

WW Experiment 3
The third WW experiment was performed in quick succession after the second,

Air Immersion, Air Backing
Experiments were performed to investigate the response of the EVE composite specimen to an UNDEX event in an experimental environment characterized by air immersion and air backing-an environment referred to hereafter as "AA". The

AA Experiment 1
It was recognized from theory (Cole, 1948) (Smith & Hetherington, 1994) (Shin, 2004) (Ngo et al., 2007) that the pressure wave decay rate in air would be greater than in water. Without full knowledge of the energy release during an RP-503 detonation in air, the appropriate pressure ranges to be expected at certain radii from the explosive were not clearly known. Because of this uncertainty, it was recognized that the oscilloscope trigger, still initiated by the amplified signal from sensor 1, needed to be lowered as far as possible while at the same time remaining above the instrument noise level. This was achieved by lowering the trigger to 100 mV. However, during detonation the electromagnetic interference induced by the firing box's 2000 V pulse prematurely triggered the oscilloscope and caused high frequency noise that consumed the blast probe signal. For this reason, the first AA experiment yielded no useful pressure data. The tips of sensors 1 and 2 nevertheless stood off at respective distances of 76.2 mm (3 in.) and 136.53 mm (5.375 in.) from the explosive. The DIC camera angles of incidence with the specimen plane were 7º for the Master camera and 6º for the Slave 1 camera, and the corresponding DIC calibration error was found to be 10%.   Key values from AA experiment 1 are listed for convenience in Table 3.2.3.1-1.

AA Experiment 2
The second AA experiment, due again to premature oscilloscope triggering, highfrequency noise, and additional error, yielded neither useful pressure data nor any DIC data. However, post mortem damage was consistent with that experienced in AA experiment 1.

AA Experiment 3
The third AA experiment was accomplished with two oscilloscopes, both with separate triggering mechanisms. The first oscilloscope (hereafter "oscilloscope 1") was triggered directly by a new firing box, with the capability of sending an independent 9 V triggering signal to oscilloscope 1 in concert with a 3000 V explosive detonation pulse. The second oscilloscope (hereafter "oscilloscope 2") was triggered by an external circuit beak. The circuit break supplied a 5 V triggering signal to oscilloscope 2 after a graphite rod, positioned 1 in. from the explosive, fractured during detonation ( pressure phases are clearly visible in the sensor 1 signal. The sensor 2 signal tends towards its negative phase, but apparent wave reflections prevent it from experiencing that phase as quickly as the signal from sensor 1. Since oscilloscope 2 operated with inferior resolution than did oscilloscope 1, its detected pulses were lesser in magnitude than oscilloscope 1's. Because of this, only the data from oscilloscope 1 is presented here. A filtered plot of sensor 1 data from oscilloscope 2 is provided in Figure 3

AA Experiment 4
The fourth AA experiment was accomplished with one oscilloscope with a graphite circuit break trigger. The oscilloscope was arranged with a sampling frequency of 10 MHz. This lower sampling frequency was used in AA experiment 4 due to the cumbersome size of the data sets from AA experiment 3, which slowed down processing to such an extent that the data had to be broken into 4 individual files. The DIC camera angles of incidence with the specimen plane were 7º for the Master camera and 12º for the Slave 1 camera, and the corresponding DIC calibration error was found to be 3.0%  Post mortem damage was consistent with that seen in the previous experiments.

AA Experiment 5
The fifth AA experiment was accomplished without pressure sensors, and DIC data was recorded only. A side view camera with a 28 mm lens was employed in an attempt to observe the phenomena responsible for the unique post mortem damage characteristic of the AA experiment set. Of particular interest was the cause of the through-thickness cleft. The DIC camera angles of incidence with the specimen plane were 7º for the Master camera and 12º for the Slave 1 camera, and the corresponding DIC calibration error was found to be 2.8%. The event was sufficiently bright and quick that the side view camera was unable to observe the cause of any damage. In

Low Temperature Water Immersion, Air Backing
Experiments were performed to investigate the response of the EVE composite specimen to an UNDEX event in an experimental environment characterized by water submersion and air backing at water temperatures of approximately 0 °C-an environment referred to hereafter as "WALT." Apart from the water temperature, the WALT experiment series was conducted under largely identical conditions as the WA series: DIC resolution and frame rate, pressure sensor type and standoff distances, RP-503 standoff distance, and waterproofing techniques remained the same between sets.
A notable difference in these parameters was the use of 60 mm DIC lens in the WALT series. Side view illumination was also provided this time by an SSG-400 filamentless 400 watt HMI spotlight, manufactured by Frezzi Energy Systems, through the rear observation panel of the water tank. Besides this, the pre-experiment specimen and instrument preparation methods were identical. Key invariable parameters for the WALT experiment set are included in Table 3.2.4-1. To achieve the required 0 °C water conditions, equation 3.1.6.2-1 was employed to determine the necessary mass of ice to be mixed in the tank, assuming a -10 °C ice temperature. These calculations, accounting for the mass and nominal temperature of the water, indicated that 150 kg (330 lb.) of ice were required per experiment to chill the water sufficiently. Cubed ice was purchased in 9 kg (20 lb.) bags, which were emptied directly into water tank, either before, during, or after filling. The water temperature was monitored with a network of 5 thermometers that were embedded in a small Styrofoam flotation raft. Cooling durations took between two and four hours to accomplish, and water was circulated via manual mixing with a wooden plank.

WALT Experiment 1
In the first WALT experiment, the recording oscilloscope was arranged with a sampling frequency of 10 MHz. The DIC camera angles were not recorded for this experiment. The calibration error was found to be 5.1%. A side view camera with a 28 mm lens was employed, and appropriate illumination was supplied through the rear observation window, as stated before. The cooling process was halted when the average registered temperature reached 1.4 °C. Individual qualifying temperatures for each thermometer are given in Table 3.2.4.1-1.

WALT Experiment 2
The second WALT experiment was conducted without having moved the DIC camera system after WALT experiment 1. Because of this, WALT experiment 2 was analyzed using the same calibration images and thus had the same calibration error. In the second WALT experiment the recording oscilloscope was arranged with a sampling frequency of 10 MHz. The DIC camera angles were, again, not recorded for the experiment. A side view camera with a 28 mm lens was again employed, and appropriate illumination was accordingly supplied through the rear observation window. The cooling process was halted when the average registered temperature reached 2.7 °C. Individual qualifying temperatures for each thermometer are given in

Low Temperature Water Immersion, Water Backing
Experiments were performed to investigate the response of the EVE composite specimen to an UNDEX event in an experimental environment characterized by water submersion and water backing at water temperatures of approximately 0 °C-an environment referred to hereafter as "WWLT." The set up and preparation for these experiments was almost completely identical to those of the WALT series, the only prescribed difference being that no sealing was undertaken between the fixture and the Lexan front observation window or between the specimen and the fixture. As with the room temperature WW experiments, the fixture was retracted slightly from the observation window so as to allow for adequate water circulation during the UNDEX event and, as with the WW experiments, only 200 µsec of DIC data could be obtained due to thick cavitation in front of the specimen. Key invariable parameters for the WWLT experiment set, since they were identical to those in the WALT series, are included in Table 3.2.4-1. The cooling method and process remained the same for the WWLT series as it had been in the WALT series, as did the required mass of ice and general cooling duration.

WWLT Experiment 1
In the first WWLT experiment, the recording oscilloscope was arranged with a sampling frequency of 100 MHz. The DIC camera stereo angle was 15°, and the calibration error was found to be 6.3%. A side view camera with a 28 mm lens was employed, and appropriate illumination was supplied through the rear observation window. The cooling process was halted when the average registered temperature reached 3.5 °C. Individual qualifying temperatures for each thermometer are given in Table 3.2.5.1-1.      Table 3.2.5.1-2.

WWLT Experiment 2
In the second WWLT experiment, the recording oscilloscope was again arranged with a sampling frequency of 100 MHz. The DIC camera stereo angle was 15°, and the calibration error was found to be 4.6%. A side view camera with a 28 mm lens was employed, and appropriate illumination was supplied through the rear observation window. The cooling process was halted when the average registered temperature reached 3.3 °C. Individual qualifying temperatures for each thermometer are given in Table 3.2.5.2-1.     Table   3.2.5.2-2.

High Temperature Water Immersion, Air Backing
Experiments were performed to investigate the response of the EVE composite specimen to an UNDEX event in an experimental environment characterized by water submersion and air backing at water temperatures of approximately 40 °C-an environment referred to hereafter as "WAHT." Because this experiment series bore exact likeness to the WALT series apart from the experimental water temperature, To achieve the required 40 °C water conditions, Equation 3.1.6.1-1 was used to obtain the required power input to raise the water temperature from 23 °C over a period of approximately 2 hours. The resulting power, just over 5 kW, was rounded to 5 kW due to electrical constraints in the experimental facility, as described in Section 3.1.6.1. Five 1-kW water heaters were suspended from rods above the water in the tank, with the heating elements fully submerged. Water circulation was achieved by an impeller. The water temperature was monitored by a network of five thermometers that were embedded in a small Styrofoam flotation raft. At any point during the heating process, the remaining heating duration could be ascertained by solving Equation 3.1.6.1-1 for ∆t.
To prevent skin burns, measures were taken to avoid coming in contact with the hot water or the water heaters. Neoprene heat-resistant gloves were worn when handling the water heaters, and the RP-503 explosive was inserted into the water via a small-bore copper tube. The explosive's lead wire was fed through the tube until the charge capsule was flush with it. The rigid tube provided an ideal means of directing the position of the explosive once placed in the water, and it was fastened in place simply by spanning a rod across the top of the water tank and taping the tube to the rod.

WAHT Experiment 1
In the first WAHT experiment, the DIC camera angles were 8° for the Master camera and 7° for the Slave 1 camera. The calibration error was found to be 7.3%. The heating process was halted when the average water temperature reached 41.5 °C.

WAHT Experiment 2
WAHT experiment 2 was performed immediately after WAHT experiment 1.
Because of this, the DIC camera angles remained 8° for the Master camera and 7° for the Slave 1 camera. The calibration error also remained constant at 7.3%. The heating process was halted when the average water temperature reached 41.2 °C. Individual qualifying temperatures for each thermometer are given in Table 3.2.6.2-1.   Post mortem damage was very similar to that observed in the WAHT 1 specimen.

High Temperature Water Immersion, Water Backing
Experiments were performed to investigate the response of the EVE composite specimen to an UNDEX event in an experimental environment characterized by water submersion and water backing at water temperatures of approximately 40 °C-an environment referred to hereafter as "WWHT." As with the case between the WALT and WWLT series, setup and preparation for the WWHT experiment set bears great similarity to those for the WAHT set. Notable differences are limited to, again, the retracted location of the experimental fixture and the presence of thick cavitation in front of the specimen during the UNDEX event, permitting only the first 200 µsec of DIC data to be confidently processed.

WWHT Experiment 1
In the first WWHT experiment, the DIC camera angles were 8° for the Master camera and 7° for the Slave 1 camera, and the calibration error was found to be 4.4%.
The heating process was halted when the average water temperature reached 40.8 °C.
Because of this, the DIC camera angles remained 8° for the Master camera and 7° for the Slave 1 camera. The calibration error also remained constant at 4.4%. The heating process was halted when the average water temperature reached 40.8 °C. Individual qualifying temperatures for each thermometer are given in Table 3.2.7.2-1.  Post mortem damage included some minor delamination, matrix cracking around the boundaries, and the specimen had a noticeable permanent concavity. Key parameters from WWHT experiment 2 are listed in Table 3.2.7.2-2.

Discussion of Results
Pursuant to the goals of this study, the foregoing results provided insight to a number of questions. Some of these are discussed here.

Air Pressure Decay
Especially owing to the intense noise encountered and the low signals detected, it was of interest to compare the recorded pressure decays in air with established theory.
Smith and Hetherington give relations for shockwave pressure (in bar) as a function of standoff distance and explosive equivalent weight of TNT, with the caveat that, due to complex fluid flow processes close to the charge, the accuracy of pressure predictions in the near-field, "is somewhat lower than in the medium to far field" (Smith & Hetherington, 1994 These theoretical pressures are given in Table 4.1.1-1.  The average overpressures from sensor 1 and sensor 2 are thus seen to have been 0.125 MPa and 0.05 MPa, respectively. From this it can be seen that the pencil probes captured data that follow the theory to a reasonable degree, considering the propensity for error at such close standoff distances as proposed by Smith and Hetherington. When calculating the pressure at the specimen distance of 76 mm (3 in.), the scaled distance parameter Z was seen to be 0.76. Thus Equation 4.1.1-1 was employed for ascertaining the overpressure at the panel specimen surface. Doing so gave 1.357 MPa (197 PSI).
Another notable pressure effect in the AA series occurred in the center point deflection profiles. It was seen that the maximum positive displacement of each experiment was of lesser magnitude than the maximum negative displacement. This seemingly counterintuitive phenomenon could be explained by the influence of the negative pressure phase as depicted in Figure 4.1.1-1, which details the filtered pressure decay seen in sensor 1 from AA experiments 3 and 4. In the figure, the onset of the negative pressure phase can be observed at approximately 100 msec. Extraneous wave reflections-perhaps from the bulky structure used to restrain the pencil probes-appear to have influenced the pressure transducer signal between 100 and 350 msec, but it's unlikely that these oscillations would have been experienced to any significant degree at the specimen interface. With this said, it can be reasonably stated that, absent the said reflections, the natural negative pressure phase began at approximately 100 msec and continued for a time duration beyond that which was recorded. This duration demonstrates the extent that the negative pressure "vacuum" could have influenced the panel specimen, as compared to the shorter positive phase. When considered over the whole explosion event, it is possible that the positive impulse applied to the panel by the shockwave was actually smaller than the negative impulse applied to the panel by the vacuum.
This would explain why the maximum positive center point panel deflection was smaller than the maximum negative deflection.

Water Pressure Decay
The recorded pressure histories collected in water are illustrated in    Taking all this into consideration, the peak overpressures for sensors 1 and 2 then are displayed in Table 4.1.2-1. The mean sensor 1 overpressure is then seen to have been 24 MPa, and the mean for sensor 2 is seen to have been 15 MPa. The extrapolated overpressures, based on the MATLAB-generated trend line function of ‫ݎ9245‬ ିଵ.଼ , are listed in Table 4.1.2-2.

Environmental Effects
The central issue of this study was to examine particular environmental effects and their influence on the near-field blast response of the EVE composite. A variety of criteria could be selected to gauge the blast response of the panel specimen, but this study chiefly considered one criterion in particular, namely center point deflection.
When considered over the whole duration of an UNDEX or air blast event, the panel center point experiences diverse forms of oscillatory behavior, which produce a range of deflections that could potentially be used to measure blast response. Of these, this study defined the maximum positive displacement as most significant and most indicative of resilience to the blast load. The maximum positive displacement was recognized as being best suited to measure the immediate, direct effects of the shockwave on the panel. This blast response was investigated by varying two environmental characteristics, namely the backing medium and water temperature.
Section 4.2 of this document will examine these effects by considering them in turn.     Table 4.2.1-1.   It is believed that the significant differences in the maximum panel deflections were caused in large part by the differing characteristic acoustic impedances of water, air, and the composite material. Characteristic impedance is a material property that influences wave reflection and transmission between two media. Waves are transmitted more easily between media with similar impedances than they are between media whose impedances are dissimilar. Equations 4.2.1-1 through -3 describe the relationship between wave amplitude and impedance, expressed as the product between a material's density, ρ, and longitudinal wave speed, c (LeBlanc et al.,

Effects of Backing Medium
2013) (Sadd, 2009) (Gracia, 2012)   This can be explained by considering the impulse imparted to the specimen in those two environments. A pressure wave decays in air at a more drastic rate than it does in water. This can be learned from the literature (Cole, 1948) (Shin, 2004) (Batra & Hassan, 2007) (Smith & Hetherington, 1994) (Ngo et al., 2007), which provides empirical formulations of pressure decay as being functions of 1 ‫ݎ‬ ଷ ൗ during air blasts and as high as 1 ‫ݎ‬ ଶ ൗ during UNDEX events, and can also be interpreted from the pressure histories included in this document. As discussed previously in Section 4.1, the pressure at the target in an air blast event was roughly 1.357 MPa (197 PSI), a little more than 37 times less than the 50.5 MPa (7324.4 PSI) encountered at the target in an UNDEX event. Ergo, although the AA series presented an environment more hostile to wave propagation, the wave had dissipated to such a degree that the impulse applied to the panel specimen was considerably smaller. These results suggest that the wave dissipation effects dominate the impedance mismatch effects when considering explosions in water versus those in air.       There is insufficient data to examine the full effect of water temperature on the water backed environments (WW, WWLT, WWHT), but it is clear that, within the known 200 µsec range, there appears to be no influence of temperature.

Effects of Temperature
The deflections exhibited by the panels from the air backed experiments (WA, WALT, WAHT) suggest a dependence between water temperature and center point deflection. The average maximum positive deflections from the air backed series are included in Table 4.2.2-1. These points were statistically analyzed using both ANOVA methods and an original MATLAB code at 90% confidence (see Appendix C). Both MATLAB and ANOVA indicated that, while the maximum positive WA and WAHT deflections were statistically the same, the maximum positive WALT deflection was statistically different than both the WA and WAHT deflections. These results are listed in Table   4.2.2-2. Under the MATLAB scheme, t o is a distribution parameter defined as where n 1,2 are the sample sizes of the experiments being compared, y is the sample mean, and S 1,2 are the sample variances. Additionally, t α/2 is an element of the t-distribution table, and is based on the sample size and desired confidence interval.
Criteria for rejection of the null hypothesis-that any two corresponding experimental values are the same-is that t o be greater than t α/2 . Under the ANOVA scheme, P is the probability that variances and differences between two data sets would still exist if the null hypothesis were true. F is a ratio of cross-group variance to within-group variance. F critical is a threshold value of F beyond which two data sets are said to be statistically different.
With these statistical differences and similarities established, quantitative disparities among the data points in Table 4.2.2-1 were more readily appreciated.
Using the room temperature average center point deflection as a baseline for comparison, it can be seen that the average high temperature center point deflection was increased by 3%, while the average low temperature center point deflection was decreased by 10%. Bearing these points in mind, it is seen that, over the temperature range studied, and though the influence does not appear to be extraordinarily great, temperature influence on center point deflection did appear to manifest itself as the water was made colder.

Effects of Backing Conditions
The backing conditions greatly influenced the damage mechanisms exhibited in the panel specimens. Backing condition effects were evaluated by cross-comparing room temperature post mortem damage.

WA Post Mortem
Post mortem damage in the WA series occurred predominantly at the clamped boundary and manifested itself chiefly as fiber breakage and delamination. In some cases the fiber breakage propagated through the panel thickness along a seam.
Through thickness breakages were not severe, though. Matrix cracking also existed in localized areas across the panel surface. Permanent deflection was observable and the panels had visible concavity.
To be expected, boundary effects apparently encouraged the development of high stress areas along the clamped edge, thus leading to pronounced fiber breakage and delamination in those areas. The Matrix cracking was likely induced by the contorted vibration modes that the panel experienced during the UNDEX event.

WW Post Mortem
The WW series specimens experienced mostly matrix cracking around their boundaries and exhibited only sparse, highly-localized occurrences of delamination at the boundary. Small amounts of matrix cracking were typical of the WW specimens, which occurred towards the panel center. The panels had almost no visible permanent deflection, and virtually no concavity was observed in them.

AA Post Mortem
The prevailing damage mechanism in the AA series was impact damage from flying shrapnel produced when the RP-503 capsule exploded. The AA panels exhibited no visible delamination around their boundaries, which instead was pockmarked with impact craters from shrapnel. Some craters were black, indicative of resin singeing after plastic shrapnel became embedded between the specimen mounting bracket and the panel. A burned laceration was marked horizontally along the whole panel surface, across which there was near-continuous fiber breakage, much of which was through-thickness. This cleft appeared as a demarcation line, below which was a dense field of burned pockmarks and impact damage, and above which was a much sparser, unburned field of impact damage. It is believed that the dense impact damage below the demarcation line could be indicative of the explosive shrapnel being deflected towards one area by the charge geometry or by other components in the blasting cap of the RP-503 charge.

Effects of Temperature on Damage Mechanisms
Damage mechanisms were not observed to have changed significantly as a function of temperature. Damage mechanisms in the high and low temperature specimens were evaluated and compared with room temperature mechanisms.

WALT Post Mortem
Although the specimen for WALT experiment 2 tore completely across a long seam on its bottom clamped edge, the prevailing damage mechanisms on both of the WALT panels caused that occurrence to be considered anomalous. The tearing, which occurred as a result of the first bubble pulse, could have been caused more fundamentally by quality variations during manufacturing. There was less-pronounced delamination around the WALT panels' boundaries and only superficial fiber breakage in those areas, apart from the tear in WALT specimen no. 2. Matrix cracking occurred in largely the same manner as it did in the WA series. Permanent concavity was noticeably lower than that of the WA series.

WAHT Post Mortem
The WAHT panels exhibited somewhat more delamination around the clamped boundary than was seen in the WALT series, but still less than exhibited in the WA series. As before, matrix cracking appeared in much the same way as it had in the WA series. The WAHT panels exhibited the most pronounced permanent concavity of all the specimens in this study.

WWLT Post Mortem
WWLT panels exhibited very little delamination around the clamped edge, except in the case of WWLT specimen 1, which exhibited quite noticeable delamination along about half of its boundary. Very faint matrix cracking was observed in areas across the panel face, and neither panel exhibited any visible concavity.

WWHT Post Mortem
WWHT specimens exhibited more concavity than did the WWLT specimens, and also displayed some matric cracking and sparse delamination around the clamped edge. Otherwise they exhibited no further unique damage mechanisms.
From these results it should be noted that water temperature appeared only to significantly influence the permanent concavity of the panels.

WA Correlation with the Gas Bubble
The air backed deflections followed a pattern indicative of heavy dependence on the progress of the gas bubble. Given the repeatability of the experiments, and given the similar displacement trends for each of the water temperatures, WA experiment 3 will suffice as a representative case for discussion. In WA experiment 3, the maximum  5. The panel's movement was arrested until the bubble radius increased to such an extent that the inertial effects on the surrounding water were reduced due to the gradual slowing of the bubble growth and its reversal to collapsing motion.
6. As the bubble's internal pressure dropped to, and fell below, ambient conditions, its reversal into collapse began dragging the surrounding water in the collapse direction. This not only created a suction current towards the bubble's center, but also relieved the compressive force the water barrier and the panel.
7. The bubble subsequently initiated its collapse period from 14-23 msec.
8. This change in fluid dynamics permitted the gradual ebb in panel deflection towards its negative maximum, observable from approximately 10-18.5 msec. After achieving its negative maximum at ~18.5 msec, the elastic response of the E-glass fiber reinforcement again reversed the panel motion away from the bubble center.  suggested that the WW, WWLT, and WWHT series experienced the least center point deflections and most benign damage mechanisms. The disparity in blast response was attributed to differences in characteristic impedance between the panel material and the immersion environment, but that, although the AA series experienced the most hostile environment from an acoustics standpoint, the water/air backed series (WA, WALT, WAHT) ultimately experienced the greatest environmental punishment due to water's ability to sustain pressure wave intensity.
Experimental results indicated that water temperature influenced panel blast response over the range of temperatures from 0 °C to 40 °C, based on available DIC data for the WA, WALT, and WAHT series. WAHT experiments displayed only 3% greater average maximum center point deflection than the room temperature WA series. In contrast, WALT experiments displayed average maximum center point deflections that not only were 10% smaller than those exhibited by the WA series, but also were proven to be statistically different than the corresponding deflections from the WAHT and WA series by two independent statistical analyses. Post mortem results for those specimens indicated no appreciable temperature influence on damage mechanisms, apart from permanent concavity. The available DIC data for the WW, WWLT, and WWHT series suggested no difference in center point deflection across temperatures; but since the available DIC data pertained to only the first 200 µsec of the blast event on account of dense cavitation, it's unclear what can be conclusively inferred from it. Minor variations in post mortem damage were insufficient by themselves to imply temperature dependent damage mechanisms.

Recommendations for Future Work
Having stated the conclusions of this study, there also are some ways that the research may be advanced in future work. Some of these are as follows: 1. Correlations may be drawn between the applied overpressure and the maximum positive panel center point displacement. These correlations may be developed by conducting similar experiments to those of this study, while varying the explosive standoff distance.
2. Greater insight regarding the temperature dependence of center point deflection and damage mechanisms may be gleaned by the following: a. Increasing the range of water temperatures, especially to investigate why the high-and room-temperature maximum center point deflections were statistically the same, whereas the low-and room-temperature deflections were statistically different.
b. Drastically raising the temperature of the panel specimen above that of the environment, to simulate blast effects on a structure heated by prolonged exposure to direct sunlight.
3. Experiments may be conducted in an environment characterized by air immersion with water backing, to simulate the blast effects of a detonation beside the interior bulkhead of a ship or submerged structure. a. Two parallel, opposite edges of the flow media were measured to be 2" shorter than the edges of the glass fiber sheets; the remaining two edges of the flow media were measured to be 1" shorter than the glass fiber sheets ( Figure A.1-1). c. The flow media was adhered to the peel ply in a similar manner as described in bullet 7. 9. Scrap material (glass fabric), approximately 7" wide, was adhered to the edges of the Peel Ply for which the flow media stood off 2" (top and bottom edges of Figure A.1-1). Rope of approximately 0.5" diameter was laid in the middle of the scrap material, running lengthwise. The scrap material was folded over the rope so as to envelop it, and was adhered in place using the 3M spray adhesive. On both edges, a small end portion of the scrap material was not adhered so as to allow the connection of vacuum tubes later on. 10. A spring coil, enveloped within a flow media sleeve, was laid across the middle of the flow media sheet described in bullet 8 (tacky tape was applied to potential sharp edges of the spring coil to prevent the puncturing of the vacuum bag, applied later on). The flow media sleeve, enveloping the spring coil, was taped in place with small periodic applications of Tacky Tape.
11. Vacuum tubes were inserted over the rope contained within the scrap material, as depicted in Figure   The vacuum tubes were laid over the Tacky Tape surrounding the glass sheet/Peel Ply setup. A resin feed tube was also inserted into the spring coil described in bullet 10. 12. A plastic sheet was measured and cut to serve as the vacuum tube for the panel manufacture. The vacuum bag was measured to fit comfortably over the Tacky Tape perimeter, with comfortable excess.
13. The Tacky Tape adhesive backing was gradually removed, and the vacuum bag was in turn pressed against the tape, leaving pleads (bunny ears) in strategic places to ensure an air-tight fit with Tacky Tape later on. This process is depicted in Appendix A.1.
Similar considerations were made for Nate Gardner's composite panels.

Thursday, August 23, 2012:
Departed URI at 7:30 A.M. Arrived at TPI Composites around 8:30 A.M. Work of the day consisted of the following: 1. A pressure drop test was performed so as to ensure an air-tight seal.
a. The vacuum tubes were connected to a vacuum chamber, and the resin feed tube was clamped shut using a vice grip clamp.
b. The vacuum chamber was connected to the company's low pressure air mains.
The vacuum valve was switched into the flow position.
c. Low pressure was induced at 15 inHg. Any audible leaks were closed by pressing the vacuum bag harder into the Tacky Tape. d. After audible leaks were closed the low pressure was switched to 30 inHg. Once the vacuum pressure reached a steady state, the peak pressure was recorded and the vacuum chamber valve was closed. After 2 minutes the pressure was recorded again.
e. Final pressure must be above 27 inHg to qualify for a good enough seal.
2. The vacuum was left running while the resin was mixed.
a. The weight of the glass fiber sheets was determined from its area and areal density. The sheets were both 36" x 36", or 1 yd 2 . By the areal density described in bullet 3 of the August 22 notes (18 oz/yd 2 ), the total weight of the two glass panels was 18 oz + 18 oz = 36 oz. Simple conversion yielded 2.25 lb.
b. The amount of resin used was approximately 5 lb.
c. The type of resin used was Ashland Derakane 8084 Vinyl Ester resin.
3. The resin feed tube was inserted into the resin bucket, the vacuum pressure was turned on, and the feed tube was unclamped.

Friday, August 24, 2012:
Departed URI at 7:30 A.M. Arrived at TPI Composites around 9:00 A.M., due to traffic. Work of the day consisted of the following: 1. Composite specimens were cut using a 1/8"-thick diamond-edged saw. The 36" x 36" composite panel was cut into sixteen 8" x 8" specimens. Extra material was also collected for possible use in sundry analysis later on.
2. Similar actions were performed for Nate Gardner's panels.
Departed TPI Composites at 10:30 A.M. and returned to URI at 11:30 A.M.         while j<2; %This loop will continue until all outlying data points are deleted, after which time j will be increased to 2 and the loop will break. n=length(SAMPLE1); %Redefine n to be the number of data points in the ammended vector SAMPLE1. StandDev=std(SAMPLE1); %Redefine the standard deviation based on the updated SAMPLE1 Mean=mean(SAMPLE1); %Redefine the mean of the updated SAMPLE1 Max=max(SAMPLE1); %Since maxima and minima are the first candidates for elimination, identify the maximum in SAMPLE1. Min=min(SAMPLE1); %Since maxima and minima are the first candidates for elimination, identify the minimum in SAMPLE1.
Delt1=abs(Max-Mean); %Determine the absolute difference between the maximum in SAMPLE1 and the mean of SAMPLE1 Delt2=abs(Min-Mean); %Determine the absolute difference between the minimum in SAMPLE1 and the mean of SAMPLE1 DELT=[Delt1 Delt2]; %Put the two resulting differences into a vector, for comparing against each other. Larger=max(DELT); %Assign the variable "Larger" to be the larger value of the above two differences.
Thomp=THOMPSON(n)*StandDev; %Evaluate the product between the Thompson "tau" value (that corresponds to the sample size) and the standard deviation.
if Thomp < Larger; %If the product is smaller than the larger difference value, then the data point corresponding to that difference is an outlier and needs to be deleted. SAMPLE2=SAMPLE1; %SAMPLE2 is an arbitrary duplicate of SAMPLE1, and is referenced later on by the DELETE vector after SAMPLE1 has been ammended. SAMPLE2 serves only to allow DELETE to reference data points from SAMPLE1 that have already been eliminated.